Adaptive predictive image coding using local characteristics

In this method, three coding schemes, namely, mean, subsampling combined with fixed DPCM, and ADPCM/PCM, are used and one of these is chosen adaptively based on the local characteristics of images. The prediction parameters of the 2D linear predictor in the ADPCM/PCM are extracted on a block by block basis using the 2D lattice filtering method. The order update 2D lattice model proposed by Parker and Kayran (1984) is extended into order update and time update recursion for more efficient computation. Simulation results show that the proposed method is effective in reducing the slope overload distortion and the granular noise at low bit rates, and thus it can improve the visual quality of reconstructed images. Other advantages of this method are the low computational burden and the suitability for the VLSI implementation.