An image segmentation algorithm based on similarity matrix
Some popular existing clustering algorithms are hard to obtain good image segmentation results. Aim to this problem, similarity matrix algorithm is designed and implemented to solve the problem in this paper. According to the capacity of each pixel being the center, member pixels and centers can be identified to complete clustering process. Then, the color difference among each pixel is defined by the Luv color model to provide a meaningful measure of distance which is considered as the input of similarity matrix algorithm. The experimental results show that the image segmentation algorithm based on similarity matrix has superior performance than K-means and Ncut algorithm.