principles of CONSTRUCTING SEMANTIC ANNOTATIONS OF IMAGE CONTENT

PRINCIPLES OF ANNOTATING THE IMAGE CONTENT

E.G. Sokolova ( [email protected])
Computing Center of the Russian Academy of Sciences. A.A. Dorodnitsyna

M.V. Boldasov ( [email protected])
Russian Research Institute of Artificial Intelligence

The article discusses the principles of creating semantic annotations to describe the content of images of scenes of discrete objects represented by photographs of open space - city views, landscapes. Annotations are created based on the ontology of concepts of a given subject area in the form of XML representations. Annotations describe the static content of images, including the composition of the image. Formalized annotations can be used both for information retrieval, for example, in library and museum image collections, and for generating natural language descriptions of images, both general and at user request. In recent years, information systems and catalogs have been created on the Internet to search for information in repositories containing data of various natures - texts, images, sound. In this case, the search is carried out according to queries in different languages. An example of this information system is international information network cultural heritage in Europe, created by the BRICKS community (BRICKS, 2004). The system includes searching for information about museum exhibits, including works of fine art. Searching for information about images is also relevant for photograph museums. In addition to searching by bibliographic data, it is relevant to search for images by their visual parameters and by the content of the image, i.e. according to the objects depicted.

To describe the content of images in collections, three means are used: text annotations in Natural Language (NL), keywords and a list of objects.

Text annotations have three disadvantages: a) the use of a specific NL complicates interlingual search; b) expressions in NL are often ambiguous; c) to compare the query text in NL and the search image, non-trivial operations are required. For example, (Emeljanov et al., 2000) proposes to search for images using text annotations, and to overcome the variability of analysis by performing equivalent text transformations. This method involves solving two very complex problems - creating modules for syntactic analysis and semantic interpretation of free text sentences, as well as a device for paraphrasing free text sentences. These tasks remain at the research stage.

A set of keywords is the simplest and most universal form of describing the content of documents, including images. This form is used everywhere, including in BRICKS. It allows you to overcome multilingualism by defining correspondences between descriptors in different languages. Typically, a collection has its own, sometimes semi-structured, indexing scheme (Hollink et al ., 2003). However, keywords specify only the topic of the document, not its content.

The list of depicted objects is used in the State Historical Library in Moscow. This is a more meaningful description than keywords, but it does not allow you to search for groups of objects that are in a certain connection with each other, for example, two-story country houses by the river, a man sitting near a table in a frock coat and bowler hat, etc. Expanding lists of objects to coherent semantic annotations of image content would make it possible to organize a search for related objects in an array of semantic annotations.

The article presents the results of a study of the principles of constructing formalized semantic annotations to describe the content of images. The means of description are concepts denoting types of objects, relationships between objects, as well as properties of objects and their meanings. The concepts are collected into an experimental model of the Ontology of Concepts of the Visible World created by the authors.

Ontology concepts are elements of semantic annotations of images represented in the XML formalism. The material for this study was photographs of urban and rural areas, natural landscapes, taken from the ground or from a small elevation - a bridge, mountain, roof of a house, etc., collected by different people from different sources- print, internet, home archive.

Semantics of the image

We will classify images of open space as a class of images, which we will call “scenes of discrete objects” (DSO). This class is characterized by the fact that behind the image there is a structure of depicted objects such as a network of frames, which organizes the objects participating in it into a certain rational scene. This scene can be described in terms of objects and their property values, as well as spatial relationships between objects. When creating a semantic annotation of an image, we use the concepts that people use when describing an image in NL. Those. the descriptions discussed are anthropocentric

Experience shows that different people, when describing a picture, pay attention to different objects, therefore, EL descriptions of the same photograph by different people can differ significantly, i.e. Human descriptions of pictures are subjective. Here, for example, are two descriptions of the content of the photograph shown in Fig. 1:

(Description 1) The picture shows a summer day after a thunderstorm.

In the foreground is a river. Judging by the shooting location, the photo was taken from a boat or bridge. The banks are flat, maybe it’s not a river, but a canal. The storm is most likely either about to begin, or has just ended: gaps are visible in the cloudy sky, the grass is illuminated by the sun. A rainbow is visible in the sky.

On the left on the bank there is a large three-story house with a red roof.

(Description 2) The painting depicts a rural landscape: a river, its two banks, a gray sky. On the left bank of the river there is a small white house. There are no other houses visible; the house stands alone among the green trees. There is a boat near the left bank. A multi-colored rainbow breaks through the gray clouds.

In Description 1, the house is called “large”; apparently the author was thinking about country houses, for which a three-story house is large. In Description 2 the same object is called a “small house”. Perhaps the author is a city dweller and for him all country houses are small. In Description 2, among the objects depicted, a boat is mentioned on the left bank, in Description 1 - not. Both descriptions, starting with the introduction of the painting, then list objects, starting with the river, which occupies a central place in the painting. But then opposite strategies are implemented: in Description 1 the sky is discussed and ends with the house, in Description 2 it begins with the house and ends with the sky.

The task of creating the most objective description in the form of a semantic annotation involves listing, if possible, all depicted objects that are potentially interesting to the user, based on objective data of different types, in particular:

– real relative the size of an object in an image, for example, the portion of the image area occupied by the object relative to the area of ​​the entire image or other object;

Perspective - a decrease in the size of objects as they move away from the observer;

Knowledge of the design, relative sizes (e.g. compared to human size) and functionality of objects.

The differences in the descriptions given in the previous example are due to the choice of different presentation strategies and the fact that different people relied on different knowledge about its subject area (DO) when describing the picture. Therefore, we can assume that descriptions of a picture are objective relative to its software. We present open space software in Ontology.

Ontology

The proposed ontology model presents the concepts used to describe the objects that make up images of open spaces, listing the types and properties of objects distinguished by humans and the relationships between them. The concepts of an ontology are defined by its articles. The ontology article consists of the following sections:

title– matching the ontology concept with an identifier word that denotes the type of object in the image (for example, concepts house wall, man, photography);

informal definition – text description in NL of the content of this concept;

genus– the concept of a given ontology, which is generic in relation to the given;

view– listing the concepts of this ontology, which are specific in relation to this;

attribute values, which are fixed in the ontology article for a given concept,

ontological relations – represent knowledge about the structure and functionality of objects, as well as about the typical arrangement of objects (for example, part, place) ;

consistent visual properties the image of the object itself, for example, image, texture, shape, color scheme, are so far indicated only sometimes in the form of an informal comment. Increasing interest and attempts to create an ontology of image semantics, for example, (Hunter, 2001), (Beloozerov et al., 2005), allow us to hope for the development of research on object recognition in an image in the future.

The ontology contains a hierarchy of open space concepts for constructing annotations of image content. Her articles are completely independent of linguistic semantics, because they contain concepts represented by images. At the root of ontology there are three types of concepts: - image, relation, attribute. Image- exactly this general concept, which includes the picture itself and its parts and the types of real-world objects depicted in the picture. Below are several articles of the ontology.

Relationshipinclude ontological and spatial relationships. Spatial relationships (eg. right, above), convey the relative position of objects in the image from the point of view of the observer. They are used in scene descriptions - semantic annotations of images.

Attributes describe the properties and states of objects. State information can be obtained from visual image analysis, e.g. standing (person), sitting (cat, bird). Thus, the types of attributes are expanded compared to the set that is usually used in semantic research. In addition to the “classic” properties or attributes, for example, color, size, attributes include concepts usually expressed by verbs, for example, running. In the Ontology they are represented as object type attribute values pose.

The root object of the ontology is the concept of “image”:

(IMAGE

(DEFINITION(“Image as a unity of medium and content”))

(GENUS (*))

(VIEW(IMAGE-CARRIER IMAGED)))

(IMAGE-MEDIA

(DEFINITION("Image as a collection of pixels"))

(GENUS(IMAGE))

(VIEW(PHOTO PICTURE))

(PART(IMAGE-BORDER PART-IMAGE (center, upper left corner, etc.) IMAGE-FRAGMENT LINE))

(PICTURED

(DEFINITION("An image perceived by vision on a flat surface - a painting, a photograph, as content - the configuration of the presented objects."))

(GENUS(IMAGE))

(VIEW(OPEN-SPACE (landscape)

CLOSED-SPACE (domestic scene)

IMAGE-OBJECT (still life)

IMAGE-HUMAN (portrait) SPECIAL-IMAGE (medical, for example, cytological, preparation, photograph of the Earth’s surface from space)))

(PART(GEO-OBJECT OBJECT))

(GEO-OBJECT T

(DEFINITION"planetary-scale objects that are ontologically present in any image of open space")

(GENUS(PICTURED))

(VIEW(SURFACE-EARTH SKY HORIZON))

(GROUND SURFACE

(DEFINITION("Part of open space on which all OBJECTS are located))

(GENUS(GEO-OBJECT))

(VIEW(DRY WATER))

(WHOLE(OPEN SPACE))

Object classification includes everything that may be in the image - houses, ponds, roads, lawns, cars, plants, sun, etc. Descriptions of the contents of the paintings used the common language ontology WordNet (WordNet), for example, (Hollink et al.). Our ontology contains fewer concepts than general language ontologies; for example, there are no abstract concepts like idea, fate, smart, etc., but there may be abstract attributes like age, since they are related to appearance object in the image. It also lacks verbs whose meanings are to a small extent covered by the attribute values pose. These restrictions are natural - they are imposed by the visual way a person perceives the world in natural conditions.

WordNet already contains many of the concepts that need to be included in the open space ontology. However, the articles describing them cannot simply be copied. WordNet follows the philosophical tradition (Vossen, 2003), i.e. speculative rather than visual contemplation of the world. They contain a lot of unnecessary information and do not contain some necessary information. For example, of the seven meanings of the noun Earth (earth) in WordNet, only two correspond to the concepts we need, these are:

[ S: (n) earth, ground (the loose soft material that makes up a large part of the land surface) "they dug into the earth outside the church" ].

[ S: (n) land, dry land, earth, ground, solid ground, terra firma (the solid part of the earth"s surface) "the plane turned away from the sea and moved back over land"; "the earth shook for several minutes"; "he dropped the logs on the ground"]

The first denotes earth as a material, e.g. a bunch land . The second turns out to be too vague, in particular, the example of use given in WordNet:

“has instance

S: (n) America (North America and South America and Central America) “

may refer to software “imaging from space”, but not from the ground. The indicated direct specific concepts of the second meaning, including such concepts relevant to the software as “open space visible from the ground” as:

S: (n) beachfront (a strip of land running along a beach) S: (n) coastal plain (a plain adjacent to a coast),

and indirectly:

S: (n) forest, woodland, timberland, timber (land that is covered with trees and shrubs),

do not include many relevant concepts, e.g. , road, lawn and so on. as well as all bodies of water (water) – river, sea, pool etc., which also represent the surface of the Earth visible to humans. Many of these concepts are represented in WordNet by their functional aspect and thus do not contain any indication that it is the surface of the Earth, such as a road:

S: (n) road, route (an open way (generally public) for travel or transportation)

S: (n) road (a way or means to achieve something) "the road to fame"

Our ontology is closer to ontologies that solve problems related to representing the meanings of a specific text. These include, in particular, ontologies for knowledge-based machine translation (KBMT), for example, (Nirenburg, Raskin, 2004). A distinctive feature of such ontologies is the presence of syntagmatic semantic relations - those that convey the relationships of objects in descriptions. These are actant relations such as Actor, Goal, relations between objects, for example, Localization, etc. Our descriptions present a “pre-linguistic” picture. They do not contain predicates and relations with participants, but do contain ontological and spatial relations, e.g. part, place, above, right.

The initial version of the Ontology, compiled for 20 images, has up to 20 attributes, 10 relationships and 100 objects.

Description of the scene using ontology concepts

In the description of the scene, ontology concepts are compared with their instances - objects, prototypes of real participants in the scene. Objects are connected by relationships.

N.N. suggested using the apparatus of conceptual (ontological) relations to describe content and, in particular, scenes. Leontieva (Leontieva, 2006). For example, the following set of two conceptual relations with filled slots:

place (road, car)

right (car, person)

describe a scene that can be expressed in NL in different ways, for example, “There is a car on the road. There is a man to her right,” or “There is a man standing on the road. To his left is a car." The predicate "stand" to describe a given representation can be derived from information about the person's posture.

The image annotation is anthropocentric, i.e. is focused on the concepts that people use to describe images. A person perceives the image immediately and in its entirety and understands how its content is structured. Descriptions in EA usually begin with general characteristics images, for example, in (Description 1) - “summer day after a thunderstorm”, in (Description 1) - “village landscape”. This overall characteristic is the result of a person's instantaneous generalization and assessment of the scene. However automatic system it can only be given after a detailed analysis of the semantic annotation. Our annotations include, whenever possible, all depicted objects in which the user might be interested, their attributes and relationships.

When describing an image, a person combines the depicted objects into groups in accordance with the composition he understands. To describe the composition we will use the concepts that a person uses. A typical way to reflect a composition is to use plans ( layer ) images. Plans are groups of objects located at approximately equal distances from the observer. The concept of plans is used in almost all descriptions of photographs. The last, farthest plan from the observer is the general background (for example, the sky, a wall or a forest). So, in the photograph shown in Fig. 2, the transparent inscription “Rosfoto Image collection” is located in the foreground, the sidewalk and the person walking on it are in the background, the background is the wall.

Plans are carried out invisibly horizontal lines in the image, in which there are groups of objects approximately equidistant from the observer. Plans can be used to estimate the size of objects from perspective. In addition to plans, multiple objects are used to describe collections of objects, such as forest, and groups - objects that subordinate other objects in the annotation, for example, group of people (person-1, person-2, ... person-n), house (roof, windows, doors). Between the root of the group, for example, house and its subordinate objects, for example, roof, ontological relations may exist, in this case, Part. Spatial relationships can be established with individual objects of the group and with the group as a whole.

Objects play different roles in describing the content of an image. Large objects (usually geo-objects) elongated in the direction of the observer’s gaze, for example, a road, a river (see Fig. 1), can participate in the formation of the composition of the image. Typically, such an object is located close to the center of the image and divides it into three parts - the object itself, objects to the right and left of it. In Fig. 1 is a river (or canal), left bank and right bank. Group objects can also be organized according to plans, for example, in the description of the image content in Fig. 1 left bank can be divided into three plans: a bank with a steepness of 20%, overgrown with grass, on it is a three-story house with a red roof, a tree to the left of it (1st plan); trees (2nd plan); a bank covered with sand with a steepness of 5 %, trees on it and a boat on the water near the shore (3rd plan). The sky with rainbow refers to the background of the entire image.

The purpose of annotation is not to create an absolutely complete description, but to present those objects and relationships between them that may be of interest to the user. If the description is not detailed enough, the abstract may be clarified. This will not require a complete rewrite of the annotation. For example, attributes and dependent objects can be added to a group object without affecting the rest of the image. Numerical characteristics can be entered into the description, for example, estimates of the size of the object, the steepness of the coast or cliff, as done above. Below is an XML approximate representation of the contents of the photograph shown in Fig. 1:

xml version="1.0" encoding="windows-1251"?>

- < picture id=" picture1">

< v-object concept=" RIVER" id=" river-1" />

- < left-v-object concept=" SHORE" id=" shore-1">

- < layer id=" layer-1">

< property concept=" COATING" meaning=" GRASS" />

< property concept=" STEEPness" meaning=" 20 " />

- < object concept=" HOUSE" id=" house-1">

< property concept=" COLOR" meaning=" white" />

- < object concept=" ROOF" id=" roof-1">

< property concept=" COLOR" meaning=" RED" />

object>

- < object concept=" ROW- WINDOWS" id=" floo-1">

< property concept=" QUANTITY" meaning=" 3 " />

object>

< object concept=" tree" id=" tree-1" />

object>

layer>

- < layer id=" layer-2">

- < object concept=" TREE" id=" trees-1">

< property concept=" QUANTITY" meaning=" SOME" />

object>

layer>

- < layer id=" layer-3">

< property concept=" COATING" meaning=" SAND" />

< property concept=" STEEPness" meaning=" 5 " />

- < object concept=" BOAT" id=" boat-1">

< property concept=" COLOR" meaning=" WHITE" />

object>

layer>

left-v-object>

- < right-v-object concept=" SHORE" id=" shore-2">

< property concept=" COATING" meaning=" THRUGS" />

right-v-object>

- < layer-last id=" layer-last-1">

< object concept=" FOREST" id=" forest-1" />

- < object concept=" SKY" id=" sky-1">

- < object concept=" CLOUD" id=" cloud-1">

< property concept=" QUANTITY" meaning=" SOME" />

object>

< object concept=" RAINBOW" id=" rainbow-1" />

object>

layer-last>

- < relations>

< relation type=" left" domain=" house-1" range=" tree-1" />

< relation type=" contact" domain=" shore-1" range=" boat-1" />

relations>

picture>

Conclusion

The article formulates the principles of creating semantic annotations of image content. It is shown that following the anthropocentric principle it is possible to create a semantic annotation of the image content based on the ontology of concepts of this software. The means of description are also the grouping of objects, the establishment of spatial relationships between them and the use of attributes that represent, in particular, some physical properties of the object in the image, for example, steepness (banks, rocks). An annotation allows you to clarify the properties and composition of objects by adding properties and objects - members of a group of objects or parts of an object. Annotations can be used for information retrieval and for searching for new information in image arrays, as well as for automatically generating descriptions of images in NL. The ontology provides the possibility of comparing stable visual properties of depicted objects to concepts. This would make it possible to subsequently explore the possibilities of recognizing objects in images and propose a formalization of rules for constructing semantic annotations of image content.

The work was carried out with partial support from the Russian Humanitarian Science Foundation, grant 04-04-00185a and the Russian Foundation for Basic Research, grant 04-07-90187v.

Bibliography:

1. Leontyeva N.N.Automatic text understanding: systems, models, resources.M.: "Academy", 2006.

2. Beloozerov N.V., Murashov D.M., Trusova Yu.O., Yanchenko D.A. . Searching for solutions in the image analysis and processing knowledge base // Pattern recognition and image analysis: Advances in mathematical theory and application. – MAIK “Nauka/Interperiodica”, 2005. – Vol. 15, No.2. –P. 361-364.

3. BRICKS Project. Building resources for Integrated Cultural Knowledge Services © 2004. // European Commission Sixth Framework Program (http://www.brickscommunity.org/)

4. Emeljanov G.V., Krechetova T.V., Kurashova E.P. Tree grammars in the problems of searching for images by their verbal descriptions // Pattern recognition and image analysis. – 2000. Vol. 10, N4. P.520-526.

5. Hollink L., Shreiber G., Wielemaker J., Wielinga B. . Semantic annotation of image collections // S. Handschuh, M. Koivunen, R. Dieng, and S. Staab, editors, Knowledge Capture 2003. Proceedings Knowledge Markup and Semantic Annotation Workshop, Florida, USA, October 2003. P. 41-48 .Also available on the Internet:

6. http://www.cs.vu.nl/~guus/papers/Hollink03b.pdf

7. Hunter J. Adding multimedia to the semantic Web – building an MPEG-7 ontology // Proceedings of International Semantic Web Working Symposium (SWWS), Stanford, July 30 - August 1, 2001

8. ( http://archive.dstc.edu.au/RDU/staff/jane-hunter/semweb/paper.html) .

9. Miller G.WordNet: a lexical database for English. Communications of the ACM, 38(11):39–41. WordNet: (http://wordnet.princeton.edu/)

10. Nirenburg, S., Raskin, V. Ontological Semantics // Language, Speech, and Communication. -MIT Press, 2004.

11. Vossen P.Ontologies // The oxfor d handbook of computational linguistics. Ed. by R. Mitkov, Oxford univ. Press. 2003. p. 464-482.

Several interior options for the updated Lada Vesta have appeared on the Internet. The blogger showed the images in one of his videos VsyakoRazno. According to his subscriber, who sent the renderings, the sketches were made as part of a closed presentation in St. Petersburg for AvtoVAZ dealers. The event included a discussion of the Lada Vesta 2020 model year project.

Previously, “RG” was the interior of the updated Lada Vesta, but the image was not clear, but now the renderings have become available in higher quality.

It became possible to examine in detail two interior design options at once. The first is made in black and orange style, the second - in black and beige. It is worth noting that this possible options new salon, which give a general idea of ​​the changes in the future modernization.

After the facelift, Lada Vesta will receive a completely new front panel, door panels and central floor tunnel. A feature of the cabin will be a huge “floating” tablet of the multimedia system, which is combined with the air conditioning unit. At the bottom of the center console there is a large niche for a smartphone (wireless charging option will be available).

The design of the manual transmission lever has been changed, and near the closed section with cup holders, as on cars of a higher class, you can see an electronic handbrake button and, presumably, a selector washer for selecting the Lada Ride Select driving mode. The seats and armrest remained the same, but the steering wheel has changed in design and has become smaller in diameter, the instrument cluster now has two wells and a large on-board computer display. The shape of the deflectors, door handles and decorative elements has also changed.

Photo: screenshot from www.youtube.com

In addition to a completely new interior, according to the canons of the facelift, the Lada Vesta 2020 will receive a fresh exterior - new bumpers and radiator grille, a different design of front optics and rear lights, as well as wheels with a modified design.

The appearance of the modernized Lada Vesta Facelift is expected in 2020, and not next year, as previously reported.

Let us remind you that at the end of 2019 Lada Vesta will receive a Japanese-made Jatco JF015E variator. It is known that AvtoVAZ is actively working on the Lada Vesta FL project, which means that the new gearbox will debut on the facelift model.

By 2020, the Vesta engine range will be represented by three power units and the same number of gearboxes. VAZ 1.6 and 1.8-liter engines with 106 and 122 hp. Accordingly, they will not go anywhere; they will be supplemented by a 114-horsepower Nissan HR16 1.6-liter engine. A 5-speed manual transmission (MT), as well as a robotic transmission (AMT) and a CVT (CVT) will be available to choose from.

Let us remind you that serial production of the Lada Vesta started in September 2015 at the Lada Izhevsk plant. Today the family consists of the Vesta sedan, the Vesta SW station wagon, their “cross” modifications, as well as the dual-fuel Vesta CNG. The appearance of "charged" and . The range of engines includes two VAZ naturally aspirated engines with a volume of 1.6 and 1.8 liters, the first develops 106 hp, the second - 122 hp. The engines are mated to a 5-speed manual transmission or automated manual transmission.

Research into the relationship between optical and semantic properties of images

^ Investigations of relationship between optical and semantic properties of images
The goal is to study the relationship between the optical and semantic properties of images in object classification problems. To achieve the goal, electrophysiological methods were used - cognitive evoked potentials, their amplitudes, are largely determined by the physical properties of the stimuli. Therefore, we applied the methods of icons - digital processing and image synthesis. 90 test images were synthesized. Of these, half belonged to one class of objects, and the other to another, i.e. had different semantic meanings for the observer. Then we used bandpass wavelet filtering, optimal for our tasks, in the high and low frequency ranges. We received two sets of tests. The spectrum of 90 images was shifted to the region of high spatial frequencies and for the same 90 images to the region of low spatial frequencies. The images were presented on a professional Sony cathode-ray monitor. Corrected the gamma function of the screen. To record cognitive evoked potentials, an encephalograph “Mitsar-EEG-201” was used. Processing was carried out using the WinEEG program. The electrodes were placed according to the 10-20 system. Observers - 43 volunteers were aged from 20 to 38 years. Based on the calculation of the statistical significance of the components from the stimulation characteristics, significant differences were established in the amplitude of the components of cognitive evoked potentials for stimuli that differ in the spatial frequency range in almost all leads. When the observer was tasked with distinguishing stimuli by semantics, and comparing responses to stimuli that also differed in semantics, we identified differences in the amplitudes of evoked potentials in the frontal and temporal cortex. If we change the task for the observer - to distinguish blurred images from non-blurred ones, but during processing compare the differences in the amplitudes of cognitive evoked potentials to stimuli that differ in semantics, then differences in amplitude will be revealed in the parietal and temporal cortex. Conclusions: a different ratio of the amplitude and phase of the main components of cognitive evoked potentials was established in different leads, to stimuli that have different semantic meaning and have different optical characteristics (spatial-frequency spectrum). By changing the task assigned to the observer, it was possible to identify the parameters of the components of cognitive evoked potentials, reflecting the optical and semantic properties of the images.

^ Muzychenko Ya.B.

St. Petersburg National Research University ITMO

Gnomons and megaliths: artistic symbols or sighting devices?

Gnomons and Megaliths: art-objects or viewfinder devices?
Gnomons are the first optical instruments for orientation in space and time. They were used both as stationary structures and as portable devices. The purpose of the gnomons was to record in time special astronomical events determined by the location of the Sun in the sky, and to determine the current time with an accuracy of several minutes.

There are known images of priests of Mesopotamia holding poles with ring-shaped objects. The symbols of the Sun and Moon, depicted there, make it possible to interpret these devices as sighting devices, in the form of a closed or semi-closed aperture mounted on a high rod. Devices of this kind, as a rule, were symbols of the gods, which indicates the crucial role that the ancients attached to observing the sky. The variety of symbolic staffs and signs worn on poles and their specific forms leave no doubt about their purpose for sighting astronomical objects.

Sumerian upturned rods resemble Egyptian depictions of astronomical observing instruments, which consisted of a semicircular element placed between “horns” on the top of a tall pole. Since observing the Sun was one of the central sacraments of the Egyptian religion and an important occupation of its priests, it is not surprising that images of ancient Egyptian gods and goddesses often included symbols of various viewing devices. Solar or lunar discs in combination with sickle-shaped, horn-shaped or stepped elements decorated their headdresses. Ancient Egyptian U-shaped headdresses are usually interpreted as cow or calf horns.

A typical example of sighting devices is the iconic symbolism of the Minoans and Carthaginians. The decorative decorations of the Minoan “horned” altars with double labrys axes indicate their obvious visual origins. They repeat the shape of the sighting apertures of Egypt and Mesopotamia, and the use of labryses as markers gives them angular accuracy in orientation

The archaic period became the time of the appearance of portable sighting systems. If gnomons are a reliable example of ancient devices for sighting, then one can only make guesses about other sighting devices, the images of which have survived to this day. Over time, such devices lost their visual function, remaining only as a decorative element. The report examines possible use symbols of ancient cultures as the first sighting devices.

^ Nesteruk A.V.

University of Portsmouth, England

Cosmology and optics: the history of one experiment

Cosmology and Optics: the history of one experiment
Modern cosmology teaches that the observable matter of the Universe, that is, what emits in the wavelengths possible for reception in the optical and radio ranges, makes up only 4% of the total mass of the Universe. A person lives in an island of light visible to him from objects in the surrounding world. The universe is "dark" and "hidden" not only because of forms of matter that do not interact with anything. But the Universe is “visible” only up to certain point in the distant past, when it became transparent to light. The time strip that separates everything that we see from what this “visible” was formed from is called the Epoch of separation of matter from radiation. We are able to receive signals from the past of the Universe only after this Epoch. Can cosmology hope to study physical processes from the Big Bang?

Theoretically, it is possible to use optical methods to detect signals from “invisible” gravitational waves coming from the past, from behind the optical horizon. The idea is simple: gravitational waves cause microscopic deviations of “geodesic lines”, which can be detected using a laser interferometer if its mirrors are in a gravitational field in a state free fall and as the deviation process manifests itself, the relative displacement of the interferometer mirrors can be detected. To experimentally test this at the State Optical Institute named after. S.I. Vavilov in the early 80s of the twentieth century, an attempt was made to build an unequal-arm interferometer with an optical path length of 750 meters. The delay of the signal at such a base due to the passage of a gravitational wave could make it possible to detect a relative displacement of the mirrors of ~10 -15, but this required stabilization of the frequency of the emitting laser at the same level, which was practically impossible in those years. There was no experience of such experiments not only in the USSR, but also abroad. GOST standards for the manufacture of mirrors for a multi-pass interferometer system did not guarantee a flat profile of their surface with an accuracy of 10 5. Air fluctuations in the optical waveguide had to be reduced to a level impossible for the pumps of those years. As a result, the level of technological precision and the obvious lack of funding (and today the price of such projects is 40 times higher) made this ambitious experiment unrealizable. But the first attempt to create a gravitational wave detector showed the amount of work required and the fact that the progress of optics depends on the tasks of not only optical astronomy, but also theoretical cosmology. “Unobservable” physics becomes the engine and inspirer of the science of light and its “vision”.

^ Novikov G.I.

Visual mechanisms for converting optical information into eye movement control commands

Visual Mechanisms of Optical Information Transform in to the Commands of Eye Movement Control
In order to study the mechanisms of foveation (gazing at a target), electrophysiological experiments were carried out, combining both traditional methods for studying the characteristics of receptive fields during optical stimulation and methods for studying the neurophysiology of eye movements. Subcortical visual centers mammals are known as key oculomotor structures initiating saccadic eye movements. In the work, based on experimental electrophysiological data, the author formed a concept of mechanisms for converting optical stimulation into commands for controlling eye movements in these structures. The superior colliculus of the cat, which has direct inputs from retinal ganglion cells into the superficial layers and simultaneously contains neurons in the deep layers that have their own motor fields, was chosen as an experimental model. The spatiotemporal characteristics of the receptive fields of visual neurons were usually experimentally analyzed in a completely immobilized animal. Eye movements in this case are impossible. A way out of this dramatic situation was made by a group of researchers who implemented a method of immobilizing one of the animal’s eyes, carried out through an elegant surgical operation to cut the oculomotor nerves. The cooperative nature of eye movements was taken into account during the experiments and led to repeatable and unexpected results. The directional and orientational characteristics of the receptive fields of single neurons were studied on the immobilized eye, and eye movements were recorded on the other eye. When comparing the results of the experiments, it was found that local electrical microstimulation of neurons with pronounced directional tuning causes parallel-type eye movements, in which the direction of eye movements does not depend on its initial position. With electrical stimulation of neurons that do not have directional properties, goal-directed eye movements occurred, in which the direction of movement depended on the initial position of the eye. The direction of movement of an object determines the nature of involuntary saccades and is the organizing element of the foveation mechanism of new objects appearing on the periphery. Proposed possible model such transformations within the framework of mathematical set theory.

^ Novikov G.I.

Institute of Physiology named after. I.P. Pavlova RAS

Three main types of maps of eye movements caused by local electrical stimulation of subcortical brain structures.

Three main type maps of eye movements induced by local electrical microstimulation of subcortical zones of the brain
The key issue in the organization of eye movements is the study of the relationship between optical and visual space and the organization of eye movements that ensure the translation of gaze to an area of ​​interest in the surrounding world. To solve this fundamental problem, electrophysiological studies on cats using electrical stimulation of the projection zones of the subcortical structures of the cat's brain recorded three main patterns of eye movements, called eye movement maps. The temporal and spatial characteristics of these eye movements were quantitatively studied. With electrical stimulation, eye movements were friendly and ordered. The type of map depended on the retinotopic coordinates of the stimulated locus of the cat's superior colliculus. The direction of eye movements recorded during stimulation of neural pools located in the peripheral zones of retinotopic projections always corresponded to the direction from the central part of the retina to the centers of the peripheral receptive fields of the stimulated neurons. The centers of overlapping receptive fields in the recorded and stimulated pools coincided.

The first type of movements (35% of the total) includes maps of eye movements with parallel trajectories. The direction of movements in this case did not depend on the position of the eyes in the orbit at the time of stimulation. We designated them as “parallel” type cards.

The second type of maps (42%) included maps of eye movements in which the trajectories of movements converged towards a certain area - the “target” zone. This type was designated as "targeted".

The third type included maps of eye movements, in which movements were directed from the peripheral projection areas of the retina to its central part. We have designated this type of map as “centered.”

Each of the three types of eye movement maps is largely determined not only by retinotopic projections into the stimulation area of ​​the corresponding subcortical structure, but is also associated with the directional characteristics of the receptive fields of the stimulated neurons.

The data obtained make it possible to model and thereby explain the neurophysiological mechanisms of foveation and visual attention.

^ Okladnikova E.A.

RGPU named after. A.I. Herzen

Cosmology in petroglyphs: celestial objects in the rock art of Eurasia (semantic analysis)

Cosmology in petroglyphs: selestial objects in the rock art of Eurasia (semantic analysis)
Applied optics has a long history. Its development was unthinkable without: 1) observations of the stars, 2) without the formation of cosmology as a system of scientific views. The Stone Age dates back to constantly improving visual-optical techniques for studying the surrounding world. Observations of celestial bodies found expression in such inventions as: paleocalendars, magical objects (“magic” Chinese mirrors, crystal magic spheres, magic crystals, shamanic drums, etc.), numerous archaeologically recorded processes (for example, burial of the deceased with the orientation of bodies in accordance with the countries of the world), artifacts different types, which now constitute the subject field of astroarchaeology. The tools of field calendar calculations were cromlechs, menhirs, steles, “shaman rings, seids and their varieties, stone labyrinths, and in a certain sense, rocks and cavities of grottoes and caves with rock paintings.

The main monuments of the Stone Age, which preserved evidence of visual observations of the movements of celestial bodies by Stone Age hunters, are: Paleolithic caves of the Franco-Cantabrian region. A series of unique rock temple complexes in the territory of Siberia, Transbaikalia and the Amur region, which are located in the open air, were discovered during archaeological research in the twentieth century (Mount Kalbak-Tash in Altai (used as an astronomical observatory, drawings of cosmological content and celestial signs were preserved on the surface -symbols); Srednyukzhinskaya pisanitsa in the Amur region (a grandiose cosmological composition with images of the sun, moon, stars); Ak-Baur grotto (Western Kazakhstan).

Conclusions: Optical observations and their immediate objective results were subject to mystical rethinking and became material for the formation of various mythopoetic and mystical cosmological traditions of antiquity. Celestial phenomena objectively observed with the help of simple astronomical instruments, including solstices, eclipses, the movement of comets, etc., received not only mythological reinterpretation, but artistic embodiment in monuments of ancient art.

^ Orlov A.V.

St. Petersburg National Research University information technologies, mechanics and optics

Optics in Russia in the 1890s - 1905.
Optics in Russia in the period 1890 and until 1905 was relatively widespread. The level of this distribution depended mainly on the consumption of optical media in the circles of specialists and scientists. Ophthalmology - eye optics has become one of the most striking industries, where the successes of domestic optics have been the greatest in comparison with other industries. However, there was not a single branch of application in which optics independently developed in its own original way, except perhaps for ophthalmology, which can be mentioned. The microscope, which had already gained worldwide attention, found widespread use in Russia, although this use was limited to the scientific and special research environment (medicine, biology, chemistry). The microscope for this time was considered the highest achievement of technical optics. Along with microscopy, increasingly noticeable developments took place in the field of photography. The use of a camera has already become firmly established in the everyday life of many Russians, and photography has been a constant success. Photography must be recognized as having a very special place in the development of optics in Russia. And precisely in this area, photographic portraiture. Russian amateurs reached heights that were in no way inferior to Western achievements. On the contrary, photography developed in Russia in a very original way and, in the person of individual amateurs, reached the level of scientific perfection. It was in Russia that the technique of photography and photographic portraiture made its national contribution to European photographic art and photographic equipment. This was a time when photography had not yet become a mass phenomenon. The camera required a static installation. The era of portable cameras was just approaching. Photography required scientific and special knowledge in the field of chemistry and physics, extensive practical experience and personal observation. But there is no doubt that without basic knowledge about the phenomena and properties of light it was impossible to master and use optical instruments. Astronomy can demonstrate the influence of optics on the development of science even more “visually” than photography. Interest in the starry sky, the Sun and the Moon in Russia was no less great than in the West. Observing the sky required long, constant, and most importantly systematic work with astronomical instruments and instruments of the highest precision and complexity, which was then achieved by the successes of optical mechanics. It can be argued that the circle of astronomers and photographers, biologists, ophthalmologists and topographers was not so wide as to say that optics in Russia really had any noticeable development. However, it was precisely because of the need for optical means that interest in optics and understanding of the tasks and problems associated with it intensified in Russia. This interest inevitably raised the problem of optical production in the Empire.

Military affairs had a very special demand for optics. In this area, it was not limited to spotting scopes and binoculars. Fields such as topography and geodesy, which immediately became the focus of military attention, developed in connection with the interests of the latter and were the least known of other fields to the general public. The oldest devices in this industry, theodolites, have evolved. The defensive importance of optics promised it great prospects in connection with the advent of rapid-fire artillery and the development of mobile lighting technology (searchlight technology). A whole branch of special optics has emerged - optical instrumentation for the fleet, on ships - rangefinders and optical sights, various kinds of rangefinders for the ground army for the purposes of observation, reconnaissance, calculating the distance to the target and coordinating artillery fire. The army received various types of sighting devices for the needs of artillery - sighting tubes mounted on guns, measuring devices and, finally, optical panoramas that allow the gun to be aimed at the target directly during the firing process. In connection with the evolution of the optical instrument was the evolution of mechanics, whose importance was constantly growing. The level of mechanical work became increasingly refined and the requirements for precision continually increased. The devices had to meet new requirements: greater portability than before, operational efficiency and quick replacement, reduction in weight and size, and operation in different weather and climatic conditions. In general, the emergence of rapid-fire artillery not only increased the tempo of battle and the “mobility” of fire, but also most directly affected new requirements for accuracy and efficiency of shooting, observation of the enemy in different conditions of the day and night, and the possibility of using optical instruments and reconnaissance instruments for observation. terrain, constantly improving officer capabilities. Binoculars are becoming lighter, and the magnification ratio is increasing, protractors and calculating rulers are being improved, special safety glasses are appearing, and ordinary Galilean tubes are being replaced by stereo tubes. Almost all sighting instruments, under the influence of the introduction and spread of optical means, are experiencing changes in favor of greater accuracy and reduction in their size. At the same time, manufacturing technologies are also changing, primarily materials: aluminum, copper, bronze, the use of which should reduce the weight of the device without reducing the strength and tightness of the structure.

Optics in Russia, however, was characterized by extremely weak development of the national industry. With the exception of the company I.Ya. Urlaub, who had independent achievements in the design of ophthalmological instruments, and the Warsaw company “Fos”, which was just beginning its development, operated in Russia only repair shops and trading houses engaged in the distribution of optical products for individual or departmental orders. The Phos partnership under the company Alexander Ginsberg and Co. or Russia's first optical-mechanical plant Phos (founded in 1899) undoubtedly had merits that were still so little appreciated in the history of domestic optics. The Phos company also had an “optical glass factory,” that is, it processed raw optical glass, mainly engaged in the manufacture of lenses. This company is credited with producing the first photographic lenses and cameras in Russia based on German models.

Finally, the problem of personnel for optical mechanics raised the question of a state program for the development of this specialty before the Empire. Initially, watchmaking was combined with an optical-mechanical specialty as part of one department of the Tsarevich Nicholas Vocational School, founded only in 1900 in a very modest size.

The first steps towards the optical emancipation of Russia were made through the joint efforts of the state and domestic entrepreneurship.

^ Paranina A.N.

RGPU named after. A.I. Herzen

Editing images and creating collages would be a very exciting process if you didn't have to spend most of your time painstakingly marking up objects. The task becomes even more difficult when the boundaries of objects are blurred or there is transparency. Photoshop tools such as Magnetic Lasso and Magic Wand are not very intelligent because they only look at low-level features in an image. They return hard boundaries, which then need to be corrected manually. The Semantic Soft Segmentation approach from Adobe researchers helps solve this difficult problem by dividing the image into layers corresponding to semantically significant areas and adding smooth transitions at the edges.

"Soft" segmentation

A team of researchers from the CSAIL laboratory at MIT and the Swiss university ETH Zürich, working under the leadership of Yagiz Aksoy, proposed to approach this problem based on spectral segmentation, adding to it modern advances in deep learning. Using texture and color information, as well as high-level semantic features extracted, a special type of graph is built from the image. Then, using this graph, a Kirchhoff matrix (Laplacian matrix) is constructed. Using the spectral decomposition of this matrix, the algorithm generates soft contours of objects. The division of the image into layers obtained using eigenvectors can then be used for editing.

Overview of the proposed approach

Model description

Let's look at the method of creating semantically meaningful layers step by step:

1. Spectral mask. The proposed approach continues the work of Levin and his colleagues, who were the first to use the Kirchhoff matrix in the task of automatically constructing a mask. They built a matrix L, which specifies the pairwise similarity between pixels in a certain local area. Using this matrix, they minimize the quadratic functional αᵀLα with user-specified constraints, where α specifies a vector of transparency values ​​for all pixels in a given layer. Each soft contour is a linear combination of K eigenvectors corresponding to the smallest eigenvalues ​​of L, which maximizes the so-called sparsity of the mask.

2. Color affinity. To compute nonlocal color proximity features, the researchers generate 2,500 superpixels and estimate the proximity between each superpixel and all superpixels in a neighborhood with a radius of 20% of the image size. Using non-local proximity ensures that regions with very similar colors remain coherent in complex scenes like the one below.

Nonlocal color proximity

3. Semantic proximity. This stage allows you to identify semantically connected areas of the image. Semantic proximity encourages merging pixels that belong to the same scene object and penalizes merging pixels from different objects. Here, the researchers take advantage of previous advances in pattern recognition and compute, for each pixel, a feature vector that correlates with the object it contains. given pixel. Feature vectors are calculated using a neural network, which we will discuss later in more detail. Semantic proximity, like color proximity, is determined at superpixels. However, unlike color proximity, semantic proximity only links nearby superpixels, encouraging the creation of coherent objects. The combination of non-local color proximity and local semantic proximity makes it possible to create layers that cover spatially separated images of a fragment of one semantically related object (for example, vegetation, sky, other types of background).

Semantic proximity

4. Creating layers. In this step, a matrix L is constructed using the previously calculated proximities. From this matrix, eigenvectors corresponding to the 100 smallest eigenvalues ​​are extracted, and then a rarefaction algorithm is applied, which extracts 40 vectors from them, from which layers are built. The number of layers is then reduced again using the k-means clustering algorithm at k = 5. This works better than simply rarefying the 100 eigenvectors to five, since such a strong reduction in dimensionality makes the problem overdetermined. The researchers chose a final number of contours of 5 and claim that this is a reasonable number for most images. However, this number can be changed manually depending on the image being processed.

Soft contours before and after grouping

5. Semantic feature vectors. To calculate semantic proximity, feature vectors calculated using a neural network were used. The basis of the neural network was DeepLab-ResNet-101, trained on the task of predicting metrics. During training, maximization of the L2 distance between features of different objects was encouraged. Thus, the neural network minimizes the distance between features corresponding to one class and maximizes the distance in the other case.

Qualitative comparison with similar methods

The images below show the results of the proposed approach (labeled "Our result") compared to the results of the closest soft segmentation approach - the spectral mask method - and two state-of-the-art semantic segmentation methods: the PSPNet scene processing method and the Mask R-CNN object segmentation method.

Qualitative comparisons of soft semantic segmentation with other approaches

It can be replaced that PSPNet and Mask R-CNN tend to make mistakes on the boundaries of objects, and soft contours constructed by the spectral method often go beyond the boundaries of objects. At the same time, the described method completely covers the object without combining it with others, and achieves high accuracy at the edges, adding soft transitions where required. However, it is worth noting that the semantic features used in this method, do not distinguish between two different objects belonging to the same class. As a result, multiple objects are represented on a single layer, as seen in the images of giraffes and cows.

Editing Images with Soft Semantic Contours

Below are some examples of using soft outlines to edit images and create collages. Soft outlines can be used to apply specific changes to different layers: adding a train motion blur (2), separate color adjustments for people and backgrounds (5, 6), separate stylization for a hot air balloon, sky, landscape, and person (8) . Of course, the same thing can be done using manually created masks or classic contour selection algorithms, but with automatic selection of semantically significant objects, such editing becomes much easier.

Using Soft Semantic Segmentation for Image Editing

Conclusion

This method automatically creates soft edges corresponding to semantically significant areas of the image using a mixture of high-level information from neural network and low-level features. However, this method has several limitations. Firstly, it is relatively slow: the processing time for an image with dimensions of 640 x 480 is 3-4 minutes. Secondly, this method does not create separate layers for different objects of the same class. And third, as shown below, this method can fail early on in cases where object colors are very similar (top example), or when merging soft edges near large transition areas (bottom example).

Cases of algorithm errors

However, the soft edges created using this method provide a useful intermediate representation of the image, allowing you to spend less time and effort editing images.

The artistic, figurative language of the ornament is diverse. Fulfilling the task of decorative significance, it often plays the role of a social, gender and age marker, ethnicity, and is a means of expressing the people's worldview. It is appropriate to recall the social function of ornament among different peoples. For example, the patterns of the Khanty and Mansi reflected tribal characteristics; The Udmurts have names for patterns indicating that they belong to a particular clan or tribe; among the peoples of Central Asia, who for a long time preserved the remnants of tribal divisions, this was reflected in the patterns. Turkmens, especially women, unmistakably determine the tribal affiliation of carpet ornamentation. Each tribe had its own stock of carpet patterns.

The composition of the patterns, their composition, and colors differed depending on who the objects were intended for - a man or a woman, for example, on Komi knitwear, in the embroidery of the peoples of the Volga region. Age differences were also reflected in the embroidery (especially headdresses) of most peoples. The embroidery on forty Russian peasant women, marking their age and marital status, was mentioned above.

The connection between ornament and beliefs is undeniable. Among the Turkmen, for example, V.G. Moshkova identified the category of “sacred” patterns, clearly recognized by carpet weavers.

Special ornaments (gulyaids) were placed on funeral carpets; some of them had apotropaic meaning.

The karga termak pattern, “raven’s claws” (in the form of three branches), widespread among the Kirghiz, was considered a talisman; it was placed on headdresses, on the felt covering of the yurt, etc.

Among the Karelians, the ornamental image of the four-leaf clover, which was a symbol of love and marriage (lemmenlehtizet) and supposedly had special power in witchcraft, was of particular importance. Also of particular importance - protection from the evil eye - was the pattern of eight-pointed stars among the Latvians.

Much success has been achieved in revealing the semantics of Russian embroidery patterns. However, not everything remains clear. Deciphering its ancient semantics will be the task of more than one generation of scientists.

To understand the language of the ornament great importance has pattern terminology. Ethnographers pay special attention to its study. The name of the pattern can come from the technique of execution, from the material; often the term indicates the place of its origin - where it was borrowed from (Udalsky pattern - from the village of Udali; the name Mikitinsky pattern, etc., arose on the same principle).

An ornament that reflected national specifics in areas with an ethnically mixed population is often called by the name of the people for which it is characteristic or from which it was borrowed: Russian pattern, Tatar pattern, Kyrgyz pattern, etc. Sometimes the name reflects the composition of the pattern, its shape, for example, among the Mordovians - long; or indicates its location on clothing - on the sides of the chest, along the bottom, on the sleeves, etc.

The Russian terminological material available is varied. Geometric patterns in embroidery on shirt sleeves are called differently: in a stick, two berries in a stick, five berries in a stick (Kostroma province); Shashetski, klubotski, simpleton, or testicle, belts, ropes (Olonets province); towns, cubes, windows (elongated rectangles), crosses, circles (Novgorod province); Lastotskino Gnedetsko - a rhombus with extended sides (Tver province. Vesyegonsky district). In Ryazan province. the last motif is called burdock, arepey. Hooks, trestles (Vologda province), large and small loaches (Tver province); Krivonog (Voronezh province) is the name of an oblique cross with curved ends. On the Ryazan and Meshchera ponevs, the same motif was called horses, horse shanks, which gave B.A. Kuftin grounds to suggest that animal patterns were absorbed into geometric ones.

Names reflecting certain ideas about animals include cat's paws (Novgorod province), frogs, big frog (Vesyegonsky district). The fauna in the terminology of geometric patterns is reflected among all Eastern Slavs and their neighbors.

A geometrical figure resembling a person or its individual parts are also reflected in the names: in a set of heads, tadpoles - figures connected in fours (Tver province); heads-diamonds, heads and fingers - rhombuses with processes (Tver, Smolensk provinces).

Some of the names of the embroidery of the region under study reflect everyday realities: the pattern of the pinnacle or wing-feathers represents, as it were, the four wings of a mill; mostinotskam - resembles bridges, baskets woven from pine chips; the pattern of the chairs is similar to the corresponding objects (Tver province); gingerbreads are distinguished by their rectangular shape (Novgorod province), tselnotskas are reminiscent of the shuttles of a weaving mill (Olonets province); rakes and combs are similar to the teeth of these objects.

There are names of patterns associated with the celestial sphere: moon, months, circles, sun (Olonets province); Zvezdotskam (Novgorod province); clouds (Kostroma province).

Zoomorphic motifs are reflected in the following names: in single file, in two single files, large and small horses, in the skate, rich in the skate, konyastitsa, headless horse (deer), roosters, tailless cockerels, floroska cockerels (Tver province. Vesyegonsky district); cockerels, chickens, chicks (Yaroslavl province), peahen bird (Olonets and other provinces); peacock, fish (Novgorod province); eagles, with eagles (Olonets, Kostroma provinces), lion-beast (Olonets, Kostroma provinces); rokastitsa - a pattern vaguely reminiscent of the crayfish on an old woman’s magpie (Vesyegonsky district); pattern for bears (Kostroma province), etc.

Plant patterns have the following names: grass, grass pattern, gardens, trees, flowers, with light, berry patch, berries, apples, pine trees, herringbone, paws (like spruce branches), apple tree color, viburnum, rowan, wide rowan, birch , oak tree

Anthropomorphic female images are called: baba (Pskov province), podbatsen'ki heads, podbatsen'ka dolls, i.e. akimbo (Tver province. Vesyegonsky district), gossips, bags with candles (Tver province. Bezhetsky district), panya (Kostroma province), nemotski (in Veliky Ustyug weaving).

As for the complex embroidery compositions, when asked what was depicted on them, descriptive characteristics were given: “The hand-knitted towel was very good - they would create patterns with birds and apple trees”; “on a large towel they will embroider lights, birch trees, horses, roosters, and sometimes a hemlock peahen.” In Novgorod villages, women explained the composition of embroidery this way: roosters, horses and dolls, horses and birch trees, birch trees on horses, dolls and manes, roosters with dolls, nesting dolls in a dress with a bell, arms spread apart. The peculiar names of these patterns are recorded in the Onega region. Arkhangelsk province. - leshaks, peahens with leshaks.

Researchers have repeatedly noted the unstable nature of the terminology of folk patterns; the same can be said about many names of the Russian embroidery patterns of the North.

The above terminology reflects the modern vision of the creators and curators of embroidery. The names for the most part clearly reveal the basic nature of the motive. Horses that turn back are aptly named werewolves, horses with high manes are manes, feathers are the four wings of a mill, konyastitsa are horses, etc. The given terminology of the 19th and early 20th centuries. was not created at the same time. Many names are found in inventories of embroidery of the 16th-17th centuries. (berry, towns, roads, circle, trees, etc.). Others are clearly of later origin - ladies, dolls; the modern term is nesting dolls.

Some of the terms show greater stability than the pattern itself, which has been transformed. The name of the frog is found in the patterns of the Russians on the Upper Volga and their neighbors - the Karelians (letempiat), in the decorations of the Smolensk region (frogs are simple and curled), on the beaded backs of peasant women of the southern Russian provinces, as well as in the patterns of Udmurt women's bibs, in Mordovian women's clothing ( vatraki - frog). Apparently, this term is not accidental in the ornamentation of Russians and their neighbors, as can be assumed from its stability.

Some other names also deserve special attention, perhaps reflecting their former semantics: godmothers, godmothers with candles, panyas, leshaks.

The original semantics of complex ancient embroidery themes has been forgotten by the population, which was shown not only by studies of the 20th century, but was also noted in the 60-70s of the last century. It is possible to understand the original semantics of these plots only with the use of comparative data about beliefs, rituals, folklore of the Russian village of the 18th-early 20th centuries, as well as various historical evidence and ancient iconographic monuments.

The ornament is characterized by constant change; its motifs have been reworked over the course of many centuries. Nevertheless, the archaic subjects of facial embroidery, which have come down to us in many variants, allow us to trace the stability of certain features and see repeating “common places” in them. Such stable elements are the images of a woman, a tree, horsemen surrounded by birds, animals, presented in a certain interpretation, having a number of features, which allows us to see in them an archaic layer in embroidery.

All parts of the complex composition are interconnected and reflect the subordination of the characters, an appeal to the central anthropomorphic, mostly female figure (or tree), which seems to send down a blessing to others, and sometimes expresses aspiration upward - to some higher being.

Exaggerated hands and their position express the relationships between the characters. The image of hands was given special significance; they seemed to have a beneficial effect on the surrounding reality, and also protected from misfortunes. The exaggerated hands at the same time seemed to intensify the request, the appeal to the central character or to a higher being. In people's minds, a hand is a powerful weapon that can transform the surrounding reality and protect a person from misfortunes. Therefore, images of the hand in the ancient art of Europe and Asia (Mediterranean, Caucasus, Siberia, etc.) are not accidental. In particular, stones with handprints were discovered in the Novgorod region.

N.P. Grinkova, considering this motif in Russian folk art, came to the conclusion that images of the hand were amulets and served as good wishes among many peoples. Among Russians, the image of a hand can be found in wooden sculpture, embroidery, and ritual wedding cookies (pie) - the hand that the mother of the bride baked for the groom's family. Such cookies were like a wish for the daughter to live happily in her new family.

Geometric signs, which played an important role in the composition, carried a large semantic load. Taking into account the works where, using large archaeological material from ancient times, the ancient meaning of rhombic figures as symbols of fertility has been traced, it can be assumed that in Russian embroidery, in a number of cases, the rhombus had the same meaning. “Judging by its place in various compositions, it could mean earth, plant and woman at the same time.”

The cosmic meaning of cruciform figures, a circle, a rosette as pagan symbols is revealed in many examples of the art of the peoples of Eurasia and other continents. Circular figures on Slavic things of the X-XIII centuries. They also give reason to consider them as symbols of fire, the solar deity, as pagan cleansing and protective signs. The main anthropomorphic figures in the embroidery compositions, as well as birds and animals, marked with circles and rosettes, seem to show their involvement in the sky.

Perhaps it is no coincidence that in Russian embroidery the placement of a rhombus and a cross, a rhombus and a rosette next to each other, which in the distant past was associated with the symbolism of the union of the feminine and masculine principles.

A feature of archaic type embroideries is not only their saturation with solar, cosmic symbols, but also plant motifs that seem to permeate them. As already stated, distinctive features The iconography of archaic subjects is stinginess, laconicism in the depiction of people and animals, the severity and solemnity of frozen poses, and the geometric nature of their interpretation. The embroidery features idol-like figures. There is a lot of evidence from medieval sources about wooden idols in Rus'; sculptural images of idols were also found, particularly in the North. A tall columnar figure surrounded by smaller ones in the embroidery in Fig. 52 has similarities with the description of Slavic deities of the 10th century. The idol found in the river also had a columnar appearance. Zbruche.

The features that are consistently repeated in various embroidery versions allow us to consider the patterns in question not just a play of the embroiderers’ imagination, but deeply traditional. They did not reflect everyday stories (although over time some of them acquired a genre character), but were fundamentally associated with ancient mythological ideas.

The antiquity of the three-part composition is confirmed by finds of things from the 10th-13th centuries. on the Slavic-Chipus borderland, in which the composition is presented in a more simplified form than in embroidery (horses on the sides of the rhombus, buildings, deer on the sides of the tree, etc.). A woman holding birds in her raised hands is depicted in a miniature from a 16th-century Ryazan manuscript. This main ornamental motif of embroidery can be traced in urban art of the 16th century. The stylistic difference between the drawing of the 16th century. and embroidery from the 18th to early 20th centuries. great. The embroidery motifs look typologically more ancient.

Embroidery of the archaic type is rightly recognized as an outstanding phenomenon of Russian folk art. This is one of the important sources for revealing the features of the ancient worldview of their creators. Researchers, following V.A. Gorodtsov, see in the subjects of archaic embroidery a reflection of the ancient agricultural cult of the Slavs. “The agricultural communal religion was, apparently, the dominant form of belief and cult among the Slavic tribes before Christianization...” It highlighted the personified forces of nature that determined the well-being of the farmer. After baptism, the cult of the great gods (“higher mythology”) turned out to be the least stable, but did not disappear completely; in particular, echoes of it were preserved in the ornament of embroideries.

The Church fought against pagan cults, and they merged in everyday life with Orthodoxy, forming a unique complex of syncretic ideas. The ornament of Russian embroidery reflected “only a particle of ancient paganism, the least dangerous for the dominant religion and the closest to the farmer.”

Russian agricultural rituals of the 18th and early 20th centuries and oral poetry provide a lot of valuable information for illuminating the semantics of ornament. Holidays associated with the agricultural calendar coincided mainly with the days of solstices and equinoxes: in winter - Christmastide, Maslenitsa, in summer - Ivan Kupala Day. There are especially many holidays in the spring - during the spring awakening of nature: Semik, Trinity, funeral (or driving) of a mermaid, Kostroma, Yarila, cuckoo, etc. The entire annual cycle of agricultural holidays, with all their differences, contained many similar and even identical moments. Spring-summer games, the most developed, were mainly women's holidays, although men also participated in them (often dressed in women's clothing). The main image of the rituals - Trinity, Mermaid Week, Maslenitsa, Carols - was given anthropomorphic (most often female) features.

The central point of many rituals was dressing a tree (most often a birch) in a woman’s dress or making a stuffed doll from plant materials (reeds, straw), burying it: throwing it into water, into a rye field or burning it. Researchers rightly trace the ancient basis of ritual funerals of a doll (sometimes a living person), a bird, or a horse to the idea of ​​a dying and resurrecting deity of vegetation. It is no coincidence that the female Maslenitsa doll in the Kaluga region was represented as “young”, with emphasized female forms, symbolizing its fruitful beginning. The actions and songs that accompanied the funeral of the mermaid, Kostroma, emphasized their agrarian-magical significance. They advocated the idea of ​​fertility of the earth and man. The erotic element in one form or another was usually also present in them.

To reveal the semantics of archaic ornamental subjects, the details of rituals are also essential. It was customary to decorate the effigy and participants with plant branches. The role of wood is great (especially in spring-summer rituals), as well as water. The action usually took place at a source; branches, wreaths, and often the scarecrow itself were thrown into the water. The connection with the sun and fire is expressed in the “salting” movement, burning a wheel, lighting a fire, etc. The life-giving and cleansing power of water, sun and plants appears very clearly.

The horse played an important role in the rituals. According to folklore materials, female anthropomorphic images are associated with horses. In a song from Muromsky district. Vladimir province. There is an appeal to Kostroma: “Your black horses spend the night in the field.” In a song from the Pskov lips. it says: “...our annual Maslenitsa...”; “to drive around on komons...”; “so that the horses are black, so that the servants are young.”

A very archaic form of the ritual of driving a mermaid, in which a horse and rider personify a mermaid; on the horse (he was represented by two people) sat a leader - a man dressed in a woman’s dress, with an imitation of female forms, wearing a mask. Sometimes the horse was made of wood, straw, and a horse skull was used for the head, decorating it with branches and ribbons. In one version of this ritual, a boy was mounted on a constructed horse, the procession moved through the streets, and then the horse was thrown into the rye. The women, accompanying the mermaid horse, “played and danced.” The ceremonial leading of the horse by the bridle in a relict form resembled a ritual procession.

The horse played an important role in other rituals of the agricultural cycle. There was a game action “filly”, “drive the filly” at Maslenitsa and later. The custom of dressing up as a horse or mare at Christmas time is well known: the horse was the first mask among other animalistic images of Christmas games. He is also one of the main characters in round dance games and songs of an agrarian-producing nature: “And we sowed millet”, “The Lenok is born”, etc.

The similarity of the rituals and their participants with the subjects and characters of embroidery is extremely great. Both reflect the pagan magical and incantatory symbolism of fertility - honoring the female fruit-bearing principle, the sun, water, vegetation.

Despite all the great similarities between the images of productive rites and the accompanying poetry and ornamental motifs, there are considerable differences between them. In the embroidery, in the samples that most fully preserved the ancient iconography, the subordination of all characters to the central figure, on which the arrangement of other figures depends, is clearly expressed. She patronizes them, protects them, brings goodness, gives them a bird - a symbol of warmth and light - or a plant. Russian calendar holidays reflect a more ancient stage of anthropomorphization of religious ideas, “they arose before differentiated ideas about deities developed.”

Spring-summer rituals received the greatest development in the southern Russian provinces, less so in the northern ones. There is information that in the North they sang carols, burned Maslenitsa, said goodbye to spring, but here these rituals were preserved in a reduced form. In the extreme northern regions, the population does not know the custom of mermaid funerals. The coastal mermaid in the North is associated with the water element. Her image - in the form of a woman with a fish tail - is different from the southern Russian and Ukrainian ideas about the mermaid as the spirit of vegetation, fields and harvest.

The celebration of Ivan Kupala Day is comparatively clearer in the north-west. In the village of Ostrechye, Kargopol district. For this day, they made a stuffed animal from straw, dressed it in a woman’s shirt, and decorated it with flowers and wreaths. The most beautiful girl in the village jumped with him over a fire lit from a “live” fire (produced by friction). The rest of the girls and boys jumped after her. On this day, fishermen “conspired” with the water, hunters with the forest, and girls with the guys. Collecting herbs, searching for treasures, etc. were timed to coincide with this day. At dawn the girls were swimming.

Agrarian rituals were very developed in the 19th century. in the southern Russian and central Russian provinces and only partially in the northern Russian ones, and the distribution of archaic facial embroidery covered mainly the northwestern part of European Russia. Thus, these two areas coincide only somewhere in certain areas. The preservation of developed agrarian rituals was determined by many reasons. Probably, a significant role in this was played by the greater development of agriculture in the southern provinces compared to the northern ones, as well as the long-term preservation of serfdom remnants in the south.

The fishing and trading northern regions (whose population knew almost no serfdom) were economically developed regions of the Russian state until the 17th century. inclusive and only later found themselves apart from the main centers of intense social life. Calendar rituals in the North had a slightly different nature - agrarian-pastoral, and partly commercial in nature and were accompanied by animal sacrifices (bull, ram), public cooking of meat and meals. These actions, timed to coincide with the dates of the Orthodox calendar, were performed at a source, in a grove or near a tree, revered as sacred.

Women's fraternities were organized on "Woman's Day", or "Women's holiday", which was celebrated by the Izhorians of the St. Petersburg province and the Russians of the Pskov province. and other places. The image of a “woman,” as the pattern on towels was called in the Pskov region, was apparently associated with this holiday, which was celebrated in the spring. Its essence, according to the researcher, was to ensure a fertile year. The brotherly feast at the holiday was closed (the shepherd was the only man allowed to the feast) and had an orgy character.

The religious ideas of the peasants of the North took shape in the process of adaptation of the Chud population by the Slavs and absorbed some local mythological ideas. The cult of trees, stones and springs was known to one degree or another throughout Russia, but it lasted especially long in the North, which, apparently, is explained by the influence of adapted Chud groups. It is no coincidence that in the church message (the first half of the 16th century) the need to eradicate paganism “in the Chud and Izhora places” was especially emphasized.

Only the population of the North retained the name of the female East Slavic deity - Mokosha, who is mentioned in the chronicle. Mokosh was considered the patroness of water, economy, family hearth and women's work. Mokusha, as they said in the Olonets region, allegedly went around houses during Lent, bothering the spinning women; as a sacrifice when shearing sheep, a piece of wool was left for her, etc. In the name of Mokosh in Cherepovets district. was called a spirit in the form of a woman (with a large head and long arms), spinning at night. Functions of Mokoshi in the 19th century. very limited compared to previous centuries, and she turned from a patroness into an evil creature.

In the villages of Pinega, the idea of ​​a benevolent being - the domozhirikh - was preserved - the patroness of the hearth. Living under the floor, she seemed to protect the house, livestock and patronized women's work, especially spinning and weaving. According to the stories of the residents, the house-wife predicted the death of one of the family members with a plaintive cry, “and as soon as the house gets a profit, the house-wife is busy, smoothing the cattle and sitting at the door.”

An interesting sculptural image of the patroness of spinning - “women with sour lips”, or “Aunt Anya”, from the village of Gorki (formerly Kirillovsky district). This sculpture, dressed in peasant dress (sundress, scarf, etc.), was placed in the front corner of the hut, where girls gathered for Christmas gatherings for spinning. People approached her asking for help in their work. Paraskeva Pyatnitsa was considered the patroness of spinning and weaving (in Orthodoxy).

The earth, in the minds of Russian peasants, had some anthropomorphic features. The very expression “nurse-mother of the damp earth,” referring to her as a spiritualized being, speaks of this. The peasants celebrated the name day of the land, when it was forbidden to disturb it (plow, dig, etc.). Anyone who violated this prohibition brought trouble upon himself. Other customs also testify to the reverence of the earth and its help in illnesses. The term "cheese" indicates its inextricable connection with water. It is also connected with vegetation: grasses, flowers, bushes and trees were called the “hair” of the earth.

The pagan symbolism of ornamental compositions depicting ancient rusalia (XI-XII centuries) and saturated with roots, seeds and plants in combination with the ideogram of water, “one might say graphically expressed the idea that is embedded in the word “mother of the damp earth.” A reflection of this image can be found in embroidery: a female figure with birds in raised hands, combined with a tree growing inside her, and with water - the lower tier of embroidery (on the woman's dress) includes an image of water in the form of a jagged line. City-chat, jagged or smoother wavy line in embroidery it was often a graphic symbol of water. Bird boats float along it. Sometimes this line runs under the line representing the earth, the soil on which bushes, grass grow, birds and animals live.

The tradition of depicting water as a jagged or wavy line can be traced in the art of the North since the Neolithic era. Water is also shown in everyday embroidery subjects, where it is often combined with images of fish, ships, and boats. The vertical wavy line symbolized rain.

The central female character of the northern embroidery ornament, no matter what it was called (mother earth, great mother of nature, bereginya, woman in labor, Mokosh), undoubtedly depicted a deity, “who embodied the idea of ​​the earth giving birth and the woman continuing the family line.” .

The motif of the great goddess with a male companion is widely known in the mythology and iconography of the ancient agricultural cultures of the Mediterranean, Asia Minor and the Caucasus. The cult of the great goddess played a significant role in the religion of the Scythian tribes of the Black Sea and Dnieper regions. Her image was not alien to other peoples of Eastern Europe.

The heir to the Slavic female pagan deity, as shown by researchers (S.V. Maksimov, N.V. Malitsky, V.I. Chicherov, etc.), was Paraskeva Pyatnitsa, revered in ancient Novgorod and the North. Northern masters embodied the image of Paraskeva and Nikola in wooden sculpture (wooden sculptures were less accepted for other representatives of the Orthodox pantheon). S.V. Maksimov saw this as a tradition of paganism - the use of “ready-made forms of the old faith.” V.I. Checherov, based on a large ethnographic material, showed the merging of the image of Paraskeva with the Mother of God in the popular imagination. They acted as intercessors of women, healers of diseases of people and livestock. At the same time, their images were inextricably linked with the veneration of earth and moisture. The cult of Paraskeva the Mother of God reveals the contours of a pagan female deity - the giver of earthly blessings. It is no coincidence that his image is so persistent in the ornament of shirts, aprons, and headdresses of a betrothed girl, a bride, or a young woman in the first years of marriage. The symbolism of fertility of some motifs, apparently, remained in the popular imagination for a long time, in some places until the beginning of the 19th century.

The peasant woman, whose life was limited to family and economic interests, was the keeper of ancient ideas and images reflected in the field of women's art - embroidery.

Images of female characters in embroidery vary in their iconography. The stability of the anthropomorphic image with horns is observed, which is unlikely to be the result of free variation of the ornament by embroiderers. The headdresses of Russians (and Eastern Slavs in general) reflected certain religious ideas. The similarity between Russian magpies and kokoshniks with a bird and kichek with the horns of a cow has been repeatedly pointed out in ethnographic literature. Young married women wore a horned kitty, exchanging it in old age for a hornless one. From this we can conclude that horns were associated with the productive period in the lives of women. According to popular beliefs, they contributed to the fertility and well-being of the family. A bare-haired woman caused the wrath of the brownie (the patron of the house, family), the death of livestock, the failure of the grain harvest and the illness of people.

Archaeological and ethnographic parallels show that a female character with horns in Russian embroidery is not a unique phenomenon in art. Horned fertility deities in female (sometimes male) form are not uncommon in the ancient fine arts of the East, Egypt, Asia Minor, Greek, Caucasus and are associated with a developed agrarian cult, the cult of the bull and cow.

Worship of the wild bull - tur (and turitsa) in Ancient Rus', and then the cult role of the domestic bull among the Russians (in particular, in the northern fraternities) were expressed quite clearly. Traces of the bull cult were discovered in Novgorod in the 10th century. The “cattle” god Belee was especially revered among the Novgorodians and in the Rostov land. Later, his role as the patron of livestock passed to Blasius, who was depicted on icons with cattle. The ancient Mokosh was also endowed with the function of the patroness of livestock. The meaning of the cow as a symbol of fertility is revealed in Belarusian folklore material. The baking of “horned cow” at a Belarusian wedding is noteworthy.

Anthropomorphic creatures in frog pose also appear to have been associated with fertility magic. This significance of such figures is revealed by ethnographic data (among the peoples of Australia and America). In the rock paintings of Karelia and Siberia, these motifs are expressed quite realistically, representing the pose of women giving birth. It is possible that the images in Russian embroidery of the 18th-19th centuries, stylistically far from rock paintings, are still echoes of ancient ideas, the essence of which was forgotten, but perhaps their meaning of good wishes still remained. It is noteworthy that these motifs are marked on women's clothing, sheets and towels (which played a large role in wedding and maternity rituals).

Women's images in embroidery, whose hands seem to be turned into combs for combing flax (or wool), presented most often in the form of a frieze, can be comparable (as L.A. Dintses also believed) with the image of the patroness of spinning and weaving, o which ideas were preserved in the northern regions in the 19th and even at the beginning of the 20th century.

Agrarian rituals are also reflected in embroidery. A row of female figures (not marked with special signs) holding branches in their hands, presented ornamentally, possibly depicts a spring festival. This topic undoubtedly interested embroiderers. For example, embroidery on a towel from Vesyegonsky district. second half to end of the 19th century. quite realistically depicts the Trinity ritual: a birch tree and girls walking towards it with branches in their hands. The difference in the patterns of the ancient version and the later one is great. The first (embroidered with a double-sided seam) is made in a conventional, generalized manner, the second (embroidered with tambour and satin stitch) is realistic (there are even such details as wreaths in the form of ovals and circles).

Apparently, also associated with ritualism is a widespread motif of a number of female figures with branches or lamps in their hands - godmothers, godmothers with candles. The custom of cumulation was timed to coincide with the seventh or trinity; Some analogies to the images of kumka with candles are found in the description of the mermaid’s funeral ritual, when its participants, female songwriters, accompanied the mermaid with homemade “candles” made from reeds or wormwood branches.

The question of the origin of the figures of horsemen (instead of horsemen) on the sides of the central female figure or tree needs further research. As one of possible solutions In this regard, I will point out the ancient custom of gender reassignment - dressing men in women's clothing when performing priestly functions. It existed in a vestigial form back in the 19th century. among the peoples of the North and Volga region. T.A. Kryukova, who noted this phenomenon in the southern Russian ritual of driving a mermaid, wrote: “The leader - a man in a woman’s robe - in a mermaid outfit is a very archaic character.”

Male characters in facial embroideries often occupy the main place in the pattern. The horseman with all the signs marking his involvement with the deities is also associated with the agrarian-pastoral cult. Addressed to some higher being, he is, as it were, a protector of people and animals, depicted in small figures in the pattern.

The archaic image of a rider on horses, fused bodies, with his hands in an emphatically blessing gesture, with a radiant halo around his head, is perhaps an echo of the ancient solar image.

“On foot” rows of male figures (men in hats) are reminiscent of the words of Kalevala about the “forest people”, the owners of the forest. Ideas about the owners of forests and water were widespread among Russians in the North - in the region of forests and lakes - and were preserved among the population until the beginning of the 20th century. Like the neighboring Karelians, these owners seemed to have many families.

It can be assumed that the rows of standing male figures and single “sprouted” male figures in embroidery are associated with images of lower mythology, which were included in embroidery throughout its centuries-old life. In the light of these assumptions, the name “leshaki”, which was used to name anthropomorphic figures in Onega embroidery, becomes more understandable. Obviously, their original meaning was forgotten and then reinterpreted.

As for the patterns, where a number of male and a number of female figures are presented, they are reminiscent of round dance games, when girls and boys were divided into two parties, performing alternately, as, for example, in the round dance song “And we sowed millet,” which was once agrarian -producing character.

The term "panya" for female figures in embroidery could be synonymous with "lady", but another explanation is possible. It is compared with the word “punks” - that’s what they were called in the Arkhangelsk province. dolls, roughly carved from wood, schematically depicting a person (often with identified female characteristics), as well as with the word “lords”, known in the North to mean ancestors who were commemorated on certain days. If we remember the important function of the towel - its role in funeral and memorial services.