“…If an objective IQA metric can simultaneously offer higher values of CC, SROCC and KROCC, and lower values of RMSE and OR, it is deemed that this metric achieves better predictive performance. In our experiments, the predictive ability of the proposed MMVD metric is compared with eleven representative FR-IQA metrics, including two extensively used traditional metrics, i.e., PSNR and MultiScale Structural SIMilarity (MSSSIM) [61], four state-of-the-art FR-IQA metrics, i.e., VIF [5], VSNR [19], FSIM [11], NQM [20], and the latest RVSIM [12], EFS [27], GDRW [2], SLY [3], SCQI [62], which are published in 2018, 2018, 2017, 2015, 2016, respectively. Matlab source codes of these methods are downloaded from websites provided in corresponding literatures, or are derived from their authors.…”