Abstract: Image's feature extraction and matching is essential for content based image retrieval. Io retrieve texture image better, this paper discussed the Rational method for extract features of texture image by gray level co-occurrence matrix. Then combined with similarity measures, the basic system based on LLCM was developed. Different measure funcdons and different combination of statistics were used to analyze their effect on retrieval. I}he result indicates that better retrieval performance is achieved by extracting four parameters from GLOM and measuring them with weighted block distance.

Key words: Image retrieval, Feature extraction, Dray level co-occurrence matrix