An analogy-relevance feedback CBIR method using multiple features
Huimin Xie, Ying Ji, Yueming Lu · 2013
Since traditional relevance feedback content based image retrieval (CBIR) methods need several rounds of search, in this paper we put forward an analogy-relevance feedback (analogy-RF) CBIR method using multiple features which only needs one. The method allows users to choose the kind of object of the query image when they input the query image, and our system can determine several analogy-RF images in the sample database. Then we can use analogy-RF images to revise the similarity of images and retrieval and sort images by the re-calculated similarity. The experiment result on COREL 1k image database shows the effectiveness of the proposed method.