An Approach for Image Thresholding Using CNN Associated with Histogram Analysis

Thresholding is one of the old, simple, and popular techniques for image segmentation, and has been widely studied. In this paper, an approach for image thresholding based on cellular neural network (CNN) associated with histogram analysis is presented. The approach realized by threshold CNN (T-CNN), in which the threshold is obtained via histogram-based automatic searching algorithm. Experimental results on real images show that the proposed approach can extract the objects from the background effectively with better visual quality than other methods.