Methods for filtering optical images corrupted by additive white noise based on using Haar transform in blocks are considered. To control quality, both standard metrics and visual quality metrics PSNR-HVS-M and MSSIM are used. Test images of different complexity for several levels of noise are analyzed. It is shown that efficiency of Haar based denoising is slightly worse than DCT based filtering but computational efficiency is considerably better.
Haar transform, filtering, processing acceleration, image quality