自适应联合小波去噪算法

Adaptive Denoising Algorithm Based on United Wavelets

对于传统小波去噪方法,由于选用了单个的小波基,很难兼顾图像的平滑区域、边缘和纹理部分。而对于多小波基的去噪方法,尽管选择多个具有不同性质的小波基,但已有的文献中只是简单地取其算术平均,没有很好地体现小波基的多样性,造成了丢失细节与过平滑的后果。针对图像的非均匀性以及每个小波基支撑区间的不同,提出了一种自适应联合小波去噪算法,对图像中的不同区域和每个小波基处理的不同结果赋予不同的权系数,这样就充分发挥了每个小波基的作用,取得了满意的实验结果。

Wavelet denoising is a popular technique in digital image processing. With traditional wavelet denoising methods, it is hard to deal with, simultaneously, smooth regions, edges and textures using a single wavelet base. Even with the use of multiple wavelet bases of different properties, in existing literature, only the mean values are applied in the process and the differences among the bases are not used properly. This leads to the loss of some details and the over-smoothness of an image. An adaptive denoising algorithm is proposed based on united wavelets which are determined by non-evenness of images and different support sets of the wavelets. A weighted mean is used and the weights on individual wavelet bases are chosen by processing different regions. The differences of wavelet bases play important roles in this method and the experimental results are quite satisfactory.