Image retrieval based on saliency detection in the application of the guide system

A new saliency detection algorithm based on image segmentation and color histogram is proposed. The saliency values of partitions are decided by the spatial location and color difference which is measured by the difference of same color pixel number between the regions. In order to enhance salient areas and inhibit background noise a selective enhancement suppression method is used. Through the comparison of using the enhanced suppression method and not using by experiments, a better saliency map can be obtained by using this method. An image retrieval algorithm based on image saliency detection is proposed, which extracts salient areas from saliency map and divides them into blocks. Then the features of the images to search are composed of average saliency value and spatial location of each block. Finally the method is applied to guide system, which can quickly and accurately retrieve tourism information.