Unsupervised texture segmentation using color quantization and color feature distributions

An unsupervised texture segmentation method is presented in this paper. In this paper, we propose a simple color quantization scheme to reduce the color space, and then local binary pattern (LBP) and color histogram (CH) are applied to measure the similarity of adjacent texture regions during the segmentation process. In addition, for improving the segmentation accuracy, an efficient boundary checking algorithm is proposed. The execution time of pixelwise modification is also reduced by the proposed approach. The proposed method achieves not only saving processing time but also segmenting the distinct texture regions correctly.