Objective Quality Evaluation For Screen Content Images

With the rapid development of mobile internet,cloud computing and Internet of Thing(Io T)technology,multi-client interactive multimedia applications are increasing,and the client screen image gets more and more attention and application.Screen content images typically contain media forms such as images,text,and pictorial.In the process of acquiring,encoding,transmitting and displaying,the screen content image inevitably introduces various kinds of distortion,which leads to the decrease of human visual perception effect.Therefore,how to accurately evaluate the screen content image quality it is a very important problem in the field of screen content image technology.The image quality assessment has a very broad research foreground.In the image acquisition system,the image quality can be monitored and adjusted dynamically according to the results of image quality assessment.In the network digital service business,the image quality assessment model can be used to monitor the quality of network transmission video and the distribution of data stream resources in real time.In video coding,a priori theoretical knowledge can be established for the development of video coding standards.The existing image quality assessment method is the object of the natural image.Because the screen content image differs from some characteristics of the traditional natural image,the traditional natural image quality assessment method is not suitable for the screen content image.Therefore,it is necessary to study real-time accurate image quality assessment method of screen content.Aiming at the image of screen content,based on the characteristics of human visual system,this study puts forward a quality assessment method of screen content based on the visual characteristics of human,including the quality assessment method of screen content image based on gradient similarity,the method of blind quality assessment screen content distortion,the screen content quality assessment method based on structure information,evaluation of screen content quality based on support vector machine(SVM).A method of image quality assessment of screen content based on the gradient magnitude similarity is proposed.According to the human visual perception of text and image area of screen content image,this paper proposes a method of image quality assessment of screen content based on the gradient magnitude similarity from the perspective of human perception and the principle of structural degradation.The algorithm firstly presents a gradient magnitude calculation method which can effectively capture the distortion characteristics of screen content image,then calculates the gradient magnitude similarity map of the reference and distorted image;Finally,the integrated structural degradation measurement is used to obtain the final screen image quality index.The simulation results of screen content images(SCIs)subjective screen image quality database(SIQAD)show that the performance of the proposed method is significantly higher than other screen content image quality assessment methods.A method for quality assessment of blind screen content is presented.In view of the problem that the existing natural image quality assessment method cannot accurately evaluate the screen content image,this study proposes a blind screen content quality assessment method based on the theory of human perception.Extracting 22 eigenvectors of four characteristics of training image,such as free energy,structural degradation,log-energy and contrast,and using support vector machine(SVM)model to predict screen image quality.The simulation results of the SCIs subjective SIQAD and quality assessment of compressed screen content images(QACS)database show that the proposed metrics show good performance.A method for evaluating the content image quality of no-reference screen content based on structure information is presented.On the basis of analyzing the feature of local image sharpness,luminance and structure information,this paper presents a method of image quality assessment no-reference screen content based on structure information,aiming at the problem that the distortion of screen content image is difficult to capture effectively.Firstly,the statistical luminance feature is extracted from the luminance map by local normalization,then the texture feature is extracted by Prewitt filter,the image structure information is computed,and the image sharpness is measured by calculating the logarithmic energy(log-energy)of the discrete wavelet transform(DWT)coefficient.Finally,the support vector machine(SVM)is used to evaluate the quality of the screen content image.In order to verify the validity of the proposed method in this study,the comparative experiment is carried out in the subjective SIQAD database.The experimental results show that the proposed method has better performance and is superior to other existing evaluation methods.This study presents a method for evaluating the screen content quality of noreference based on support vector machine(SVM).In order to describe and predict the visual quality of screen content image,this paper analyzes the structure degradation feature,structural information and Gaussian distribution of screen content image,and proposes a method of no-reference image quality assessment based on support vector machine.Firstly,the characteristic of structure degradation feature,structural information and Gaussian distribution of screen content image are extracted,and then the SVM is used to train the mapping function from eigenvector to subjective mass fraction,and the quality evaluation score of screen content image is obtained.The effectiveness of the proposed method is verified by testing the subjective SIQAD database.The results show that the proposed method has good performance compared with other existing evaluation methods.The paper contains 27 figures,28 tables,and 146 references.