Authors : Amine Samet , M.A.B. Ayed , Nouri Masmoudi and Lazhar Khriji
Abstract: The goal of this study is to determine a human visible difference between two images that may be used as an excellent image quality evaluator. This study describes a new perceptual image quality metric: the New eighted Mean Square Error (NwMSE) is easy to calculate and applicable to various image processing applications. It produces clear distortion maps and quantitative error measures informed by considerations of the Human Visual System (HVS). Many experiments on various image distortion types validate that NwMSE is compliant with subjective evaluations and performs significantly better than the widely used distortion metrics the Mean Square Error (MSE), the Structural SIMilarity (SSIM) index, the Image Quality Index (IQI) and the weighted Mean Square Error (wMSE). We show that the NwMSE correlates with MOS results and with human perception of image quality.
Amine Samet , M.A.B. Ayed , Nouri Masmoudi and Lazhar Khriji , 2005. New Perceptual Image Quality Assessment Metric . Asian Journal of Information Technology, 4: 996-1000.