彩色含噪图像的空间多分辨力阈值收缩求解方法

Spatial Multi-resolution Shrinkage Thresholding for Denoisy Color Image

采用了一种新的基于小波变换的自适应多阈值图像去噪方法——空间多分辨力阈值收缩去噪法,方法首先根据信号在不同子带和不同方向上的分布特性选择不同的最佳噪声阈值,然后对彩色图像fC提取3幅分量图像fR,fG,fB,采用不同阈值分别对每幅分量图像进行去噪,最后重建去噪后的RGB合成图像。通过大量的实验证明,空间多分辨力阈值收缩去噪法能针对不同信号成分采用不同的噪声阈值,与硬阈值去噪方法和软阈值去噪方法相比,不仅提高了去噪后图像的信噪比,也使图像更加清晰,并能更好地适合人眼的主观视觉特性,可在客观和主观上同时获得更佳的去噪效果。

A new denoising method of adaptive spatial multi-resolutin shrinkage thresholding algorithm is presented. Optimum all thresholds are chosen according to different characteristris in different scales and different directions. Then ＇three component images fR, fc, fB are derived from a color image fc for denoising. Finally the denoised RGB image are recovered. Experiments are carried out to verify the SNR and HVS coherence. And comparing with those of the hard-threshold denoising and the soft-threshold deniosing, this method improves not only the SNR, but also visual perception.