Improved support vector clustering algorithm for color image segmentation

Color image segmentation has attracted more and more attention in various application fields during the past few years. Essentially speaking, color image segmentation is a process of clustering according to the color of pixels. But, traditional clustering methods do not scale well with the number of training samples, which limits the ability of handling massive data effectively. With the utilization of an improved approximate Minimum Enclosing Ball algorithm, this article develops a fast support vector clustering algorithm for computing the different clusters of given color images in kernel-introduced space to segment the color images. We prove theoretically that the proposed algorithm converges to the optimum within any given precision quickly. Compared to other popular algorithms, it has the competitive performances both on training time and accuracy. Color image segmentation experiments performed on both synthetic and real-world data sets demonstrate the validity of the proposed algorithm.