Texture classification using local and global histogram equalization and the Lempel-Ziv-Welch algorithm

This paper presents a new texture classification method using histogram equalization and the Lempel-Ziv-Welch algorithm. Two versions of the classifier are presented. The first version uses global histogram equalization and the second one uses local histogram equalization. After local or global histogram equalization, each texture sample is encoded by LZW using pre-constructed dictionaries and assigned to the class whose dictionaries minimize the coding rate. Both versions are tested under different conditions of gray-level variations.