A method for Content-Based Image Retrieval using visual attention model

In this paper we present a new method for Content-Based Image Retrieval, in which regions of interest (ROIs) are being extracted from images using visual attention models. After finding salient map of regions in image, which humans pay attention to those region, we calculate Histogram of Orientation Gradient (HoG) and some useful features for those regions to make a feature vector in order using in retrieval process. Whereas Saliency Detection finds important regions in image, but it is insufficient to compare two objects, Because two objects may have different color, orientation and some of another aspect. For this we use those features to make a similarity measure to take account another aspect similarity between two objects. The experimental results demonstrated that the proposed method which uses those features to seek image in a database more efficiently rather than traditional methods.