Content Based Image Retrieval using Color and Full Texton Index Co-occurrence Matrix (FTiCM) Features
G. BinduMadhavi1, V. Vijaya Kumar2, K. Sasidhar3
1G. BinduMadhavi, Research Scholar of JNTUH, Department of Computer Science and Engineering, Asst. Professor. Anurag Group of Institutions (Autonomous), Hyderabad, India.
2V. Vijaya Kumar, Professor, Dean – Department of Computer Science and Engineering, Anurag Group of Institutions (Autonomous), Hyderabad, Telangana, India.
3K. Sasidhar, Professor, Head of Enterprise Content Management, Srinidhi Institute of Science and Technology, Hyderabad.

Manuscript received on 15 April 2019 | Revised Manuscript received on 19 May 2019 | Manuscript published on 30 May 2019 | PP: 1262-1275 | Volume-8 Issue-1, May 2019 | Retrieval Number: A3211058119/19©BEIESP
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (

Abstract: In this paper a powerful feature descriptor is derived based on color and texture features. This paper initially transforms the RGB color image into HSV color apace. The color histograms are derived on H, S and V color planes. The structural features are derived on V-color space by dividing into micro girds of size 2×2 and each micro grid is replaced with full texton index (FTi). The FTi’s are extracted based on shape of the texton derived from four, three and two identical pixels. The proposed full textons represents all possible patterns on a 2×2 grids and this representation over comes the disadvantages and ambiguity of TCM and MTH approach. The texture features are derived on the FTi image by computing co-occurrence matrix and by deriving gray level co-occurrence matrix (GLCM) features. The GLCM features derived from Full Texton Index co-occurrence Matrix (FTiCM) are integrated with the histogram features derived on the individual color planes. The proposed color histogram-full texton co-occurrence matrix (CHFTiCM) model is experimented on Corel-1k, Corel-10k, MIT-VisTex, Holidays and CMU-PIE and Holidays datasets. The performance of the proposed CBIR is measured in terms of average retrieval rate (ARR) and average precision rate (APR) and compared with the state-of-art methods. The results indicate the superiority of the proposed method over the existing methods.
Index Terms: RGB, HSV, V-Color Space, Texton, MTH, TCM.

Scope of the Article: Image Processing and Pattern Recognition