Improving the Quality of Reconstruction of Noisy Images in the Wavelet Region

Abstract

The authors suggest an approach to improving the quality of reconstruction of noisy compressed images. To reduce the amount of transmitted data, they offer to use only a subband image after a low-frequency wavelet filter with the highest concentration of signal energy. To restore a noisy image in the wavelet region, which is a random Markov field, it is proposed to use a two-dimensional nonlinear filtering algorithm based on a mathematical model of a two-dimensional Markov chain. The considered method of recovering noisy compressed images is effective at signal-to-noise ratios up to -6 dB and compression ratio up to 64 times when computing resources on the transmitting side of the channel are reduced.