Study to the image denoising algorithm based on multiwavelet transforms
Xiaowei Zhang, Lei Zhu, Xiongbo Zheng · 2007
Compared to the scalar wavelet, the multiwavelet has the properties of orthogonality, short support, real symmetry, high order vanishing moment and so on, however, it does not work well in image denoising techniques, primarily due to not making full use of the characteristic property of images in the multiwavelet domain. By transforming the noisy images to the multiwavelet domain, and applying a special difference scheme of the Laplacian operator and also considering of the image’s fractal dimension of high frequency ubband in the multiwavelet domain, the paper proposes an adaptive multiwavelets thresholding algorithm AMT algorithm, which can automatically determine the wavelet shrinkage thresholding in the multiwavelet domain without a priori knowledge of image, for instance, the variance of image noise. The result of the simulation experiment indicates, that the effect of the AMT algorithm is perfectly well, especially for high degraded images.