Video Coding Parameter Optimization Based on Module Inter-relevancy

Algorithm customizable modules in video encoder play important roles in coding performance, and their algorithms can be optimized by adjusting multiple coding control parameters under a certain rate distortion complexity constraint. There are inter-correlated relationships among these modules. This paper analyzes the inter-module influence mechanism among algorithm customizable modules, and gives method to quantitatively measure the influence extent. According to the correlation degree among modules, the algorithm decision priority order is dynamically determined, and then coding parameter options are configured to achieve video coding parameter determination and optimization. X264 is used as the verification platform, and intensive experimental results are given to verify the proposed idea and model. The experimental results show that there is complicate inter-influence mechanism among modules, and module algorithm variation affects the performance of multi-objective performance, and this affection is not a simple linear relationship. The proposed parameter configuration and algorithm optimization method achieves considerable encoding complexity reduction, 80% computation saving, at the cost of slightly coding performance losses, no larger than 0.1 dB PSNR degradation.