A Framework for Fine-Granular Computational-Complexity Scalable Motion Estimation
Zhi Xiang Yang, Hua Cai, Jiang Li · 2005
This paper presents a novel motion estimation (ME) framework that offers fine-granular computational-complexity scalability. In the proposed framework, the ME process is first partitioned into multiple search passes. A priority function is used to represent the distortion reduction efficiency of each pass. According to the predicted priority of each macroblock (MB), computational resources are then allocated effectively in a progressive way to achieve fine-granular computational-complexity scalability. Experiments show that our proposed scheme achieves progressively improved performance over a wide range of computational capabilities.