Adaptive rate control for dynamic bandwidth in video transcoding

To overcome the overflowing and information loss caused by huge volume compressed video transmitting over dynamic changeable bandwidth, in this work, we propose a frame layer adaptive rate control scheme suitable for the video transcoding from mpeg2 to H.264 which is able to convert the bit rate of the transcoded bitstream according to the variable bandwidth. The original compressed information such as quantization parameter, input frame bit rate and the input and output bandwidth can be utilized to derive the new frame target bits. The proposed rate control strategy estimates the current transcoding frame targets bits employing rate complexity model which is capable of tracking input frame characteristic. Also the accumulated output frame complexity computed from the transcoded output video stream is considered to keep the constant quality without large motion jerkiness and quality flicker. In addition, a low complexity quantization parameter estimation model is proposed. Experimental results show the new rate control for video transcoding can achieve better video quality and more accurate video rates compared with the h.264 rate control scheme adopted by the cascaded transcoder. Also, the new adaptive rate control transcoder nearly improve speed 70% than the cascaded transcoder.