Wavelet shrinkage and generalized cross validation for image denoising

We present a denoising method based on wavelets and generalized cross validation and apply these methods to image denoising. We describe the method of modified wavelet reconstruction and show that the related shrinkage parameter vector can be chosen without prior knowledge of the noise variance by using the method of generalized cross validation. By doing so, we obtain an estimate of the shrinkage parameter vector and, hence, the image, which is very close to the best achievable mean-squared error result--that given by complete knowledge of the underlying clean image.