Unsupervised Color-Texture Image Segmentation Based on A New Clustering Method

Image segmentation is a classical problem in the area of image processing, motion estimation, and soon. Although there exist a lot of clustering based approaches to perform image segmentation, few of them study how to obtain more accurate image segmentation results by designing a suitable clustering method. In this paper, we select an appropriate distance measure in the composite feature space of color and texture. Then the distance measure is incorporated in a clustering method that utilizes the spatial information of each feature vector. Finally, the proposed scheme performs morphology filtering to obtain the final segmented regions. Experimental results show that proposed scheme can constantly achieve higher segmentation accuracy compared to some state-of-art image segmentation algorithms.