Adaptive nonlinear enhancement of coarsely quantized video sequences

A computationally simple algorithm to enhance the visual quality of coarsely quantized video sequences is presented. This algorithm uses nonlinear filters to smooth discontinuous edges and granular noise associated with the quantization process. The nonlinear order statistics filtering approach does not attempt to preserve the edges, but to fill in and smooth the edge discontinuities. With a minor modification, these filters also function as controllable low-pass filters. The structure of these filters is suitable for cost-effective VLSI implementation. Simulation experiments show the efficiency of the enhancement algorithm.<>