单通道视频编码的可变码率控制方法

为提高实时视频编码的图像质量,提出了一种低复杂度单通道视频编码的可变码率(VBR)控制算法.通过简化经典率失真R-D模型推导得到图像复杂度和空间活动性的线性公式,根据运动估计过程的残差信息计算当前编码图像的复杂度.在编码过程中,动态地改变该公式的参数来适应各种不同的视频场景.利用图像复杂度和编码比特数的线性模型来分配当前图像的目标编码比特数,同时结合先前实际编码比特率和期望比特率之间的偏差调整目标比特数.实验结果表明,该算法可以有效地进行码率控制,对场景复杂度变化有很强的适应性,得到相对平稳的图像质量.与TM5算法相比,平均峰值信噪比(PSNR)提高了0.7～1.0dB.

Variant bit rate control algorithm for single-pass video encoding

To improve the image quality of real time video encoder, a low complexity variant bit rate (VBR) control algorithm for single-pass video encoding was proposed. A linear equation between complexity and spatial activity of image data was derived by simplifying the classical rate distortion (R-D) model. The equation parameters were yielded dynamically to adapt to various video scenes during the encoding procedure. The complexity of current picture was computed with the information in the motion estimation. The proposed algorithm employed the linear relationship between bit number and picture complexity to allocate target bit number of current picture. Meanwhile, the deviation between previous actually encoded bit rate and expected bit rate was used to adjust target bit number. Experimental results show that the proposed algorithm can implement a smooth quantitative video quality and improve perceptual quality dramatically. Compared with TM5,the average peak signal-noise ratio (PSNR) gain is about 0.7~1.0 dB.