Research On Image Quality Assessment Algorithm Based On Structural Similarity

With the rapid development of multimedia technology and network technology,the Technology of digital image and digital video has been widely used in industrial production,scientific research,medical and health,education,entertainment and so on.However,Image is subject to distortions during acquisition,processing,coding,storage and transmission,which is caused by acquisition system,processing algorithm,transmission equipment and so on.Therefore,evaluating image quality accurately becomes a meaningful research topic in the field of image processing.On the basis of previous study,in this thesis,full reference image quality assessment algorithm is studied.1)An improved image quality assessment algorithm based on structural similarity is proposedAs Structural Similarity(SSIM)image quality algorithm cannot evaluate severely burred image accurately enough and does not correlate well with human subjective perception,this thesis improves SSIM algorithm from two aspects,the main works are as follows:(1)As SSIM algorithm cannot reflect the changes of edge information of an image efficiently,an image quality assessment algorithm called Gradient-based Structural Similarity(GSSIM),which adds gradient information into the SSIM algorithm,is proposed.The new algorithm,compared with SSIM algorithm,includes not only the comparison function of luminance,contrast,and structure,but more importantly,this algorithm increases the comparison function of gradient.Experimental results on LIVE2 image database show that the performance of GSSIM algorithm is superior to SSIM algorithm.(2)Many studies show that human eyes have different sensitivities for different distortion regions of an image.Visual saliency map can reflect the visual importance of different regions of an image efficiently.The paper uses visual saliency map as a weighting function to optimize the GSSIM algorithm,and finally obtains an improved image quality assessment algorithm based on structural similarity.Experimental results on LIVE2 image database show that the improved algorithm has a good consistency with the subjective assessment of human beings and it owns much lower computational complexity compared with other algorithms.2)An image quality assessment algorithm based on color image is proposedCurrently most of the image quality evaluation algorithms are only designed for gray images(or the luminance components of color images),which ignore the color information of image.Although human eyes are more sensitive to the changes of the gray information than the color information,if image quality evaluation algorithms ignore the color information will also affect the image quality evaluation results.Based on the SSIM algorithm,considering the impact of the color information on the perceptual image quality,the paper gives a color image quality assessment algorithm which based on HSI model.To validate the effectiveness of the improved algorithm,experiments are performed on LIVE2 image database.Experimental results show that the join of color information could improve the accuracy of objective quality assessment model.