一种改进阈值的邻域小波系数图像去噪方法  

An improved threshold for image denoising based on neighboring wavelet cofficients

作　　者：吴成茂[1] 胡伟[1] 王辉[2] 

机构地区：[1]西安邮电大学电子工程学院,陕西西安710121 [2]西安邮电大学自动化学院,陕西西安710121

出　　处：《西安邮电大学学报》2014年第1期42-45,共4页Journal of Xi’an University of Posts and Telecommunications

基　　金：陕西省教育厅科学研究专项基金资助项目(2013JK1129)

摘　　要：为了能更好地恢复出原图像,对基于邻域小波系数收缩的NeighShrink法进行研究。对小波域图像去噪模型加以分析,指出原方法因其阈值不能适应尺度分解所得小波系数中噪声的强弱特性,故难以保证原图像的恢复效果,随之对原方法给出一种改进,通过修正NeighShrink法中的阈值,建立起一种用指数函数调节阈值的自适应方式来匹配小波系数的幅度特性。在不同强度高斯噪声条件下的实验结果表明,改进后的方法在有效去除噪声的同时能够更好的保留原始图像信息,且相对于原方法,改进后的方法能提高峰值信噪比。The NeighShrink cannot recover original image significantly because its threshold can- not adapt to the scale decomposition of wavelet coefficients obtained by the strength of the noise characteristics. In this paper, an improved threshold method is proposed. In this method, an a- daptive denoising way of setting up the threshold of NeighShink is proposed to regulate threshold by matching the amplitude characteristic of wavelet coefficients with an exponential function. Ex- perimental result show that the proposed method can retain the original image information much better after efficiently removing noise under different Gaussian noise intensity condition, and can obtain higher peak-to-signal noise ratio compared with NeighShrink and IIDMWT.

关 键 词：小波变换 阈值 邻域小波系数 图像去噪 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]