An Effective Color Quantization Method Based on the Competitive Learning Paradigm.

摘要: Color quantization is an important operation with many applications in graphics and image processing. Most methods are essentially based on data clustering algorithms one which the popular kmeans algorithm. A common drawback of conventional generation empty clusters (dead units). In this paper, we apply Uchiyama Arbib’s competitive learning algorithm [1] to problem color quantization. contrast batch k-means algorithm, requires no cluster center initialization. addition, it effectively avoids dead unit by utilizing a simple splitting rule. Experiments commonly used test images demonstrate that presented method outperforms various stateof-the-art terms effectiveness.