The Video Quality Assessment Technique Based On The HVS Characteristic

Digital video image compression and transmission technology will result indifferent degrees of distortion. Therefore, assessment of video quality in video compression, processing, and video communications have very important significance, especially the automated quality assessment technology plays an important role in various applications. The existing quality assessment techniques are mainly divided into two categories, the subjective quality assessment(SQA) and the objective quality assessment(OQA). The SQA directly reflects the subjectiveperception for video quality of the human eye, and it also is the standard to evaluate a quality assessment is good or not. However, it is cumbersome, timecomsuming and can not be used for realtime automated assessment. The existing objective quality assessment technique is simple and easy to implement quality, but the absence of consideration or with little regard to human visual system(HVS) characteristics result in that the assessment is not consistent with the subjective perception.At the same time, there are many limitaions in the OQA.This article studies and analyses the existing video quality assessment techniques at home and abroad indepth, then proposes the No-Reference video quality assessment model and develops the prototype system basing on fully taking into account the human visual system characteristics. First, the model extracts the most interesting areas of human eyes from the image, which are consideredas the focus region of the evaluation model. To improve the efficiency of themode, only the focus areas' block effects are detected as an important basis for the assessment. considering the texture detail for the image, this model also extracts the energy,entropy, correlation and other image features.All the above information acting as the impact factor of the model are entered into the SVM(Support Vector Machine) which are trained with the sample images, then the SVM will obtain the quality of assessment of the single image. For the roughness of the existing technology when implementing assessment from single image to video, this model achieves the assessment of the video through proposing a move-weighted quality assessment based on quality of the single image.Finally, this article designs and develops the prototype system and analyses the results. It can automating evaluates the video real-time, and verifies the validity and advancement in accordance with the consistency to the objective quality evaluation, cpu utilization and memory utilization.