Perceptual video coding (PVC) can eliminate visual redundancy and effectively improve the compression performance of ultra-high definition video. The conventional PVC schemes, involving the perceptual characteristics related to the video features, do not account for interference in the human visual system caused by external environment changes. Therefore, this paper presents a PVC scheme based on a perceptual noise channel model. In this approach, visual dither is utilized to imitate the adjustment process of human eyes while environment changes. Considering reliable external visual stimulus decoding in the perceptual noise channel and visual characteristics, dynamic estimation of visual dither quantization is performed based on a message passing algorithm on a factor graph. Furthermore, a weight factor based on the perceptual noise channel capacity is incorporated into the rate distortion optimization process. The experimental results show that the proposed PVC scheme achieves significant bitrate reduction with better subjective video quality than competitive PVC schemes.

感知视频编码（PVC）可以消除视觉冗余，并有效提高超高清视频的压缩性能。涉及与视频功能有关的感知特性的常规PVC方案不能解决由外部环境变化引起的对人类视觉系统的干扰。因此，本文提出了一种基于感知噪声通道模型的PVC方案。在这种方法中，视觉抖动被用来模仿环境变化时人眼的调节过程。考虑到感知噪声通道中可靠的外部视觉刺激解码和视觉特征，基于因子图上的消息传递算法对视觉抖动量化进行动态估计。此外，基于感知噪声信道容量的加权因子被合并到速率失真优化过程中。实验结果表明，与竞争性PVC方案相比，所提出的PVC方案可显着降低比特率，并具有更好的主观视频质量。