Rate Distortion Optimization Method Based On Gaussion Distribution

In today’s hybrid video coding, rate distortion optimization（RDO）plays animportant role.Through an accurate RDO method we can get a kind of code parameters,and obtain a better balance between the bit rate and the require of the definition of thecoded picture, and so on assume to the requirement of the channel bandwidth and theCoding buffer and meet the requirement of the picture.At present, some rate distortion model have been put forward, such as the classicalmodel used in the JM17.2, r percentage model, and the model which based on thedistribution law model of the residual coefficients is Laplace, and so on. However,,thefirst and the third one’s assumed condition is not very precise, the0percentage modelcan not adaptive to the input bit stream. Therefore, these RDO models are not veryperfect.In this paper, we put forward a new RDO method.Compared of method before thispaper, the R-D model in this paper have a advantage in both the adaptability and theaccurate. In the analysis of this paper, we consider that when the QP is low, theregularity of the residual is Laplace, and when the QP is high, the regularity of the DCresidual is Gauss and the regularity of the AC residual is Laplace. Because the DCresidual plays an important role in the coded stream, we optimize the Lagrangianoptimization and use it in RDO.