科技情报开发与经济
科技情報開髮與經濟
과기정보개발여경제
SCI/TECH INFORMATION DEVELOPMENT & ECONOMY
2015年
6期
144-145,153
,共3页
RJMCMC算法%人体运动目标跟踪%数据矩阵%粒子滤波
RJMCMC算法%人體運動目標跟蹤%數據矩陣%粒子濾波
RJMCMC산법%인체운동목표근종%수거구진%입자려파
RJMCMC algorithm%human movement target tracking%data matrix%particle filter
提出了一种基于粒子RJMCMC算法的人体运动跟踪仿真算法，首先利用Simulink仿真将人体运动过程中各个关节点逐一收集起来，建立运动特征值矩阵，然后利用粒子RJMCMC算法来实现对矩阵每个关节点的跟踪，通过将RJMCMC算法与粒子滤波算法中重采样过程的融合，更加合理地对权值分布进行优化，从而降低权值的锐度，提高粒子样本的多样性。
提齣瞭一種基于粒子RJMCMC算法的人體運動跟蹤倣真算法，首先利用Simulink倣真將人體運動過程中各箇關節點逐一收集起來，建立運動特徵值矩陣，然後利用粒子RJMCMC算法來實現對矩陣每箇關節點的跟蹤，通過將RJMCMC算法與粒子濾波算法中重採樣過程的融閤，更加閤理地對權值分佈進行優化，從而降低權值的銳度，提高粒子樣本的多樣性。
제출료일충기우입자RJMCMC산법적인체운동근종방진산법，수선이용Simulink방진장인체운동과정중각개관절점축일수집기래，건립운동특정치구진，연후이용입자RJMCMC산법래실현대구진매개관절점적근종，통과장RJMCMC산법여입자려파산법중중채양과정적융합，경가합리지대권치분포진행우화，종이강저권치적예도，제고입자양본적다양성。
This paper puts forward a human movement tracking simulation algorithm based on particle RJMCMC algorithm, which will collect all the key points in the process of human motion one by one to establish a movement characteristic value matrix by using simulink analysis at first, and then implement the tracking for all key points in the matrix by using the particle RJMCMC algorithm, and optimize the weight distribution more reasonably through mixing together the resampling processes in RJMCMC algorithm and particle filter algorithm to reduce the weight value’s sharpness and improve the diversity of particle samples.
