Design Of System Based On Video Stream Extraction And Quality Analysis

With the rapid development in computer technology and network communication, video applications cover all fields and the quality of video is crucial for various applications. A similarity in most video appliacations based on network is that video streams are transmitted from one node to another node through wired or wireless network and the quality is easy to be affected in the process of generation and transmission, so it is important to assess the quality in different nodes.There are lots of objective evaluation models about video quality currently, but an intact and pracyical video quality assessment system which contains these models is absent so far. Thus,on the basis of summarizing,analyzing and developing the previous research achievements,this paper aim to design an intact and practical video bitstreams features extraction and quality analysis system. The main contents of this paper are as follow:Firstly,analyze existing objective quality assessment methods that based on video bitstream features. Special emphasis is placed on the systematical analysis and comprehensive summary of some bitstream features that related to video quality.Then, design a subjective video quality assessment system that based on existing main subjective video quality assessment methods,thus provide a practical solution for estabilishing a video subjective assesment database that used for follow-up quality analysis experiment.Next, analylze the widely used video coding library FFmpeg and study its policy in the video decoding process for bit-streams extraction, thus provide guidance for the extraction of video bitstreams features information.Finally, design and realize the video bitstream features extraction and quality analysis system based on FFmpeg decoder and use this system to analyze the video quality of H.264 format videos. In the test based on coding distortion, the system extracts some bit-streams from the videos as samples, which are encoded with random quantization and analyze the quality of videos with assessment based on these bit-streams. In another test based on network transmission, the live555 software is used for streaming media server and simulates the process of packet loss in VMWare. The system extracts some bit-streams from videos which are transmitted by server and analyze the quality of network videos with assessment based on these bit-streams. Both tests show that the system can extract the majority of bit-streams features inerrably and evaluate video quality with the corresponding objective assesment model accurately.