篇名 | Using Analytic Hierarchy Process to Assess Network Video Quality |
---|---|
卷期 | 31:1 |
作者 | Zhiming Shi 、 Chengti Huang 、 Jianeng Tang |
頁次 | 106-118 |
關鍵字 | analytic hierarchy process 、 correlation coefficient 、 principal component analysis 、 quality of experience 、 video quality assessment 、 EI 、 MEDLINE 、 Scopus |
出刊日期 | 202002 |
DOI | 10.3966/199115992020023101009 |
Nowadays, people watch network video everywhere. Network video has become hot service of Internet. However, many factors may impair the network video quality. The video quality of experience (QoE) is difficult to define. The research of video quality assessment has become a hot topic for service providers. But the objective assessment method is unsure and needs a lot of work. This paper proposes a comprehensive objective assessment method of network quality based on analytic hierarchy process (AHP). The method considers quality of content (QoC), quality of terminal (QoT) and quality of service (QoS) as impair factor. Firstly all the impair factors are extracted and preprocessed. Secondly they are analyzed and optimized by correlation coefficient (CC) and principal component analysis (PCA). This step can find which factors have more close relationship with the video quality and reduce the redundant factor. Thirdly the AHP is used to measure the weight of optimized impair factors. Lastly the proposed method is constituted of different impair factors and gives the objective scores. This method has many advantages: 1. Many factors are considered in a method, such as video parameters, network parameters and performance of terminal. This method is more comprehensive. 2. The extraction of parameters has been optimized by CC and PCA to reduce the dimension. The factors are more concise and clear. 3. The use of AHP is an innovation in this domain. It can effectively establish the mapping relationship between the impair factors and QoE, and get accurate objective results. Meanwhile it can adjust the weight to improve the objective scores. This paper gives the detailed experimental results, and verifies the effectiveness of the method. People watch the videos under different network environment and give the subjective score. Next the proposed method calculates the objective score. So the similarity between the subjective and objective score can be compared. At the same time, other objective methods are used to compare with this method. The experimental results show that this method can better improve the similarity between subjective and objective score.