多维数据关联的机器人运动步态识别方法

Gait Recognition Method of Robot Based on Multidimensional Data Association

针对机器人运动步态识别方法存在有效识别率较低等问题,提出一种基于多维数据关联的机器人运动步态识别方法。首先通过背景减除法进行图像提取,采用形态学方法去除图像中的噪声及空洞;然后使用多维数据关联提取机器人运动步态轮廓形状静态特征,将计算动作能量图(AEI)以及步态能量图(GEI)进行对比分析,获取GEI更多的动态能量信息;最后通过FEI进行Fan—Beam变换,采用核主成分分析进行空间降维,获取运动目标的频率动态特征,将两个特征进行融合实现机器人运动步态识别。实验结果表明,研究方法能够快速、准确实现机器人运动步态识别,实际应用效果好。

Aiming at the problem of low effective recognition rate in robot motion gait recognition methods,a robot motion gait recognition method based on multi-dimensional data association is proposed.Firstly,the image is extracted by background subtraction method,and the noise and holes in the image are removed by morphological methods;Then the static features of the robot motion gait contour shape are extracted using multi-dimensional data association,and the action energy map(AEI)and gait energy map(GEI)conduct comparative analysis to obtain more dynamic energy information of GEI;Finally,perform Fan-Beam transformation through FEI,and use nuclear principal component analysis to reduce the spatial dimension,obtain the frequency dynamic characteristics of the moving target,and fuse the two characteristics to realize the robot movement gait recognition.The experimental results show that the research method can quickly and accurately realize the robot movement gait recognition,and the practical application effect is good.effect is good.