Feature Performance Evaluation in Content-Based Image Retrieval

In content based image retrieval（CBIR）, image has various inherent aspects, which reflect its contents, therefore, it needs different similarity measures to improve the retrieval performance. Features and their corresponding similarity measure selections are important research topics in CBIR. In this paper, image features used in CBIR system on different similarity measures are evaluated, and a variety of features on different image collections are also evaluated, which can be the basis of the CBIR system development. According to the experimental results, feature’s retrieval performance is not only related to similarity measure but also relies on image collection.