具有自适应窗口的双变量模型图像去噪方法

BIVARIATE MODEL IMAGE DENOISING WITH ADAPTIVE WINDOWS

针对小波域去噪方法 BiShrink-local(双变量萎缩局部方差估计)会造成图像的细节信息丢失的问题,给出一种具有自适应窗口的双变量模型图像去噪方法。该方法一方面继承原方法的优点,另一方面又利用区域生长原理,通过判断图像的小波系数值是否属于同质,从而更好地区分噪声和细节信息。通过实验表明,该方法能对含有高斯噪声的图像进行较好地去噪,同时在保持细节方面优于原来的方法。

Bivariate shrinkage with local invariance estimation——a wavelet-based denoising method will cause image detail information lost.To solve this issue,we propose a bivariate model image denoising method with adaptive windows.The proposed method utilises the principle of region growing at one hand,and inherits the advantages of BiShrink-local on the other hand,by judging whether the wavelet coefficients of image are of same properties,it can preferably differentiate the noisy signal and detailed information.Experimental results show that the method can denoise better against the image with Gaussian noisy,at the same time,it outperforms the BiShrink-local in retaining the details.