The Research of Image Quality Assessment Methods

摘要：

In digital transmission, images may undergo quality degradation due to lossy compression and error-prone channels. Efficient measurement tools are needed to quantify induced distortions and to predict their impact on perceived quality. In this paper, an artifcial neural network (ANN) is proposed for perceptual image quality assessment. The quality prediction is based on image features such as EPSNR, blocking, and blur. Training and testing of the ANN are performed with the mean opinion scores (MOS) provided by the Laboratory for Image and Video Engineering (LIVE). It is shown that the proposed image quality assessment model is capable of predicting MOS of the five types image distortions.

关键词： Image quality assessment Artifcial neural network MOS LIVE

作者: Xiaonan Cui Zhiyuan Shi Jianan Lin Lianfen Huang

作者单位: Dept. of Electronic Engineering Xiamen University Xiamen, China Dept. of Communication Engineering Xiamen University Xiamen, China

会议类型: 国际会议

会议名称: 2010 Second Asia-Pacific Conference on Information Processing(2010年第二届亚太地区信息处理国际会议 APCIP 2010)

会议地点: 南昌

会议语种:英文

页码: 35-38

在线出版日期: 2010-09-17（万方平台首次上网日期，不代表论文的发表时间）