Aligned-PSNR (APSNR) to improve accuracy of conventional PSNR for video stream over wireless and mobile network

Abstract

Peak Signal-to-Noise Ratio (PSNR) is one of widely used method to measure video quality. This method is inherent with inaccuracy issue while applied to measure video quality that streamed over wireless and mobile network. Some methods have been proposed in order to tackle the issue. Modified-PSNR (MPSNR) is one of the proposed methods that show better performance than conventional PSNR. However, MPSNR is inappropriate if there are more than 5 consecutive frame losses due to its fixed window concept. This paper proposes a concept of dynamic window size, called Aligned-PSNR (APSNR). Experiment has been conducted to evaluate performance of the APSNR toward the conventional PSNR. Results from these two methods are benchmarked with Subjective Video Quality Assessment (SVQA) by utilizing Pearson product-Moment Correlation Coefficient (PMCC). Based on PMCC calculation, APSNR shows better performance (0.92) than conventional PSNR (0.89) in representing SVQA result