Research On Synthesized Image Quality Assessment Algorithm Based On Human Visual Perception

With the rapid development of science and technology in recent years,more and more new technologies were used to the real life.From grayscale images to color images,from 2D image to 3D image,the whole process shows nowadays with the rapid development of multimedia technology.What is worth mentioning in multimedia technology is 3D technology,such as Virtual Reality,3DTV and Free-viewpoint TV,etc.These can better meet the requirements of human visual feeling and simulate the visual perception of human beings.But these technologies need people to get more images,the acquisition of image takes a lot of manpower,material resources and financial resources.Therefore,based on the depth of the image rendering technology emerges,the main objective of this technique is to reduce the amount of image acquisition,storage and transportation.However,it is inevitable that the quality of images will be reduced in the process of image acquisition,image storage,image display and rendering technology,which will destroy the visual experience of human beings.Therefore,it is particularly important to evaluate the quality of degraded images.Because the production and development of images are due to the needs of people,people are the "perception" of the image display.Therefore,if human beings directly score the image quality,the score is very accurate,this method is called subjective image quality assessment.However,this method costs human resources and financial resources,so people want to be able to score the image directly through the computer.This method is called objective image quality assessment,and it does not require human participation in the whole process.This paper introduces a no reference synthesized image quality assessment algorithm and a full reference synthesized image quality assessment algorithm.Because people are the " perception" of the image display,the two image quality assessment methods in this paper are designed based on the characteristics of human visual perception,by extracting the visual perception characteristics of images,the simple and effective objective image quality assessment algorithm is designed.The specific research content of the paper includes the following two aspects:(1)In the design of the no reference synthesized image quality assessment,the color information and hole information of the synthesized image are considered.Since human is sensitive to chromatic information,the chromatic information is extracted which is represented by the features of saturation and hue.Specially,we calculate the first derivative of saturation and hue maps by using local binary pattern(LBP)algorithm and extract features from LBP maps.In addition,inspired by the characteristic of the human visual system(HVS)and the synthesized image-specific distortion type,the proposed method extracts hole maps as weighting maps for LBP maps.Finally,the support vector regression(SVR)model is used to train all the extracted feature vectors,fit a regression function,and then predict the visual quality score of images.Compared with 8 state-of-the-art no-reference methods for natural and synthesized images,the proposed method shows improved performance on IRCCy N/IVC DIBR and MCL-3D databases.(2)Due to the rapid development of free-viewpoint television(FVT),Depth-Image-Based Rendering(DIBR)technology has been widely used to synthesized images of virtual view-points.However,the types of distortion in the synthesized images are different from those of natural images,such as discontinuity,flickering,stretching,etc.To measure the distortions occured in the synthesized images,we propose a full-reference(FR)quality assessment method by local variation measurement consisting of three-modules.Firstly,since the distortion in the synthesized image mainly occurs in the structural region of the image,the neutrosophic(NS)domain is employed to extract the structure map of image and evaluate the quality score of local image structure.Secondly,by considering that the texture of the synthesized image might be damaged due to the warping of 2D image or the loss of information in the hole region,we evaluate the quality score of local texture by using the features obtained from frequency domain.Thirdly,to measure the stretching distortion which is unique in the synthesized image,the visual quality of extracted stretching area is measured based on the entropy.Finally,a pooling operation is used to combine the quality scores of the three modules to obtain the final predicted quality score.Experimental results show that the performance of the proposed algorithm is better than start-of-the-art full-reference and no-reference image quality assessment metrics.