Complementary feature extraction approach in CBIR

Since last few years, Content Base Image Retrieval (CBIR) system has got more attention from its generic to specific use. CBIR depends upon visual low-level feature extraction i-e color, texture, shape and spatial layout. In this paper, a Local Binary Patterns (LBP) has been employed for texture analysis of image and also it is compared with average RGB color image descriptor method. And then a complementary feature extraction approach using average RGB color and LBP texture method has been proposed for CBIR. Euclidean distance is used as similarity measure for finding similar images in the database. The experimental results are generated using MATLAB. The obtained results proved that the accuracy and efficiency of proposed method in terms of overall precision, recall, f_measure and retrieval time are quite higher than single color and texture feature extraction approach.