Texture image segmentation by optimal Gabor filters

In this paper, we present a new texture image segmentation algorithm using multi-channel Gabor filters and fuzzy c-means clustering. We propose a design procedure of Gabor filter parameters for optimal discrimination of texture boundaries and a systematic iterative filter selection scheme to obtain the texture features of an input image. Based on the extracted texture features and the number of texture categories, the fuzzy c-means clustering algorithm is then used to identify the texture category. The experimental results indicate, that our proposed algorithm is much superior to the existing algorithm based on the log-octave filter bank paradigm.