Hybrid intelligent algorithms for color image segmentation
Xuexi Zhang, Yimin Yang · 2008
The single arithmetic of color image segmentation inevitably has some deficiencies and defects, and we can combine different algorithms according to the actual situation or have a hierarchical division of image segmentation. This paper suggests a hybrid intelligent color image segmentation method. Region growing is used to finish initial segmentation, the final segmentation images is realized by MST method which looks every region produced by region growing as a node, and an particle swarm optimization is used to get the best thresholding of MST. Region growing focuses on local variations of an image with fast speed while MST can extract the global property of an image, and particle swarm optimization can improve the algorithm speed. The method presented in this paper combines their advantages. Experiment results show that the new method has good effectiveness and efficiency.