Detail Preserving 3D Motion Compression Based on Local Transformation
Chang Liu, Zhaoqiang Lai, Jiaxi Hu, Jing Hua · 2010
With the advance of 3D acquisition techniques, more and more 3D motion data with fine details become available. In order to efficiently represent the data, we explored different motion compensation algorithms through conventional techniques and analyzed their performance both in theory and by experiments. A novel motion compression framework based on local transformation is proposed to solve the existing problems. Experiments show that our algorithm can achieve very high compression rate while preserving the fine details. The algorithm is easy to implement and is versatile to different applications including non-skeleton driven motion compression.