IEEE Access

The Structural Similarity (SSIM) Index is a very widely used image/video\nquality model that continues to play an important role in the perceptual\nevaluation of compression algorithms, encoding recipes and numerous other\nimage/video processing algorithms. Several public implementations of the SSIM\nand Multiscale-SSIM (MS-SSIM) algorithms have been developed, which differ in\nefficiency and performance. This "bendable ruler" makes the process of quality\nassessment of encoding algorithms unreliable. To address this situation, we\nstudied and compared the functions and performances of popular and widely used\nimplementations of SSIM, and we also considered a variety of design choices.\nBased on our studies and experiments, we have arrived at a collection of\nrecommendations on how to use SSIM most effectively, including ways to reduce\nits computational burden.\n

A Hitchhiker’s Guide to Structural Similarity