Method for Image Informational Properties Exploitation in Pattern Recognition Environment
The problem of finding optimal image transformation according to informational image properties is considered. During the research, image informational properties that distinguish images from all other recognition objects were investigated. The notion of image equivalence was introduced. Notion of image equivalence as identity of images with respect to some transformation set was explored. The proposed equivalence definition allows decomposition of image set into subsets of a certain type and establishes correspondence between image equivalence classes and some subsets of operations. On this basis, the method for selecting efficient image recognition algorithm according to image informational nature was elaborated. The proposed method provides a possibility of improving efficiency of selecting image analysis algorithms and automation (partial or complete) of image processing. It allows taking into consideration internal information that image conveys, syntactic and semantic.
- 1.I.B. Gurevitch. Descriptive Technique for Image Description, Representation and Recognition // Pattern Recognition and Image Analysis: Advances in Mathematical Theory and Applications in the USSR. — 1991.-Vol.1, No. 1.-P. 50–53.Google Scholar
- 2.I.B. Gurevich, Yu.G. Smetanin, Yu.I. Zhuravlev, Descriptive Image Algebras: Determination of the Base structures, Pattern Recognition and Image Analysis: Advances in Mathematical Theory and Applications 9(4), 1999, 635–647.Google Scholar
- 3.I.B. Gurevich, Yu.I. Zhuravlev. An Image Algebra Accepting Image Models and Image Transforms // Proceedings of the 7th International Workshop “Vision, Modeling, and Visualization 2002” (VMV2002), November 20–22, 2002, Erlangen, Germany [G. Greiner, H. Niemann, T. Ertl, B. Girod, H.-P. Seidel (Eds.)], IOS Press, B.V.Amsterdam, Infix, Akademische Verlagsgasellschaft, Aka GMBH, Berlin, 2002. — P. 21–26.Google Scholar
- 4.I.V. Koryabkina. Informational Specificity of an Image in Pattern Recognition Environment // Proceedings of the IASTED International Conference in cooperation with The Russian Academy of Sciences: Siberian Branch “Automation, Control and Information Technology”, June 10–13, 2002, Novosibirsk, Russian Federation.-P.435–438.Google Scholar