Image Retrieval Based on Steerable Pyramid Binary Image Projection
Chen Xiu-xi · Jisuanji fangzhen · 2014
To solve the existing problems such as incomprehensive in texture expression,huge computation and ineffectiveness in content based image retrieval,this paper proposed a new image retrieval method.In this method,first a gray-scale image was normalized in orientation.Then,it used steerable pyramid to decompose the image into sub-band images in different scales and orientations.These sub-band images were converted to binary images by the adaptive threshold value T.The binary images were then summed in row and column to get texture features.Vector intersection method was defined in the process of matching.Experiments show that the proposed method is fast.It gets a good retrieval result and is robust to common interferences.