Research On Video Rate Control Algorithm Base On Compressed Domain

With video products being used in all walks of people’s life, the demand for highquality products is also expanded, under this background, video coding andtransmission technology has made tremendous development. Compared with the othercoding standards in the terms such as the compression ratio, quality, and networktransmission,H.264video coding compression standard, namely the tenth part ofMPEG-4, has achieved significant improvements. So H.264is gradually replacing theother standards to be the latest and the most common-used standard. Rate control isone of the key technologies of various practical application of video codingtechnology, it adjusts the output bit rate through adjusting the coding mode andchanging the quantified parameters, thus,the output bit of bit rate can be reasonablyallocated and better reconstructed picture quality can be achieved. As can be seenfrom this,it means a lot for rate control in video coding.The GO12algorithm used in H.264standards is researched in this paper, as forits shortcoming, the improvement plan is proposed.The G012algorithm use linear prediction model to predict MAD(Mean AbsoluteDifference),in this way, the problem of “the paradox of laying hens” is kicked. Forthe two model coefficients is gotten through the linear regression method and thetemporal and space correlation is not used,thus when processing videos that hasfast-changing backgrounds,this method will produce extreme prediction errors andforce the prediction model to be invalid. To cope with the fast-changing backgrounds,the luminary moving vector of the current frame and the previous frame is used as theweight to MAD of the present Macro block.when processing issue of P frame basic unit bit allocation, the algorithmadopted the average distribution method. When processing frame images with highcomplexity, this allocation program will make the BU of images with highcomplexity has low reconstructed image quality and the BU of images with lowcomplexity has high reconstructed image quality, which make the image in the sameframe has not even clarity and the whole video quality is effected. As for this flaw,combined with the temporal and space correlation, this paper use the correspondingPSNR of the previous frame to roughly stand for the image complexity of the currentimage. The BU object bits of the current coding frame is rationally allocatedaccording to the predicted complexity.Test and analyzing work has been done with the algorithm proposed in this papercombined with the software JM8.6. The experimental results show that comparedwith the JVT-G012algorithm, the rate control algorithm proposed in this paper whichpredicts MAD on the foundation of weighted image brightness difference andallocates BU based on the PSNR, can ensure the rate near the object rate more accurately and also get higher PSNR. More prominent effects can be gotten especiallyin processing the videos that has fast changing backgrounds.