Robust Pre-processing Technique Based on Saliency Detection for Content Based Image Retrieval Systems

The perceptual quality which attracts viewer attention by making the objects differ from its neighbours is said to be Visual saliency. They concentrate on the region of interest. The paper aims at improving the performance of content based image retrieval using saliency detection approach. Several methods have been developed to extract the saliency information from an image. We use the state of the art Quaternion transform for to detect the saliency. The paper focuses on the content based image retrieval systems based on scale invariant feature transform and region segmentation. Experimental results prove that the proposed technique outperforms the existing techniques and produce better retrieval results.