Ranked feedback query technique for CBIR systems

Content-based image retrieval with relevance feedback has been widely adopted as the query model of choice for improved effectiveness in image retrieval. The performance of this method depends on the efficiency of the feedback mechanism. Current feedback methods are takes long time to make relevance feedback and fewer users friendly. In this paper, a novel feedback technique called ranked feedback query technique to reduce the delay in feedback processing, and therefore improve the system performance. Initially the color, texture and edge structure features are extracted from the images to analyze the similarity between them. Then the system starts retrieving images, from the results, the users are allowed to specify the ranks for each image based on their target. We compare the performance of this approach with conventional relevance feedback techniques in terms of computation efficiency and retrieval accuracy. The results indicate that the new technique substantially reduces response time for user feedback while maintaining the quality of the retrieval.