New image denoising approach based on Tetrolet transform
Kang Yao-hong · Jisuanji yingyong yanjiu · 2010
In order to denoise image,this paper proposed a new approach by minimizing the mean squared error (MSE) on the Tetrolet transform domain.This approach took advantage of a multi-scale framework and directionality of Tetrolet transform to preserve the significant information of original image like edges and details.And instead of the global soft-thresholding algorithm,performed the Stein's unbiased risk estimate approach.Numerical results show the effectiveness of the technique.The higher PSNR values and the better visual quality show that the proposed new algorithm is effective.