Dominant sets clustering for image retrieval

In image retrieval algorithms, retrieval is according to feature similarities with respect to the query, ignoring the similarities among images in database. To use the feature similarities information, this paper presents an application of dominant set clustering (DSC) to image retrieval system. Combining the low-level visual features and high-level concepts, the proposed approach fully explores the similarities among images in database using DSC and optimizes the relevance feedback results from traditional image retrieval system by clustering the similar images. To test its retrieval performances, we presented an image retrieval system using the memorized support vector machine (SVM) relevance feedback. The results of experiments on the images from Corel Image Database show that the proposed approach can greatly improve the efficiency and performances of learning machine, as well as the convergence to user's retrieval concept. Comparisons on retrieval precision, total feedback time of method with and without DSC were also made, which indicated an improvement by 6.79% over the average precision and less total relevance feedback times after using DSC.