System Of Video Quality Assessment Based On Visual Characteristics

In recent years, with the rapid development of computer technology, a lot of impossible ideas before become a reality,the way people access information changes escpecially most.Initially people use voice to communicate for information, Later it is developed to use images, and the great progress of network technology makes video begin to develop.Since the video is the most intuitive and effective way to communicate, so the demand for video business growing with each day.A lot of video produced lead to a huge amount of information to be processed.In order to reduce the amount of information transfer, video compression coding is needed.But the compression coding can cause the video quiality deteriorate for the loss of part of the information.At the same time, the network environment for video transmission, because of the instability of the network,will have an impact on the video quality.Faced with a lot of video,the video quality assessment is needed in order to select good quality video.Therefore, the study of video quality assessment method becomes particularly important. There are two kinds of video quality assessment methods of existing, subjective assessment is the people’s subjective feelings, but its realization is difficult, objective assessment method of realization is simple, but the result is not consisit with hunman’s subjective feelings.The purpose of this paper is to come up with an effective and objective video quality assessment method, and on this basis to develop a software system that can evaluate the quality of video, and ultimately get the result that can be in conformity with the subjective feelings. This paper firstly introduces the origin of topic and the significance of the research on video quality assessment.Secondly, the existing video quality assessment method is researched and analysised in depth, and then this paper summarizes and analyzes the advantages and disadvantages of various methods. Because the human visual characteristics have an important role in the video quality assessment, this paper studies the human visual system, and summarizes the human eye "brightness contrast masking" and "motion masking" and "scence switching masking" features, these features are fully considered in the method.Followed by a description of the design of the overall structure of the system and structure of each module.The design of this system consists of three parts, including voideo decoding module, video conversion processing module and video quality assessment module, wherein the video quality assessment module is the core module of the system. Finally this paper carries on the detailed elaboration on the quality of video module in this system.The finally video quality scores obtained by this system are determined by three parts, including block effect considering eye characteristic, block effect calculation considering eye "contrast" characteristics and including the image texture features.After get image texture features, the block effect and texture are treated as the mass fraction of each frame image of the trained SVM. According to the "motion masking" feature to get the movtion weight of each frame image, accoring to the "scence switching masking" to get the switch weight of each frame.The finally video quality score is get by the weight calculation of each fram image. A sofeware system is introduced based on this method.Finally, the system is verified.This method is proved to be more accurate through analysis of large number of video files and the evaluate results, the result has a good agreement with the hunman subjective feelings.