JND-based Multiple Description Image Coding
JND-based Multiple Description Image Coding
复制标题
基于JND的多描述图像编码
DOI：
10.3837/tiis.2017.08.010
复制
发表时间：
2017
影响因子：
1.5
通讯作者：
Wan Wenbo
中科院分区：
文献类型：
--
作者：
Zong Jingxiu;Meng Lili;Zhang Huaxiang;Wan Wenbo
根据人类视觉模型的特点，提出了一种新的多描述图像编码方案。由于人类视觉的固有特性，人眼只能感知特定阈值的变化，即JND阈值。因此，本文将JND模型应用于MDC系统的改进。本文首先计算DCT系数，然后将其与JND阈值进行比较。对于小于JND阈值的数据，可以忽略不计，提高了编码效率。与其他已有方法相比，该方法的实验结果更优越。
In this paper, a novel multiple description image coding (MDC) scheme is proposed, which is based on the characteristics of the human visual model. Due to the inherent characteristics of human vision, the human eye can only perceive the change of the specific thresholds, that is, the just noticeable difference (JND) thresholds. Therefore, JND model is applied to improve MDC syetem. This paper calculates the DCT coefficients firstly, and then they are compared with the JND thresholds. The data that is less than the JND thresholds can be neglected, which will improve the coding efficiency. Compared with other existing methods, the experimental results of the proposed method are superior.