Automatic Color Clustering Based on Competitive Network
Weiming Yin, Yuxiang Shao · 2009
Traditional clustering algorithms have difficulty in the adaptive determination of the proper clustering number and the quantificational evaluation of image segmentation. To solve these problems, an improved method based on competitive network is presented in this paper. First, a criterion is put forward to determine the optimal clustering number. Then, both chromatic and monochrome features are extracted from pixels to carry out dual clustering in succession. Moreover, a quantificational indicator is provided to evaluate the segmentation quality objectively. The experiments results indicate that, this method can not only keep the skeleton of an image using just a few colors, but also is robust for complicated images.