The Cycle Spinning-based Sharp Frequency Localized Contourlet Transform for Image Denoising

摘要：

Contourlet transform provides flexible number ofdirections and captures the intrinsic geometricalstructure of images.The efficient directional filterbanks with low redundancy of contourlet are veryattractive for image processing.However,non-idealfilters are used in the original contourlet transform,especially when combined with laplacian pyramid,which results in pseudo-Gibbs phenomena aroundsingularities for image denoising.Sharp frequencylocalized contourlet transform(SFLCT)is a newconstruction contourlet to overcome this drawback byreplacing the laplacian pyramid with a new multiscaledecomposition which significantly improve thedenoising performance than the original form.Unfortunately,the downsampling of SFLCT makes itlack translation invariance.Thus,we employ a cyclespinning(CS)method to improve the denoisingperformance of SFLCT,named as cycle spinningbased SFLCT(CS-SFLCT),by averaging out thetranslation dependence. Experimental resultsdemonstrate that the proposed CS-SFLCT outperformsSFLCT,contourlet and cycle spinning-basedcontourlet for denoising in terms of PSNR and invisual effects.

作者: Qu Xiaobo Yan Jingwen

作者单位: Dept.of Communication Engineering,Xiamen University.Fujian,P.R.China,361005 Key Lab of Digital Signal and linage Processing of Guangdong Province,Shantou University.Guangdong,P

会议类型: 国际会议

会议名称: 2008 3rd International Conference on Intelligent System and Knowledge Engineering(第三届智能系统与知识工程国际会议)(ISKE 2008)

会议地点: 厦门

会议语种:英文

页码: 1247-1251

在线出版日期: 2008-11-17（万方平台首次上网日期，不代表论文的发表时间）