摘要
在游泳运动中,对游泳运动员错误姿势进行准确识别与校正,可提高运动员平日训练质量。进行姿势识别校正时,游泳运动过程中容易发生人体的仿射形变,导致出现明暗度不高的动作特征点,而传统方法是对这些特征点进行提取,与正确的姿势进行比对实现姿势的识别与校正,导致不能实时检测与校正运动员出现的错误姿势。提出一种深度图像骨骼跟踪的游泳运动员姿势识别校正方法,采用阈值法对图像进行预处理,通过Kalman滤波器对采集的图像进行滤波处理,采用高斯分布函数从滤波后的图像中获取动作特征点,通过SURF方法筛除边际点和明暗度不高的动作特征点,采用欧氏距离法确定两个临近特征点间的距离,利用反馈监控原理对错误姿势进行识别校正。仿真结果表明,改进的深度图像骨骼跟踪的姿势识别校正方法,能实现运动运动员动作的追踪监控,完成对游泳运动姿势的检测和识别,准确度高,稳定性强。
A correction method of posture recognition for swimmers in depth image skeleton tracking is put forward by using threshold method to preprocess the image. The collected images are filtered through the Kalman filter. Gauss distribution function is used to obtain the feature points from the filtered image. The edge of the point and light and shade of the action feature points are screened out through the SURF method. The distance between two adjacent feature points is determined by using the Euclidean distance method, which uses the feedback monitoring principle to i- dentify and correct the error posture. The feedback control principle is used to identify and correct the wrong posture. The simulation results show that the improved depth image skeleton tracking method can realize the tracking control of the posture recognition of the athletes. It can complete the detection and recognition of the swimming movements. The accuracy is high, the stability is strong.
作者
刘桥
LIU Qiao(Institute of Physical Education of Changzhou University, Changzhou Jiangsu 213164, China)
出处
《计算机仿真》
北大核心
2017年第4期227-230,共4页
Computer Simulation
关键词
游泳比赛
姿势动作
骨骼跟踪
图像检测和识别
Swimming competition
Posture action
Skeletal tracking
Image detection and recognition