Image Retrieval by Salient Region Detection

An image retrieval method based on the salient region detection technique is proposed in this paper. Firstly, the variance image is calculated to reflect the grayscale changes in the image. Then, the method of maximum between-cluster variance (MBV) is used for the segmentation of the variance image to extract the salient region. Afterwards, color clustering features and texture features of the salient region are extracted to form a feature vector which describes the image features. Finally, using the similarity measurement method, the similar image results are produced. Experimental results have shown that the proposed salient region detection based image retrieval algorithm can achieve better retrieval performance.