Effective selection of mixed color features for image segmentation
Junfeng Luo, Jinwen Ma · 2016
Image segmentation is a basic task in image analysis and understanding and feature extraction is important but difficult. In this paper, we propose an effective feature selection method for color image segmentation which selects a group of mixed color features or channels from some different color spaces according to the principle of the least entropy of pixels frequency histogram distribution. Actually, 3 color channels that have the least entropies among all the channels in the whole alternative color spaces for a color image are selected to be our extracted mixed features. Publicly available segmentation database BSDS500 and 3 different segmentation algorithms are adopted to demonstrate the advantage and improvement of our selected mixed features on color image segmentation.