A relevance feedback approach based on similarity refinement in content based image retrieval

In content based image retrieval systems, the suitable visual features are extracted from images and stored in the feature database Then the feature database are searched to find the most similar images to the query image. In this paper, three types of visual features by 270 components were used for image indexing. Here, we use a weighted distance for similarity measurement between two images. This paper presents a new relevance feedback approach based on similarity refinement. In the proposed approach, weight correction of feature’s components is done by a proposed rule set using the mean and standard deviation of feature vectors of related (positive) and non-related (negative) images. Also, the weight of each type of features is adjusted according to the related images’ rank in the retrieval with this type of feature. To evaluate the performance of the proposed method, a set of comparative experiments on a general image database containing 10000 images of 82 different semantic groups are performed. The results confirm the efficiency of the proposed method comparing by well-known conventional methods.