SSIM Motivated Quality Control for Versatile Video Coding

Meng Wang, Shiqi Wang, Junru Li, Li Zhang, Yue Wang, Siwei Ma

Abstract: In this paper, we investigate the encoder control algorithm driven by the visual quality in Versatile Video Coding (VVC) standard. The proposed framework shifts the traditional bit rate-centered paradigm to the quality-centered paradigm, and adopts the repeatedly proven perceptual quality measure structural similarity index (SSIM) as the core in guiding the behaviors of encoder. In particular, a reduced-reference evaluation metric which could statistically establish the relationship between the SSIM and quantization parameter, is adopted for quality control. Extensive experiments verify the effectiveness of the proposed method, leading to precise quality control in the VVC Test Model (VTM-9.0).