Quality Assessment Of High Dynamic Range Image And Video

The development of high dynamic range(HDR)imaging technology has changed traditional way the images are displayed,bringing people a more realistic visual experience.The technology has been widely used in many fields such as film,medical,home entertainment,etc.In the process of HDR images from acquisition to display,coding distortion is common,but the objective quality assessment methods is scarce.This paper studies the effects of human perception on the quality of HDR images and video with coding distortion.The main research contents are as follows:(1)A no-reference HDR image quality assessment method based on tensor domain perceptual features is proposed.Firstly,tensor decomposition is used to obtain the tensor sub-bands with luminance distortion and chrominance distortion.Then,the Auto-Regressive(AR)model is adopted to simulate the process that human brain adopts to predict the tensor sub-bands and obtain the perceptual prediction image of the tensor sub-bands.Finally,the AR coefficients are employed to represent perceptual predictive characteristics of HDR image in tensor domain,and the quality score of the HDR image is obtained via support vector regression(SVR)model,combining with the dynamic range and proportion of brighter areas of the tensor sub-bands and the perceptual prediction image.Experimental results tested on two public HDR image databases of Nantes and EFPL show that the proposed method can achieve high consistent alignment with subjective assessment.The performance indices of SROCC and PLCC are all above 0.93,RMSE are 0.340 7 and 0.377 1,respectively.(2)An HDR video database with HEVC coding distortion is constructed.Firstly,encoding frameworks of non-backward compatible and backward compatible are used to support HDR and LDR users simultaneously.Then,four levels of quantization parameters(QP)are used to simulate the video transmission under different bit rate requirements.Finally,subjective scoring of the encoded video is obtained using degradation category rating(DCR)method.Subjective experimental results show that subjective scoring has a good monotonous with QP levels.The results of different objective quality evaluation methods tested on the HDR video database proposed show that the HDR-VDP-2.2 and PU-VIF have the highest correlation with subjective scores,and performance indices of SROCC are 0.838 0 and 0.847 5 respectively.(3)A blind-reference HDR video quality assessment method based on the texture information of different region and sparse features is proposed.Firstly,according to the characteristics that the sensitivity of human eyes to the change of texture under different backgrounds are different,the video is divided into dark area,normal brightness area and bright area,and the Sobel operator is used to extract the texture information of each area of the video.Then,the sparse dictionary learning method is used to extract the sparse features of the video,including training and testing phases.In the training phase,pre-processing undistorted video(including local normalization,cubing and classification),multi-linear principal component analysis(MPCA)method is used to extract the principal components of video blocks of each class,and vectorizes the principal components of the video blocks for sparse dictionary learning.In the test phase,the distortion video is preprocessed and the distorted principal component of the video cubes are extracted.The sparse features of distorted video cubes are obtained by using the sparse representation method to represent the vectorized distorted principal component with the help of the sparse dictionary obtained in the training phase.Then the sparse features of all the cubes are fused to obtain the sparse features of the distorted video.Finally,the sparse features of each distorted video cube is obtained.Lastly,the quality score of the distorted HDR video is obtained via SVR model,combining with the corresponeding regional texture information and sparse features.The test results on the HDR video database proposed show that the proposed method can achieve good consistent alignment with subjective assessment.The performance indices of SROCC,PLCC and RMSE are 0.827 3,0.853 7 and 0.589 2,respectively.