Method for improving the quality of restoration of noisy compressed images.

A method for improving the quality of restoration of noisy compressed images is proposed. To reduce the amount of transmitted data, it is proposed to use only a subband image after a low-frequency wavelet filter with the highest concentration of signal energy. Taking into account the nature of the statistical relationship between the wavelet coefficients within the same subband and in the same scale, it is proposed to approximate subband images by two-dimensional Markov chains with several states, and bit planes - by Markov chains with two states. To restore a noisy image in the wavelet region, it is proposed to use a recurrent two-dimensional nonlinear filtering algorithm that effectively implements statistical redundancy. The paper presents the results of modeling the proposed method. The superiority of the algorithm of two-dimensional nonlinear filtering in comparison with the wavelet filter is shown when restoring images without compression. The considered method of recovering noisy compressed images retains its efficiency at signal-to-noise ratios up to -6 dB and the compression ratio up to 64 times while reducing the computing resources on the transmitting side of the channel. The proposed approach to restoring noisy compressed images reduces computing resources on the transmitting side of the channel by eliminating the channel encoder. [ABSTRACT FROM AUTHOR]