计算资源受限的视频编码多模式决策

由于视频编码技术趋向于采用越来越复杂的分块模式,多模式决策技术也随之成为一种非常重要的编码技术.多模式决策的优劣不仅会大幅度地影响视频编码的计算消耗,而且也对编码性能的高低起到关键的作用.为使多模式决策在计算能力相差悬殊的平台上都能获得优化的率失真性能,给出一种计算复杂度自适应的优化多模式决策算法.首先,利用视频序列中不同宏块模式间的时空相关性,预测这些宏块多模式决策后的拉格朗日代价和计算复杂度的斜率(Lagrangian cost and complexity slope,简称J-C slope).J-C slope越大,说明在该宏块上的模式决策消耗每单位的计算资源可以获取的率失真收益越多.在计算资源有限的情况下,多模式决策应该按照J-C slope的大小顺序执行,也就是性价比优先的顺序,以便保证计算资源优先分配给率失真收益大的宏块.另外,还通过建立J-C slope阈值与实际计算复杂度的关系模型,设计了一种根据给定计算约束自适应调整计算复杂度的算法.根据实验结果,该方法不仅可以准确地控制多模式决策的计算复杂度,而且还能在不同的计算约束下获得优化的率失真性能.

Multi-Mode Decision Under Computational Resource Constraints for Video Coding

As more and more flexible block modes are introduced into video coding, mode decision technology becomes an important coding tool. The performance of mode decision has great impact on both coding performance and computational complexity. This article proposes a complexity controllable multi-mode decision algorithm to attain optimized coding performance under different computational complexity constraints herein. Instead of speeding up multi-mode decision merely, the algorithm predicts the Lagrangian cost and complexity slope (J-C slope) of MD for each macroblock (MB) by exploiting their temporal and spatial correlations. The larger J-C slope is, the higher coding gain over each unit of computational cost will be. In the environment of limited computational resources, multi-mode decision should be performed in a cost effective order, i.e. the order of their J-C slopes, to achieve optimized coding performance. In addition, an adaptive method is proposed to adjust the computational complexity of the algorithm dynamically by discovering the relationship of J-C slope thresholds and their corresponding complexity. Experiments demonstrate the proposed algorithm can both precisely adjust the computational complexity and optimally perform mode decision under different computational constraints.