一种基于人视觉系统的DVC改进算法

The Improvment of DVC Algorithm Based on Human Vision System

为了克服无线多媒体传感器网络中的视频传感节点的处理能力、存储空间和能量严重受限的问题,本文提出一种基于人的视觉系统的分布式视频编码的方案。由于人眼睛的潜在的感受性和视觉掩蔽性能,最小可觉差值以下的变化几乎不能被人的视觉系统感知到。因此,原始帧和边信息之间在感觉阀限之外的信号误差不需要得到纠正。本文基于这种思想,将JND模型应用到WZ编码器中,并对其仿真,实验结果表明,本文所提出的算法在不改变主观质量,甚至提高主观质量时可以较大程度对视频序列进行压缩,降低比特数,具有较好的率失真性能。

In this paper,we propose a distributed video coding （DVC） scheme with the human visual system （HVS）to overcome the limited memory capacity and low power consumption of video sensor nodes in Wireless Multimedia Sensor Network.HVS can seldom senses any changes below the just noticeable distortion （JND） threshold around a pixel due to their underlying temporal/spatial sensitivity and masking properties.On this view,it is unnecessary to spend parity bits on correcting the un-noticeable signal differences between the original frame and side information.Experimental results show that the proposed scheme can significantly improve the WZ coding efficiency in terms of bit rate having subjective quality performance guaranteed.