Research On No-Reference Video Quality Asessment

In the Internet era, with the rapid development of computer and communication technology, the applications of video are becoming popular. The applications of video include remote medical service, internet TV, video conference and so on. There are many applications using video such as QQ, Wechat and Miaopai. The video service has become the necessary way of getting message and communication.But when the video is transferred, there are many reasons to affect the video’s quality. In the receiving terminal, there are many types of distortion, such as block effection, White Gaussian noise and ringing effection. The distortion will affect the user experience, and limit the application of video. So it is necessary to have a no reference video quality assessment adapt to many kinds of distortion. Video is composed of a sequence of images, so this paper first researches the image quality assessment method. And based on the image quality assessment, this paper researches the video quality assessment method. There are some research achievements:1)This paper classifies and summarizes the existed image quality assessment. Through analyzing the design type of quality assessment method and summarizing the selected type of characteristic values, this paper gets the key of designing the no reference image quality assessment.2) This paper proposes a new no reference image quality assessment algorithm based on significant extraction and natural scene statistics. This algorithm first extracts the salient region and get the statistical characteristics of whole image and the salient region. Through computing the difference between the statistical characteristics and natural scene statistics and giving more weight on the salient region, this paper gets the quality of image. This algorithm has a good performance with the subject quality.3) Based on the above no reference image quality assessment algorithm, and through studying the characteristics of video motion and the characteristics of human visual system, this paper proposes a new no reference video quality assessment algorithm. This algorithm first gets the quality of the sequence of images, and uses the characteristics of video motion and the change of image’s quality to correct the single frame’s quality. At the end, this algorithm obtain the signal frame’s quality to get the video’s quality. Experimental results show that this algorithm is consistent with the subjective perception.