An image retrieval method based on texture spectrum feature of salient regions
Wang Run-hong · Journal of Minjiang University · 2012
This paper proposes an image retrieval method based on texture spectrum feature of salient regions.At first,the method calculates the salience value of every pixel and constructs the salience map.Secondly,it determines the salient regions according to the salience map for the segmented regions.Then,it extracts the texture spectrum feature of image for each salient region.Finally,the weighting function of texture spectrum feature of salient regions is obtained.Experiment results show that this approach can efficiently combine the visual sensitivity with the texture feature,and improve retrieval quality without increasing computational time.