Image Retrieval using Fractional Energy of Row Mean of Column Transformed Image with Six Orthogonal Image Transforms
H. B. Kekre1, Sudeep D. Thepade2, Archana A. Athawale3, Paulami Shah4
1Dr. H. B. Kekre, Sr. Professor, MPSTME, SVKM’s NMIMS Deemed-to-be University, Vileparle (W),Mumbai, India.
2Dr. Sudeep D. Thepade, Associate. Professor, MPSTME, SVKM’s NMIMS Deemed-to-be University, Vileparle (W), Mumbai, India.
3Dr. Archana Athawale, Asst. Professor, Thadomal Shahani Engg. College, Bandra (W), Mumbai, India.
4Paulami Shah, M.E. Student, Thadomal Shahani Engg. College, Bandra (W), Mumbai, India.
Manuscript received on August 19, 2011. | Revised Manuscript received on August 29, 2011. | Manuscript published on September 05, 2011. | PP: 168-173 | Volume-1 Issue-4, September 2011. | Retrieval Number: D0125081411/2011©BEIESP
Open Access | Ethics and Policies | Cite
© The Authors. Published By: Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: The thirst of better and faster retrieval techniques has always fuelled to the research in content based image retrieval (CBIR). The paper presents innovative content based image retrieval (CBIR) techniques based on feature vectors as fractional coefficients of row mean of column transformed images using Discrete Cosine, Walsh, Haar, Slant, Discrete Sine, and Hartley transforms. Here the advantage of energy compaction of transforms in low frequency coefficients in transform domain is taken to greatly reduce the feature vector size per image by taking fractional coefficients of row mean of column transformed image. The feature vectors are extracted in six different ways from the transformed image, with the first being considering all the coefficients of row mean of column transformed image and then six reduced coefficients sets (as 50%, 25%, 12.5%, 6.25%, 3.125%, 1.5625% of complete row mean of column transformed image) are considered as feature vectors. The six transforms are applied on the colour components of images to extract row mean of column transformed RGB feature sets respectively. Instead of using all coefficients of transformed images as feature vector for image retrieval, these six reduced coefficients sets for RGB planes are used, resulting into better performance and lower computations. The proposed CBIR techniques are implemented on a database having 1000 images spread across 10 categories. For each proposed CBIR technique 40 queries (4 per category) are fired on the database and net average precision and recall are computed for all feature sets per image transform. The results have shown performance improvement (higher precision and recall values) with fractional coefficients compared to complete transform of image at reduced computations resulting in faster retrieval. Finally Discrete Cosine Transform (DCT) surpasses all other discussed transforms in performance with highest precision and recall values for 50% of fractional coefficients.
Keywords: CBIR, Cosine Transform , Walsh Transform, Haar Transform, Sine Transform , Slant Transform, Hartley Transform, Fractional Coefficients, Row Mean.