基于Kinect的动态手臂三维姿势的识别与仿真  

Dynamic Arm Three Dimensional Posture Recognition Based on Device and Simulation

作　　者：吴誉兰[1] 章小宝[1] 

机构地区：[1]南昌大学科学技术学院,江西南昌330029

出　　处：《计算机仿真》2016年第7期369-372,共4页Computer Simulation

摘　　要：动态手臂中的三维姿势识别对智能化技术的发展起着推动作用。在动态手臂三维姿势的识别时,光照、舞动姿态、遮挡物导致动态的手臂动作被分割成一系列不规则的动作片段,无法从这些不规则片段中对动态手臂骨骼关节点的三维坐标进行整体定位,而定位的精度决定手臂的关节点间的距离和角度,导致传统方法对动态手臂中的三维姿势识别时,不能精确计算手臂的关节点间的距离和角度,存在动态手臂中的三维姿势识别误差大的问题。为解决上述问题,提出一种基于Kinect的动态手臂中的三维姿势识别方法。上述方法先利用Kinect算法获取手臂骨骼关节点的三维坐标,通过两点法设定动态手臂中的三维姿势关节点角度的约束条件,同时设定动态手臂中不同的三维姿势关节点角度阈值,并利用人体肤色模型获取双臂的比例长度与活动区域,提取动态手臂中各关节角度、姿势位置信息,利用隐马尔柯夫模型对动态手臂中的三维姿势进行训练识别。仿真结果表明,基于Kinect动态手臂中的三维姿势识别方法识别精确度高,能满足智能化技术的应用需求。In this paper, we proposed a recognition method for three-dimensional postures in dynamic ann based on Kinect. Firstly, we used Kinect algorithm to obtain the three-dimensional coordinate of articulation point. The constraint condition of coordinate was set up through two-point method. We also set the different angle threshold value of coordinate. Then we used skin color model to obtain ratio length and movement area of arms on that basis and extracted the position information of each joint angle and posture. Finally, we used the hidden Markov model to train and recognize three-dimension postures of dynamic arm. The simulation results show that the modified method has high recognition precision. It can fulfill application requirement of intelligent technology.

关 键 词：动态手臂 姿势识别 隐马尔柯夫模型 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]