Efficient Item Image Retrieval System
S. Adebayo Daramola1, Ademola Abdulkareem2, K. Joshua Adinfona3
1Dr. S. Adebayo Daramola, Department of Electrical and Information Engineering, Covenant university, College of Science and Technology, Ota, Nigeria
2Engr. Ademola Abdulkareem, Department of Electrical and Information Engineering, Covenant university, College of Science and Technology, Ota, Nigeria.
3F.Joshua Adinfona, Department of Electrical and Information Engineering, Covenant university, College of Science and Technology, Ota, Nigeria.
Manuscript received on May 03, 2014. | Revised Manuscript received on May 05, 2014. | Manuscript published on May 05, 2014. | PP: 109-113 | Volume-4 Issue-2, May 2014. | Retrieval Number: B2225054214/2014©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: Content based image retrieval system is a very effective means of searching and retrieving similar images from large database. This method is faster and easy to implement compare to text based image retrieval method. Ability to extract discriminative low level feature from these images and use them with appropriate classifier is factor in determining retrieval result. In this work efficient item image retrieval system is proposed. The system utilizes Haar wavelet transform, Phase Congruency and Support Vector Machine. Haar wavelet transform acted on image to form four sub-images. Texture feature is extracted from smaller image blocks from detailed bands and it was combined with shape feature from approximation band to form feature vector. Feature distance margin is achieved between query image and images in the database using Support Vector Machine (SVM). The effectiveness of the system is confirmed from output retrieval results.
Keywords: Content, Texture, shape, Support Vector Machine, Phase Congruency.