부산대학교 도서관

Due to various reasons an image may get distorted. To measure the perceptual distortions, image quality assessment (IQA) metrics are required. Peak-signal-to-noise-ratio (PSNR) is one of the popular available IQA metric. It is usually measured by computing distortion in terms of mean-squared-error (MSE) between the reference image and its distorted version. Mean-squared-error (MSE) is measured over the entire image by giving equal weight to each corresponding pixel difference of the reference and distorted image. However, as per investigation done in this paper, it is found that instead of calculating MSE/PSNR on overall image, PSNR measured over the region of maximum distortion only is a better choice. The PSNR computed over the most distorted region (PSNR-MDR) reflects the image quality better than it computed over entire image. The PSNR-MDR is not only superior to the PSNR but also competitive to celebrated metric, SSIM (Structural Similarity Index).