A New Multi-query Relevance Feedback Approach in Content Based Image Retrieval

In this paper, a new multi-query relevance feedback approach is proposed for content based image retrieval. To realize the multi-point queries, a hierarchical clustering method is applied on the related (positive) images and the clusters’ representative (here center) are selected as new queries vectors. Then, these queries vectors are moved in the feature spaces towards related images. In contrast with other existing multi-query relevance feedback methods, the proposed method acts separately on each feature space and the output results are combined to perform the final result. To evaluate the performance of the proposed method, a set of comparative experiments on a general image database containing 10000 images of 82 different semantic groups are performed. The results confirm the efficiency of the proposed method comparing to well-known conventional methods.