基于楔波的图像压缩  

Image Compression Based-on Wedgelets

作　　者：黄仁泰[1] 李金徽[2] 张宗念[3] 

机构地区：[1]东莞理工学院计算机科学与技术系,东莞523106 [2]东莞理工学院网络中心,东莞523106 [3]东莞理工学院电子工程系,东莞523106

出　　处：《科学技术与工程》2008年第14期3757-3760,3769,共5页Science Technology and Engineering

基　　金：广东省东莞市科技攻关项目(2007108101010)资助

摘　　要：分析了楔波近似理论的基础上,采用任意局部退化模型代替局部恒定退化模型,用楔波近似方法压缩图像,并与基于小波变换和基于Contourlets的两种图像压缩方法进行了对比。结果表明:在高比特率区域,楔波压缩图像的性能超过了小波变换和contourlets方法;而在低比特率区域,contourlets和楔波则更优秀;特别是几何特征十分突出的图像,楔波压缩的效果最佳。但是楔波的不足是重构图像中有明显的边缘,这可以通过采用高阶系统如platelets加以改善。The theories of wedgelets approximation are analyzed and compressed the images with a fast algorithm of wedgelets approximation by replacing local constant regression model by arbitrary local regression model. The resuhs of wedgelets approximation experiments are presenfed and compared it with that of contourlets and wavelets. The new scheme outperforms wavelet in the high bit rate area. By contrast, contourlets and wedgelets perform better in low bit area, and wedgelets perform best when dealing with images that of a predominantly geometric nature. Wedgelets artifacts by employing a higher order system such as platelets be can reduced.

关 键 词：楔波 图像 压缩 小波 轮廓波 

分 类 号：TN919.81[电子电信—通信与信息系统]