A Relevance Feedback Framework for Image Retrieval Based on Ant Colony Algorithm

Guang-Peng Chen, Yu-Bin Yang, Yao Zhang, Ling-Yan Pan, Yang Gao, Lin Shang

Abstract: To utilize users’ relevance feedback is a significant and challenging issue in content-based image retrieval due to its capability of narrowing the “semantic gap” between the low-level features and the higher-level concepts. This paper proposes a novel relevance feedback framework for image retrieval based on Ant Colony algorithm, by accumulating users’ feedback to construct a “hidden” semantic network and achieve a “memory learning” mechanism in image retrieval process. The proposed relevance feedback framework adopts both the generated semantic network and the extracted image features, and then re-weights them in similarity calculation to obtain more accurate retrieval results. Experimental results and comparisons are illustrated to demonstrate the effectiveness of the proposed framework.