A Learning Approach to Content-based Image Retrieval Combining Radial Basis Functions and Semantic Space

This paper introduces a short-term and long-term learning approach for Content-Based Image Retrieval with relevance feedback. The proposed system combines Radial Basis Function (RBF) network and the Semantic Space methods. The RBF Subsystem captures the non-linear relationship between the low-level features and the semantic meaning within an image, while the Semantic Space Subsystem stores semantic relationships between images. User’s relevance feedback is utilized for updating the low-level feature vector for the RBF Subsystem and the high-level semantic feature vector for the Semantic Space Subsystem. Using the COREL Image Database, extensive tests evaluate the performance of the proposed approach and demonstrate high retrieval rates. The system results in higher accuracy than either of the subsystems alone.