The SSIM was developed for the comparison of the quality of images subject to various compression algorithms [9], and measures how similar images are with respect to their structure within a
particular convolution window. In each convolution window and for each image(x,y), local means(μx,μy), standard deviations(σx,σy), and an auto- covariance (σxy) are calculated and compared. The SSIM is a local measure,which operates on a 2D scalar field,returning a 2D scalar
field that can be averaged to find the mean SSIM (MSSIM) for a particular image pair.Thes evalues are calculated by