The Rate Control Research Based On H.264/AVC

Rate control plays a key role in video communication, for its ultimate purpose is to transmit high quality video sequence under the limited buffer capacity and channel bandwidth. So, precise rate control model is needed, especially for low bit rate video streaming requirement.The quality and quantity of the data set are critical to accuracy of the model. With respect to the quality of the data set, these data points may sometimes deviate from RD relationship and mismatches rate control model parameters and depresses the precision. If the error is being passed on in coding process, it begets a so-called bad parameters phenomenon (BPP). In this paper this phenomenon is studied and a revision algorithm is presented.In order to improve the video perceptual quality, the texture sensitivity is studied based on the Human Visual System (HVS) characteristics. A novel method used for improving the subjective visual quality is proposed.The fluctuation of encoded data potentially mismatches rate control model parameters and depresses the precision of the rate control model due to nonstationary characteristics of video sources. In this paper, a novel sliding-window method for data selection is proposed. The data points for updating a model are adaptively selected such that the statistical behavior is improved. Especially, at MB rate control scheme, our sliding-window mechanism is adept at adaptively selecting the encoded data of MB from spatial and temporal, in this way, which well exploits the spatial, temporal and statistical redundancies. Furthermore, we propose a simple and effective algorithm to remove some erroneous data points from the data set. Moreover, a new model parameters updating algorithm is exploited to enhance the model accuracy.When compared with the rate control scheme JVT-G012 which is adopted by the JVT H.264/AVC reference model JM8.6, our proposed rate control scheme can achieve a significant average peak signal-to-noise ratio (PSNR) gain of up to 0.4 dB for the test video sequences, the buffer is better regulated, and reduces the computation complexity. Our method is especially suitable for low bit-rate video communication.