Ssim is called Structural Similarity Index, which is a structural similarity, which is a measure of two image similarity. The indicator was first presented by the University of Texas at Austin's image and video Engineering Laboratory (Laboratory for image and videos Engineering). If two images are pre-compressed and compressed, then the SSIM algorithm can be used to evaluate the compressed image quality.
Ssim How to characterize similarity:
First, give a set of formulas:
UX, Uy respectively represents the image x and y mean, σx due, σy respectively represents the standard deviation of the image x and Y, σx*σx, σy*σy (really can not play superscript ah, understand Viva) respectively represents the image x and y variance. Σxy represents the image x and y covariance. C1,C2 and C3 are constants that maintain stability in order to avoid a denominator of 0. Usually take c1= (k1*l) ^2, c2= (k2*l) ^2, C3=C2/2, k1=0.01 in general, k2=0.03, l=255 (is the dynamic range of pixel values, generally taken as 255)
The final Ssim index is:
When we set C3=C2/2, we can rewrite the formula into a simpler form:
So the structure similarity index defines the structure information from the angle of the image as the attribute of the object structure independent of luminance and contrast, and the distortion is modeled as a combination of brightness, contrast and structure of three different factors. Using mean as the estimation of luminance, standard deviation as the estimate of contrast, covariance as the measure of similarity of structure.
How to obtain the mean, variance and standard deviation:
In practice, the Gaussian function is generally used to calculate the mean, variance and covariance of the image, rather than the way of traversing pixels, in order to get higher efficiency.
The following link we will use a simple program to implement the SSIM algorithm, and make a comparative experiment:
http://blog.csdn.net/chaipp0607/article/details/70160307