A SCENE ADAPTIVE HYBRID VIDEO CODING SCHEME BASED ON THE LOT

Conventional standard video coding schemes based on independent coding of nonoverlapping image blocks produce undesirable effects (blocking effect and color bleeding, etc.) at low bit rate, In this paper, a hyb rid coding scheme evolving from the standard MPEG codec that combines lapped orthogonal transform (LOT) and overlapped block motion compensa tion and quantization approaches based on human visual system sensitiv ity is proposed to remove the artifacts. The LOT is applied to compres s the resulting prediction errors of the overlapped motion compensatio n. The LOT coefficients are quantized by a dynamic intra/inter scene a daptive quantizer that is designed according to the human visual syste m and scene analysis to optimize the decoded image quality. This techn ique involves automatically deriving the most favorable quantizer matr ix from the motion and scene analysis of image sequences and scaling t he resulting matrix furthermore by considering the buffer status. If s cene change is detected from the motion information, the proposed adap tive quantization fan also help solve the difficulty of the buffer con troller and maintain the picture quality as constant as possible. Perf ormance of the proposed hybrid coding scheme with scene adaptive quant izer is demonstrated by computer simulations using standard test seque nces and compared to that of the convention MPEG-2 coding schemes. Sub jective and objective tests show that image quality is improved consid erably and the required bit rate is much less than that of the convent ional method. Particularly, the image quality can be kept almost consi stent, and the undesirable artifacts are reduced considerably.