International Journal of Computer Science and Technology

International Journal of Computer Science and Technology | Vol. 6, No. 6, June 2004 | pp. 41–48

DOI: 10.46882/2004/IJCST/000059

Article Type: Research Article

Title: A Relational Content-Based Image Retrieval System Using Multi-Feature Fusion and Relevance Feedback

Names of Authors: David Miller¹, Zhengyang Liu²

Authors’ Affiliations: ¹Department of Data Science, New York University, New York, USA; ²Department of Computer Science, National University of Singapore, Singapore

Abstract: Querying unlabelled digital image repositories using keyword search configurations often returns noisy and inaccurate results due to the semantic gap problem. This paper presents a Content-Based Image Retrieval (CBIR) engine that incorporates multi-feature fusion alongside a dynamic relevance feedback loop. The framework extracts global color histograms, Gabor filter texture vectors, and Zernike moment shape indicators to construct an uncompressed index file. We design a user-guided interactive tracking routine that updates feature weight parameters on the fly using a modified vector space model. Evaluated on an open dataset containing 10,000 photographic entries, our CBIR layout scored a mean average precision (MAP) rating of 84.6%, outperforming standalone color index searches by a margin of 21.5%. The query execution duration requires less than 42 ms per search, confirming its usability for production-grade digital asset systems.

Keywords: Content-Based Image Retrieval, Feature Fusion, Relevance Feedback, Texture Analysis, Computer Vision

Manuscript Timeline: Received: March 04, 2004; Revised: April 15, 2004; Accepted: May 12, 2004; Published: June 03, 2004

Citation: Miller, D., & Liu, Z. (2004). A Relational Content-Based Image Retrieval System Using Multi-Feature Fusion and Relevance Feedback. International Journal of Computer Science and Technology, 6(6), 41–48.

Subscribe to read the full article: https://internationalscholarsjournals.org/subscribe-to-read