A Statistical Approach To Image Segmentation On A Cylindrical Surface

This paper discusses the development of a machine vision system using Bayesian Decision Theory in classifying target and non-target areas: on a cylindrical surface. Algorithms developed using Bayesian theory are used to extract a cylindrical surface form a textured background and then to segment the image into target and non-target areas for subsequent processing. The advantages of the algorithms are that no apriori grey level threshold values are required, illumination constraints are reduced, and the results are consistent.