A New Frame Work for CBIR Using Odd and Even Tetra Texton Matrix
This paper derived a new frame work for texton extraction on a 2X2 grid. The proposed Extended Complete Texton Index (ECTI) frame work derives all possible texton structures with 2, 3, and 4 identical pixels without any ambiguities. This research initially replaces the 2X2 grid with ECTI ranging from 0 to 14, Later this paper divides the ECTI image into a 3X3 windows. The 3X3 ECTI windows are divided into Tetra Texton Odd Unit (TTOU) and Tetra Texton Even Unit (TTEU). Based on the relative frequencies of TTOU and TTEU on each 3X3 ECTI window this paper derived Odd & Even Tetra Texton Matrix (OE-TTM). The GLCM features are derived on OE-TTM act as the feature vector. The proposed OE-TTM frame work integrates structural, texture and statistical features. To test the efficacy of the proposed method, it is tested on five bench mark databases consisting of wide verity and categories of natural textures, human faces and sceneries. The proposed frame work is tested on state-of-art existing CBIR methods and the results are shown the efficiency of the proposed method.