IMPROVED IMAGE DECOMPRESSION FOR REDUCED TRANSFORM CODING ARTIFACTS

The perceived quality of images reconstructed from low bit rate compre ssion is severely degraded by the appearance of transform coding artif acts. This paper proposes a method for producing higher quality recons tructed images based on a stochastic model for the image data, Quantiz ation (scalar or vector) partitions the transform coefficient space an d maps all points in a partition cell to a representative reconstructi on point, usually taken as the centroid of the cell. The proposed imag e estimation technique selects the reconstruction point within the qua ntization partition cell which remits in a reconstructed image that be st fits a non-Gaussian Markov random held (MRF) image model. This appr oach results in a convex constrained optimization problem that can be solved iteratively, At each iteration, the gradient projection method is used to update the estimate based on the image model, In the transf orm domain, the resulting coefficient reconstruction points are projec ted to the particular quantization partition cells defined by the comp ressed image. Experimental results will be shown for images compressed using scalar quantization of block DCT and using vector quantization of subband wavelet transform, The proposed image decompression provide s a reconstructed image with reduced visibility of transform coding ar tifacts and superior perceived quality.