Video Quality Assessment Algorithm Based On HVS Properties

With the distribution and dissemination of a large number of digital videos,video quality assessment has become a hot focus.Unlike many signal processing applications,the ultimate recipient of video signal is always human eye.Therefore,the video quality assessment algorithm must consider the perception of the human visual system,and it is necessary to construct a quality assessment algorithm consistent with the subjective feelings of the human.Therefore,this paper proposes a video quality assessment algorithm based on human visual properties.The main work is as follows:(1)This paper proposes a framework for video quality assessment methods.Using the human visual properties to adjust the prediction quality of video frames,the support vector machine is used to learn the mapping relationship between image prediction quality and final subjective quality,and is used to predict the final objective quality score of the test set.(2)The visual attention and masking effects of the human eye have a great impact on human visual perception.Based on visual attention,this paper proposes a duration time feature.The self-information of the duration time is calculated as a representation of the visual information content.Based on the masking effect,this paper proposes a texture masking feature.Based on the traditional Local Binary Patterns algorithm,a block-level texture masking algorithm is proposed,which makes it possible to calculate the perception uncertainty caused by texture information more accurately.(3)In the training process of support vector machine,the problem of multi-parameter optimal value is usually solved by exhaustive method.In this paper,a method for fast decision of optimal parameters is proposed,which can quickly obtain the optimal multi-parameters' value under target complexity and carried out experimental verification.The quality assessment algorithm proposed in this paper is tested on the LIVE video database.The results verify that the prediction score has good consistency with the subjective score,and the feasibility of the proposed algorithm is verified.