Signal and image denoising in transform domain and wavelet shrinkage: A comparative study

In this work, nonlinear local transform domain filtering is reviewed, and its relation with wavelet denoising is discussed. A postprocessing stage is applied to a number of transform domain denoised signals to obtain a better estimate of the original signal. Simulations are made over different Gaussian noise corrupted one-dimensional signals and images, in DCT and wavelet transform domains. Their performances with respect to threshold, transform basis and window size are compared.