Low-Complexity Video PSNR Measurement in Real-Time Communication Products

This paper presents a low-complexity video quality measurement framework for real-time communication (RTC) products. The framework provides a reliable PSNR measurement with low complexity, making it suitable for various devices ranging from low-end to high-end. The proposed measurement model decouples the distortion caused by scaling and compression in PSNR calculation, allowing the RTC client to compute PSNR within the video pipeline. Further optimizations are conducted to improve computation efficiency. Experimental results demonstrate the PSNR metric obtained from the proposed framework is closely aligned with the ground truth PSNR. The framework has been implemented on a mobile application and its reliability as a metric at scale has been validated.