摘要
在对运动易损动作进行识别时,一直由于无法克服外界条件带来的影响,存在识别精度低、误差大的问题。提出基于三维视觉的运动易损动作识别方法,对运动动作的三维视觉空间进行三维梯度值求解,获取三维视觉梯度直方图特征;并进行降维出来,得到低维隐空间下的运动模型,在此基础上,对其进行序列映射,求出测试序列与训练序列的吻合度。通过匹配计算,获取匹配度,实现运动易损动作的识别。实验结果表明,所提方法具有很高的识别精度,所需时间较短,识别性能高。
When identifying vulnerable action of sports,the effects of external conditions have been unable to overcome,such as low recognizing precision,error bigger problem. Vulnerability is proposed based on the three dimensional visual motion gesture recognition methods,the movements of the three-dimensional visual space,for solving three dimensional gradient values,obtain three-dimensional visual gradient histogram feature,and dimension reduction,get low dimensional space motion model of Cain,on this basis,the sequence map,the inosculation of testing and training sequences,by matching calculation,compatibility,were achieved the identification of vulnerable movements. The experimental results show that the proposed method has high identification accuracy,shorter the time required,high recognition performance.
出处
《科学技术与工程》
北大核心
2016年第17期235-239,共5页
Science Technology and Engineering
基金
海南省教育科学规划课题(QJY1251514)
海南省高等学校科学研究项目(HNKY2016-11)资助
关键词
三维视觉
易损动作
识别
3 D vision
vulnerability actions
identification
