A Stochastic Technique for the Removal of Artifacts in Compressed Images and Video

Abstract

The perceived quality of images and video sequences reconstructed from low bit rate compressed bit streams is severely degraded by the appearance of coding artifacts. This chapter introduces a technique for the post-processing of compressed images based on a stochastic model for the image data. Quantization partitions the transform coefficient space and maps all points in a partition cell to a representative reconstruction point, usually taken as the centroid of the cell. The proposed technique selects the reconstruction point within the quantization partition cell which results in a reconstructed image that best fits a non-Gaussian Markov Random Field image model. This approach results in a convex constrained optimization problem that can be solved iteratively. Efficient computational algorithms can be used in the minimization. This technique is extended to the post-processing of video sequences. The proposed approach provides a reconstructed image with reduced visibility of transform coding artifacts and superior perceived quality.