The Multi-valued Astronomical Image Segmentation Based on Pulse Coupled Neural Networks
Qing Liu, Yang Xiaoping, Xiaoshu Ma · 2013
In order to solve the problem of multi-valued segmentation in the astronomical image, the astronomical image is iteratively processed by Pulse Coupled Neural Networks. In the processing, the maximum mutual information is taken as the optimization segmentation, and the difference of mutual information is regarded as the classified criterion from the relationship between original image and image segmentation. The experimental results show that this algorithm can effectively determine the number of class of the astronomical image, and can be more adaptability and can maintain the edges, details of images.