Frame Complexity Guided Lagrange Multiplier Selection for H.264 Intra-Frame Coding

Rate-distortion (R-D) optimized mode decision plays an essential role in H.264/AVC encoding. Among all the possible coding modes, it aims to select the one which has the best trade-off between bitrate and compression distortion. Specifically, this tradeoff is tuned through the choice of the Lagrange multiplier. However, in the H.264 reference software, the value of Lagrange multiplier is only determined by the quantization stepsize, with no consideration of the characteristics of the input signal. In this paper, we address the problem of optimal Lagrange determination for intra-frames in a signal dependent environment, and a novel frame complexity guided Lagrange multiplier selection method is introduced. Our contributions are two-fold. First, we propose a novel signal dependent R-D model which uses the average input frame gradient to measure the signal complexity. Second, based on the proposed signal dependent R-D model, we derive the closed-form solution of the optimal Lagrange multiplier. Experimental results suggest that up to 8.18% bitrate reduction can be obtained with negligible additional computation complexity.