Improving the Performance of Content-based Image Retrieval Systems with Color Image Processing Tools

Most of existing Content-Based Image Retrieval (CBIR) systems operate under the query-by-example (QBE) paradigm, by which an example image is presented to the system and the user queries for images that are similar to the given example. Performance of these systems is highly dependent on the properties of the example image. In many cases, the user has an image that could be used as an example if she could quickly retouch it before submitting the query. This paper describes the development of a tool, Mirage, that encapsulates several useful color image processing and manipulation operations. Mirage is available as a research prototype and can be used either stand-alone or integrated into a CBIR system, MUSE. A summary of results of experiments using Mirage to improve the performance of MUSE under the QBE paradigm is presented. These results show that the addition of Mirage to MUSE improves the retrieval performance, both from the point of view of precision as well as recall, without posing significant additional burden to the user.