论文部分内容阅读
深入研究了基于模型的编码方法，提出了基于内容的网格模型生成方法，通过从边到点的模型生成思想，能得到更好地反映图像结构的模型．在六边形匹配算法的基础上，提出了快速的模型运动估计算法，改进后的运动估计算法在基本上保证了原有运动估计的信噪比的同时，大大地缩短了运动估计的计算时间．模型补偿后的差值图像由一个自适应位分配策略的编码方法编码．实验结果表明，基于模型的编码方法在低码率下能得到较好的解码质量，可用于视频电话等低码率实时编码应用中．
In-depth study of the model-based coding method, proposed content-based grid model generation method, through the generation of ideas from the edge to the point, can get a better model of the image structure. On the basis of the hexagon matching algorithm, a fast model motion estimation algorithm is proposed. The improved motion estimation algorithm substantially guarantees the signal-to-noise ratio of the original motion estimation and greatly shortens the calculation time of the motion estimation . The model-compensated difference image is encoded by an adaptive bit allocation strategy encoding method. Experimental results show that the model-based coding method can achieve better decoding quality at low bit rates and can be used in real-time coding applications with low bit rates such as video telephony.