VVC Rate Control Based on Distortion Propagation Factors

In order to improve the rate-distortion (R-D) performance of rate control algorithms, this paper proposes an R-λ model rate control algorithm that incorporates temporally dependent rate-distortion optimization. Specifically, the Lagrange multiplier and quantization parameter (QP) for the frame to be encoded are first determined using frame-level rate control. These parameters are then applied to perform integer-pixel inter-frame prediction encoding on fixed-size blocks, obtaining motion-compensated prediction (MCP) errors and coding distortions. Based on these values, the distortion propagation factor (DPF) for 32×32 pixel blocks is calculated. Finally, coding tree unit (CTU)-level Lagrange multipliers and QPs are determined using the DPFs and the frame-level Lagrange multiplier. The proposed algorithm adjusts CTU-level optimization targets within the constraint of frame-level bit budget, thereby effectively reducing temporal distortion propagation. This not only ensures effective rate control but also significantly enhances the encoder's R-D performance. Experimental results demonstrate that, compared to the versatile video coding (VVC) benchmark encoder, the proposed algorithm achieves average R-D performance gains of 4.44% and 4.30% under low-delay B-slices and P-slices configurations, respectively, with average bitrate errors of 0.39% and 0.37%. These results underscore the effectiveness and potential of the proposed method for enhancing video coding performance.