Research and Application of Texture Feature Extraction Based on Multi-features

Texture is a significant visual feature which is commonly used to identify and distinguish the image .. This paper summarized and analyzed the current common method of texture feature extraction ,including Gray Level Co‐occurrence Matrix (GLCM ) ,Local Binary Pattern (LBP) and Discrete wavelet transform (DWT ) .With the weight of configuration parameters ,a new texture extraction method of multi‐features is proposed and implemented ,which com‐bines the three basic methods mentioned .The image texture description ability of different methods is compared through the tests on the image retrieval system .The results show that the average precision of texture feature extraction method based on multi‐feature combination increased 20% comparing to GLCM algorithm ;increased 9% comparing to LBP algo‐rithm ;increased 10% comparing to DWT algorithm ;and increased 15% comparing to Xu's texture feature extraction method when images were retrieved from Corel library .The new texture feature extraction method proposed in this paper combined the advantages of each method ,and had good rotation invariance and scale invariance .However ,it is necessary to extract the texture features of GLCM ,LBP and DWT at the same time ;the time required for the new texture feature extraction method is the sum of the three algorithms ,which limits its application in some practical cases .