Classification model for flat nonconvex images using diagonal segments and tuples for system of automatic recognition of three-dimensional objects

Abstract

During the past five years, many researchers have developed different approaches to the classification of images, which are used for a variety of scientific tasks [1, 2, 3 and 4]. The research is aimed at solving task of classification of three-dimensional objects by the images of their projections in systems of automatic recognition of randomly located parts and products in the industrial belt. This article describes the classification model of flat nonconvex images by their form. Author offers twelve classes. The criteria for classification in this model is combination of diagonal segments in the four quadrants of the bounding rectangle of image projection of the object. Illustrations of each class, classification scheme as well as the research results of developed model on images of projections of real three-dimensional objects are provided.