基于第二代轮廓波变换的消噪方法  

Second generation contourlet transform based denoising approach

作　　者：陈新武[1] 田金文[2] 沈绪榜[2] 

机构地区：[1]信阳师范学院物理电子工程学院,河南信阳464000 [2]华中科技大学多谱信息处理技术国防重点实验室,武汉430074

出　　处：《计算机工程与应用》2008年第2期5-7,共3页Computer Engineering and Applications

基　　金：国家自然科学基金(the National Natural Science Foundation of China under Grant No.60572048/F010204) 。

摘　　要：提出一种基于第二代轮廓波变换的消噪方法,该方法在阈值选取上采用基于自适应冲激响应的多尺度多方向阈值,在硬阈值处理之后采用自适应维纳滤波。该方法充分利用了第二代轮廓波变换在频域和时域的局部化、多尺度、多方向和低冗余特征;能够利用更多较小的轮廓波变换系数以减轻轮廓波硬阈值消噪带来的纹理效应,并能更好适应人类的视觉特征。实验结果表明提出的消噪方法在峰值信噪比和视觉效果上都优于第一代轮廓波和小波的消噪效果。The second generation contourlet transform based de-noising method is proposed,its threshold selection is determined by impulse response of every multi-scale muhi-direction followed by wiener adapted fiher.The method adequately utilizes the frequency and time domain localization,muhi-scale,muhi-direction and low redundancy characters of second generation contourlet transform;more smaller contourlet transform coefficients can be used to alleviate the artifacts effect produced by contourlet hard threshold,and is more suitable for human visual system.Experimental results show that the proposed de-noising method is superior to those of the method based on wavelet and first generation contourlet transform.

关 键 词：图像消噪 第二代轮廓波变换 多尺度几何分析 冲激响应 峰值信噪比 

分 类 号：TN911.7[电子电信—通信与信息系统]