基于量子粒子群优化算法的运动捕获数据关键帧提取  

Extraction of keyframe from motion capture data based on quantum-behaved particle swarm optimization

作　　者：杨涛[1] 孙怀江[1] 叶俊[1] 

机构地区：[1]南京理工大学计算机科学与工程学院,南京210094

出　　处：《计算机应用研究》2014年第8期2523-2527,共5页Application Research of Computers

摘　　要：关键帧提取是人体运动捕获数据分析与处理的重要研究内容,为此提出一种基于量子粒子群优化算法的运动捕获数据关键帧提取方法。量子粒子群优化算法具有较快的搜索能力,编码方式采用有序整数编码来保证搜索中运动序列的时序性。该方法既可以提取出确定数目的关键帧序列,也可以根据目标函数来提取关键帧序列。其中目标函数由重构误差和关键帧数目来定义,重构误差由原始运动与重构运动之间的平均帧间距离来度量。实验结果表明,该方法能够有效地从运动捕获数据中提取出具有最优重建误差的关键帧序列,并具有较好的视觉效果。Keyframe extraction is the important research content of human motion capture data analysis and processing. This paper presented a new extracting keyframes method to motion capture sequence base on quantum-behaved particle swarm optimization algorithm,quantum-behaved particle swarm optimization algorithm had a faster search capabilities,the encoding of this method was an orderly integer coding,it could ensure the timing of the motion sequence. This method could either achieve the optimum keyframe sequence in the case of determine the number of key-frames,and could also achieve the optimum key frame sequence in the case of the optimal objective function value,defined the objective function by the number of keyframes as well as the low reconstruction error with quaternion＇s spherical linear interpolation. It measured the reconstruction error by the average distance between frames,computed the distance between the original motion and the reconstruction one by the weighted differences of joint positions and velocities. Experimental results show that this method can effectively extract keyframes,reconstruction effect according to all other non-keyframes better than other methods and have better visual effect.

关 键 词：计算机动画 运动捕获数据 关键帧 量子粒子群 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]