Finite point sets in recognizing location and orientation of machine parts of complex shapes

Abstract

In the article, a method of recognizing the position, location and orientation of irregular machine parts with a complex outline of the external contour is suggested. Recognition is performed on the basis of a raster image that has undergone preliminary processing in a vision system. The developed method is based on matching the finite set of points retrieved from the external contour of the classified object. The originality of the approach consists of the form of the object pattern taken into account in determining the degree of similarity to reference objects. The similarity index was defined in the category of fuzzy sets. A cellular automaton has been proposed to generate the outline of the perimeter quickly while simultaneously smoothing it out. The process chain necessary for the recognition function to be implemented has been presented. The developed method has been illustrated with an example related to recognizing the position, location and orientation of a rear wheel pin. The results were compared to the classic Blair-Bliss, Danielsson and Haralick shape coefficients. The sensitivity of the developed method to a change in the scale and rotating the object as well as errors in outlining its edge has been tested. The obtained results confirmed the advantageous features of the developed method, both in the aspect of the recognition quality and the practically viable time of waiting for the results of processing.