Structural similarity

Common card · No. 225569

In digital media and the study of algorithms, the structural similarity (SSIM) is a method for predicting the perceived quality of digital video and images, or for measuring the similarity between two images. The SSIM is a full reference metric, meaning its measurement of image quality is based on an initial, uncompressed image as reference. SSIM is a perception-based model that considers image degradation by evaluating perceived change in structural information, while also taking perceptual phenomena into account. The latter includes aspects of visual masking, both by luminance and contrast. This distinguishes it from other techniques such as mean squared error (MSE) or peak signal-to-noise ratio (PSNR) that instead estimate absolute errors. Structural information is the idea that the pixels have strong inter-dependencies especially when they are spatially close. These dependencies carry important information about the structure of the objects in the visual scene. Luminance masking is...

Read on Wikipedia