Abstract:
Shape is a prime carrier of information characterizing an object of interest. The main subject of shape description is to extract the features of shape that are efficient to estimate a region. Many shape description schemes have been proposed and used in computer vision in the past. In this paper, main attention is focused on a practical problem of typical binary computer vision, i.e., traffic sign recognition. New shape description methods are proposed and used to characterize the shapes of traffic signs, called as quantitative descriptions of geometric features of shape, which are based on the theories of set transform and binary rank statistics. In this procedure, mathematical morphological transforms and binary rank order transform are employed, and four kinds of shape matching methods are presented. Experimental results show that the proposed methods are efficient and robust.