Research On Quality Assessment Methods Of Enhanced Image

With the development of modern technology,people have higher requirements for image quality,and image enhancement technology can improve image quality to a certain extent.The quality of enhanced image directly affects human visual perception and the accuracy of obtaining information,so the research on its quality evaluation is of practical significance.The purpose of image quality evaluation is to design an objective evaluation algorithm which is consistent with the subjective evaluation score as much as possible.Based on this,this paper evaluates the quality of enhanced images with changing contrast and proposes two evaluation methods.(1)A full reference enhanced image quality evaluation algorithm is proposed.Firstly,a rotation-invariant filter mask is defined in eight directions by fractional differential,and the modulus of the image is calculated,then the final quality score is calculated by similarity deviation,and a large number of experiments are carried out by selecting contrast images in the standard database.The results show that the comprehensive evaluation performance of the algorithm is better than that of the state-of-the-art full reference evaluation algorithm,and the effectiveness of the algorithm is verified.(2)A no reference enhanced image quality evaluation algorithm is proposed.In this algorithm,two sets of features are extracted,namely the asymmetric generalized Gaussian distribution fitting of the normalized coefficient of mean subtraction contrast and the image feature extracted based on fractional differential.Then the image quality score is predicted by support vector regression training model.Finally,a large number of numerical experiments are carried out on TID2008,TID2013 and CSIQ database for the contrast change image and on CID2013 database.The results show that the proposed algorithm has good evaluation performance and certain competitiveness.