Automatic tuning of the pulse-coupled neural network using differential evolution for image segmentation

Abstract

The pulse-coupled neural network (PCNN) is based on the cortical model proposed by Eckhorn and is widely used in tasks such as image segmentation. The PCNN performance is particularly limited by adjusting its input parameters, where computational intelligence techniques have been used to solve the problem of PCNN tuning. However, most of these techniques use the entropy measure as a cost function, regardless of the relationship of inter-/intra-group dispersion of the pixels related to the objects of interest and their background. Therefore, in this paper, we propose using the differential evolution algorithm along with a cluster validity index as a cost function to quantify the segmentation quality in order to guide the search to the best PCNN parameters to get a proper segmentation of the input image.