Texture Based Image Retrieval Using GLCM and Image Sub-Block

This study proposed an approach to retrieve images based on texture features using GLCM and image subblocks. Each image is divided into three rows and three columns with equal sizes. Texture features are extracted based on GLCM (Gray Level Co-occurrence Matrix) using four statistical features that is contrast, homogeneity, energy and correlation. The features are calculated in four directions (0 0 , 45 0 , 90 0 , and 135 0 ). A total of 16 texture values are calculated per image sub-blocks. In this retrieval system Euclidean distance and City block distance are used to measure similarity of images. This retrieval system performance is measured in terms of its recall and precision. The performance of retrieval system is also measured based on AVRR (Average Rank of Relevant Images) and IAVRR (Ideal Average Rank of Relevant Images) that is proposed by Faloutsos. The retrieval results show that the performance using City Block distance has achieved higher than the performance using Euclidean distance. Keywords— AVRR, City Block distance, Content Based Image Retrieval (CBIR), Euclidean distance, GLCM, IAVRR,