Different shape and color targets detection using auto indexing images in computer vision system

Abstract

One of the main challenges in computer vision is to determine the number of different types, shapes, locations and color targets within the image plane for use in computer control systems. In this study, an algorithm introduced to detect the number of targets (one or two), their shapes (square or circle) and colors (red, green or blue). A new technique presented as a digital indexing code table to present the studied color targets images. The indexing table technique depends on decimal and binary numbers. In this study, there were 42 different cases represents all the input images. There is a special case considered for the similarity of input images in case it has the same shape and color, but a change in rotation and space between two image targets. This solved using referencing to indicate the same target in each case. Thus, the classification results were 100% for the presented algorithm for all input cases.