为了在滤除噪声的同时尽可能地保持图像的边缘信息,文中提出平稳Bandelet这一概念,并将其应用到图像去噪中。标准Bandelet变换虽然可以充分利用图像自身的几何特征,但它和小波变换一样缺乏平移不变性,导致处理后的图像边缘处容易出现振荡,这影响了它的图像去噪性能。为此文中在Bandelet中引入了平稳小波变换的思想,构造出平稳Bandelet,并将其用于图像去噪。实验结果表明,平稳Bandelet保留了分解过程中对图像去噪非常有用的的冗余信息,这使其去噪性能优于标准Bandelet和小波。%In order to achieve very good denoising as well as preserving image edge details,a novel Bandelet based on the stationary wavelet,which is called stationary Bandelet,is proposed and applied to image denoising in this paper. Standard Bandelet transform can make full use of the geometrical characters of image. But denoising with the standard Bandelet transform exhibits visual artifacts which have a bad influence on result image. It is due to the lack of translation invariance. Concept of stationary wavelet transform is presented to the standard Bandelet transform to solve the problem in the paper. Experiments show that the redundant information is reserved in the sta-tionary Bandelet transform,which is very useful in image denoising. So the denoising performance of stationary Bandelet is better than standard Bandelet and wavelet.
展开▼
