The invention relates to a relevant self-adaptive
bit rate control method based on H.264 / AVC. A data quantity used for forecasting a
model parameter is self-adaptively adjusted by utilizing the relevance between video frames or
macro blocks in the method to ensure that the average data quantity of the model computation is decreased, thereby the
computation complexity is decreased. The combination of the frame layer control and the
macro block layer control is adopted in the
bit rate control method, the variation range of the quantification grade of the
macro block in the video frame is contracted, a new MAD model is established on the frame layer, and the weighting solution is operated by adopting the relevance during the process of the model solution and the RD solution; a new MAD forecasting model is established on the macro
block layer. According to the correlated characteristic of a space time between the macro blocks, the solution proposal of the MAD model and the RD model is designed. The testing indicates that compared with the G012 standard
algorithm, the invention has the advantages that the complexity of the
algorithm is low, the
bit rate control precision is improved, a buffering area is smoothed, and the PSNR value of the coding video is improved averagely by 0.4dB.