Research And Application Of Image Blur Detection In Objective Video Quality Evaluation

With the rapid development of network technology, multimedia technology gets a wide range of application in real life, such as video printable, video blog, video conferencing. However, in the process of collection, processing and transmission, the video quality will decrease because of the noise and interference which are caused by the influence of collection system, processing algorithm and transmission equipment. The degree of degradation of these video could reflect the performance of related system or the service quality of the transmission channel, so it’s important to evaluate the quality of video and it has practical significance and application value.Whether the loss of information caused by compression-coded, or the distortion caused by errors or loss frames in the network transmission, will result in blur in the video, so it is very important to detect the blur in the video, this thesis study the application of blur detection of image in the objective video quality evaluation, the main results are as follows:(1) The design and implementation of video subjective evaluation platformVideo evaluation include subjective evaluation and objective evaluation, because the ultimate observer of the video is human, so the results of subjective evaluation will have a higher accuracy, and it can be used to validate the merits of the objective evaluation algorithm, this thesis study the characteristics of the existing players at first, then according to requirements analysis, design and implement the video subjective evaluation platform, the results of experiment show that the platform is easy to operate, and it’s easy for testers to grade and researchers to get the statistical score.(2) A full reference blur ratio algorithmThe algorithm first uses edge detection algorithm to get the vertical edge image and calculates blur value of every frame of video, then marks the blur value in the coordinate of the original and distortion video, as usual, there are three kinds of distributions:dot, line and scatter, at last, uses three solutions to match the distributions: slop of mean, linear regression and minimum enclosing circle, the results of experiment show that this three kinds of solutions could give higher reference between subjective evaluation and objective evaluation of image and video.(3) Video blur dynamic optimization algorithmEdge detection is an important part of blur detection, it plays a crucial role in getting the result of the blur detection, this thesis study the classic edge detection algorithm and shortage, proposes a novel edge detection algorithm based on dynamic threshold, then combines the three solutions which previously proposed, presents a new dynamic optimization algorithm of evaluation model, the results of experiment show that the new algorithm could get a higher correlation with subjective evaluation.This thesis study the application of image blur in video quality evaluation, proposes a novel algorithm based on full reference blur ratio and a video blur dynamic optimization algorithm, improves the matching of the results of video objective and the results of subjective, that is, raises the accuracy of results of objective evaluation, it’s very important for the study of video evaluation.