Abstract: Two groups of Cellular Neural Networks（CNN） have been used to segment Industrial Computerized Tomography（ICT） Images.The first group，namely rough image segmentation，has been adopted to obtain thresholding images，then with the adaptation of the second group some more sophisticated segmentation information will be procured，with edges included.This article deals with the Pseudo-steady-state of net，which has been served as the termination condition of the iteration.Experimental results of engineering image series demonstrate the efficiency of the methods presented in the article.