Research On Image Quality Assessment Based On Symmetry Phase Congruency

Images can transmit various kinds of information.There exist more and higher demands for the important carrier of information with booming development of the Internet era.More excute and efficient evaluation systems and methods are in need for the fact that there may be different degrees of distortion in the process of transmission or compression.More accurate and efficient evaluation systems and mathods are in demand to evaluate the image quality.The image quality assessment algorithm mainly includes two methods,subjective evaluation and objective evaluation,and this paper quantitatively analyses similarity of images through objective method of mathematical modeling.The work done in the thsis is as follows:(1)The image quality assessment algorithm based on feature similarity(FSIM)has certain limitations to some extent when detecting image step edges,while symmetry phase congruency(SPC)can better restore these edge features.A new algorithm named symmetry fearure similarity(SFSIM)is put forward by combining the FSIM with the symmetry phase congruency which is sensitive to step edges,and calculating the gradient magnitude through the isotropic Sobel operator.The proposed algorithm makes use of symbols responding to the phase adjacent pixels by SPC and the more accurate weighing coefficient of isotropic Sobel operator to detect and locate the image edge feature.The simulation experiment on various kinds of distorted images are carried out in international standard image database,and the results indicate that the SFSIM is more sensitive to sharper image edge features and Gaussian blurred images on the basis of retaining the excellent performance of the original method and the evaluation results get to be better as well.(2)Medical image processing,whose message body is image,combines computer technology with sorts of medical imaging equipments.Image conveys main information of medical image processing system,and it is image quality that indicates the system performance.Magnetic resonance(MRI)images are able to clearly show the differencebetween normal tissue cells and tumor cells.The brain MRI images are segmented in accordance with different pixel blocks through super pixel division.The human brain is symmetrical in structure.Numerical values are obtained by comparing the similarity of pixel blocks in the left brain and the right symmetrically,and then the location of cancer in MRI images will be located by analyzing the evaluation results.This can be used to detect the location of cancer,which improves the accuracy when judging.