Image Retrieval Base on Dual-Tree Complex Wavelet Transform and Gray Co-occurrence Matrix
Zhang Zun-wei · Xihua Daxue xuebao. Zhexue shehui kexue ban · 2010
There are shake,translation change,mixing and direction lack problems about traditional DWT wavelet transform.In order to work out the disadvantage of DWT wavelet transform,this thesis use the dual-tree complex wavelet transform to decompose query images and target image in image retrieval,which can distill six directions features.Then gets four statistical features by using the gray co-occurrence matrix of four directions,which can make up the deficiencies of the lack of different scale spatial distribution of texture.At last,uses Canberra distance to measure the similarity to output image search result.The experiment which used all kinds of texture image in Brodatz image database proves that this method obtained a better result about texture images.