摘要
针对传统动作识别方法一直存在提取成功率较低、提取时间较长等问题,提出了一种基于图像识别的自动跟踪方法对高强度运动下的人体动作进行识别。首先,应用双重卷积理论对高强度下人体动作的原形图像进行阈值分割,对人体动作进行特征提取。然后,结合高斯分布模型对获得的人体动作图像目标、背景和前景信息进行处理,得到人体动作图像背景的高斯分布模型,并采用卡尔曼滤波获取人体动作图像的跟踪轨迹。最后,应用贝叶斯分类理论,对人体动作图像的灰度信息构建目标模型,求解出人体动作图像的最优峰值点,实现多个目标的分割与跟踪。实验结果表明,通过图像识别的自动跟踪方法对人体动作特征提取具有良好的精确,且提取速度显著提高。
Aiming at the problems of low extraction success rate and long extraction time in traditional motion recognition methods,an automatic tracking method based on image recognition was proposed to recognize human motion in high intensity motion.Firstly,double convolution theory was applied to threshold segmentation of the original image of human motion under high intensity,and feature extraction of human motion was carried out.Then,combined with the Gauss distribution model,the target,background and foreground information of human action image were processed,the Gauss distribution model of human action image background was obtained,and the tracking trajectory of human action image was obtained by Kalman filter.Finally,based on Bayesian classification theory,an object model was constructed for gray level information of human action image,and the optimal peak point of human action image was solved to realize segmentation and tracking of multiple targets.The experimental results show that the automatic tracking method of image recognition is accurate and the extraction speed is improved significantly.
作者
张辉
Zhang Hui(Xinlian College,Henan Normal University,Zhengzhou Henan 450000,China)
出处
《计算机仿真》
北大核心
2019年第9期469-472,共4页
Computer Simulation
关键词
图像识别
自动跟踪
高斯分布
卡尔曼滤波
Image recognition
Automatic tracking
Gaussian distribution
Kalman filtering
作者简介
张辉(1981-),男(汉族),河南洛阳人,硕士研究生,讲师,主要研究领域为体育教育训练学。
