Research Journal of Applied Sciences

Year: 2015
Volume: 10
Issue: 8
Page No. 371 - 375

Concerning the Issue of Object Classification by Form

Authors : Vyacheslav B. Fofanov and Alexey N. Zhiznevsky

Abstract: During the solution of applied problems related to the detection and classification of objects creating a real scene as the source of information the images are used often. On the images the scene objects are represented by their projections on a plane perpendicular to the observation trend. Therefore, during the classification of objects an important telltale sign is the form of their projections which are the subsets (figures) on a plane. In contrast to the visual classification when the comparison of plane figures by form may be performed almost at a subconscious level, an automatic classification requires a formal definition of this term. The development of an object projection according to a real scene image called segmentation is quite a challenging issue. The projection obtained from the segmentation usually differs from previously prepared projection with which the comparison occurs. This means that the method of projection comparison by form must be resistant to the errors arising during segmentation. The research consists of two parts. In this first part, the formalization of a plane figure form as the probabilities of chord length distributions, cut out by the figure from a random line. The independence of the form on shifts and turns is proved. It is shown that the comparison of figures by form is reduced to the test of two sample homogeneity hypothesis. In the second part of the proposed definition of the form is used for the classification of vehicles. The initial information for solving this problem are the images taken from aircrafts.

How to cite this article:

Vyacheslav B. Fofanov and Alexey N. Zhiznevsky, 2015. Concerning the Issue of Object Classification by Form. Research Journal of Applied Sciences, 10: 371-375.

Design and power by Medwell Web Development Team. © Medwell Publishing 2022 All Rights Reserved