Image quantization using Self-Splitting Competitive Learning
Yajun Zhang, Zhi-Qiang Liu · 2002
We have developed a new, robust clustering algorithm, Self-Splitting Competitive Learning (SSCL). It has shown great abilities in detecting not only isolated clusters, but overlapped clusters, curves and spherical shells. We apply SSCL to quantization of color images. The clustering algorithm iteratively partitions the color space into natural clusters without a prior information on the number of clusters. The algorithm starts with only a single color prototype and adaptively splits it into multiple prototypes during the learning process based on a split validity measure. It is able to discover all natural groups; each is associated with a color prototype. The experimental results show remarkably better performance as compared to several other existing clustering algorithms.