PROCEDURE DEVELOPMENT FOR IMAGE IDENTIFICATION ON PICTURE BASED ON STRUCTURAL APPROACH
Authors:
1.
Institute for design-technological Informatics of the RAS
(Professor, ved. research fellow)
employee
Moscow, Moscow, Russian Federation
employee
Moscow, Moscow, Russian Federation
Type:
Article
Pages:
from 171
to 177
Status:
Published
Received:
02.03.2017
Accepted:
02.03.2017
Published:
02.03.2017
Language:
Russian
Keywords:
structural approach, picture decoding, image structure, image signs
Abstract (English):
The paper deals with the object identification in the picture based on a structural approach. The analysis of existing methods is presented and a new procedure of image identification with the use of a structural approach is shown. In the introduction there is described a significance of picture decoding and a concept of a structure in the picture is designated. The analysis of such a structural approach to image identification in a picture as a symbolic description which is a transition from the simplest signs of a picture to a set of graphic symbols. The second structural method analyzed in the paper is a syntactic approach to the image description. Here the whole picture is divided up to the simplest elements. The elementary graphic primitives are emphasized which make a pic-ture. And further the objects of the picture are de-scribed by a set of primitives. The regulation for the description of an object as a chain of primitives sets with the aid of the socalled language grammar of image description. Merits and demerits of existing structural methods are described, one of which is a complexity of object description on the picture of a real world. The offer on the necessity of the development of a new method taking into account the signs of really existing object and also describing objects in a formal view is put forward. In the basic part a developed procedure defining objects in the picture as a specified polychromatic multitude emphasizing such properties as a significance of brightness, an object position and a number of additional characteristics is described. The definition of objects in the picture takes place by means of the creation of a structured description of known objects of the picture the identification on the basis of the use of a neuronet approach and a specification of the structure of objects found. As a result of the identification according to this procedure there is obtained a multitude of objects which was necessary to identify in the picture presented in a structural form. Such an approach can be used for various level of hierarchy at the description of objects in the picture it also allows taking into account not only brightness characteristics and a form, but a number of additional signs of objects. In the conclusion the distinguishing features of this procedure and directions of its development are shown.
The paper deals with the object identification in the picture based on a structural approach. The analysis of existing methods is presented and a new procedure of image identification with the use of a structural approach is shown. In the introduction there is described a significance of picture decoding and a concept of a structure in the picture is designated. The analysis of such a structural approach to image identification in a picture as a symbolic description which is a transition from the simplest signs of a picture to a set of graphic symbols. The second structural method analyzed in the paper is a syntactic approach to the image description. Here the whole picture is divided up to the simplest elements. The elementary graphic primitives are emphasized which make a pic-ture. And further the objects of the picture are de-scribed by a set of primitives. The regulation for the description of an object as a chain of primitives sets with the aid of the socalled language grammar of image description. Merits and demerits of existing structural methods are described, one of which is a complexity of object description on the picture of a real world. The offer on the necessity of the development of a new method taking into account the signs of really existing object and also describing objects in a formal view is put forward. In the basic part a developed procedure defining objects in the picture as a specified polychromatic multitude emphasizing such properties as a significance of brightness, an object position and a number of additional characteristics is described. The definition of objects in the picture takes place by means of the creation of a structured description of known objects of the picture the identification on the basis of the use of a neuronet approach and a specification of the structure of objects found. As a result of the identification according to this procedure there is obtained a multitude of objects which was necessary to identify in the picture presented in a structural form. Such an approach can be used for various level of hierarchy at the description of objects in the picture it also allows taking into account not only brightness characteristics and a form, but a number of additional signs of objects. In the conclusion the distinguishing features of this procedure and directions of its development are shown.
Keywords:
structural approach, picture decoding, image structure, image signs
structural approach, picture decoding, image structure, image signs
1. Gonzales, R. Digital Image Processing / R.Gonzales, R.Woods. - М.: Technosphere, 2005. - pp. 1072.
2. Pavlov, S.N. Systems of Artificial Intelligence: manual in 2 Parts. / S.N.Pavlov. - Tomsk: El Continent, 2011. - Part 2. - pp. 194.
3. Prett, W. Digital Image Processing: [transl. from Engl.] / W.Prett. - М.: Mir, 1982. - B. 2. - pp. 480.
4. Fu, К. Structural Methods in Image Identification / К. Fu. - М.: Mir, 1977. - pp. 320.
