Multi Object and Dynamic Query Based CBIR System using DCT Incorporated with HOG and HTF
G. Suresh1, N.C. Sendhil Kumar2, R. Murugesan3, P. Mukunthan4
1G. Suresh, Professor, Sri Indu College of Engineering and Technology, Hyderabad (Telangana), India.
2N.C. Sendhil Kumar, Professor, Sri Indu College of Engineering and Technology, Hyderabad (Telangana), India.
3R. Murugesan, Associate Professor, Annamacharya Institute of Technology & Sciences, Tirupathi (Andhra Pradesh), India.
4P. Mukunthan, Professor, Sri Indu College of Engineering and Technology, Hyderabad (Telangana), India.
Manuscript received on 27 November 2019 | Revised Manuscript received on 15 December 2019 | Manuscript Published on 30 December 2019 | PP: 590-592 | Volume-9 Issue-2S3 December 2019 | Retrieval Number: B11341292S319/2019©BEIESP | DOI: 10.35940/ijitee.B1134.1292S319
Open Access | Editorial and Publishing 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 (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: This work contributes multi object detection and dynamic query image based retrieval system. Generally, finding relevance and matching user expectations is very critical based on query key information and these results irrelevant responses which will produce low similarity index. Consequently, CBIR system took a major responsibility of identifying new objects, retrieving similar objects or contents based on multi query and dynamic keywords with improved recall and precision as per requirement of the users. At this juncture, Discrete Curvelet Transform with the incorporation of HOG and HTF based approach is proposed to handle commercial image, medical images and types of multi model images. This proposed approach mainly focuses on extracting scaled features for finding correlation among the query and database images. To start with the process, query image is decomposed into multi level sub images to extract set of texture features at two levels. These features are estimated by Gray Level Co-occurrence Matrix (GLCM) and HOG descriptor based techniques is adapted to find scaled vectors with reduced dimensionality. This method outperform compared as compared to existing method is authenticated from experimental results.
Keywords: Image Retrieval, HOG, Curvelet Transforms, GLCM.
Scope of the Article: Multi-Agent Systems