Improved Quantization and Transform Coefficient Coding for the Emerging Versatile Video Coding (VVC) Standard

One key component of all block-based hybrid video codecs is transform coding of prediction residues, which consists of an orthogonal block transform, scalar quantization of transform coefficients, and entropy coding of the resulting quantization indexes. For improving coding efficiency relative to the state-of-the-art video coding standard HEVC, we propose the following modifications: (1) Replacing scalar quantization with trellis-coded quantization; and (2) utilizing additional statistical dependencies between quantization indexes for entropy coding. The proposed approach was integrated into the first test model VTM-1 for the new standardization project Versatile Video Coding (VVC). Our coding experiments showed average bit-rate savings of 4.9 % for intra-only, 3.4 % for random access, and 2.8 % for low-delay configurations.