三维视觉图像跟踪运动员对受伤目标监测仿真

摘要

A monitoring method of three-dimensional (3D) visual image for athletes' injured action is proposed based on support vector machine.Firstly,the target image block of body motion with given size of module integrated with median filtering theory is de-noised.The second order statistic magnitude of gray co-occurrence matrix is used to predict attribute feature of 3D athletes' injured image.Then,the optimal classification decision function of injured action feature is obtained,and the discriminant model is built.Finally,the athletes' injured action integrated with distance monitoring function is monitored.Simulation results show that the method can improve monitoring precision effectively.%对于三维视觉图像跟踪运动员对受伤动作目标的监测,可有效的保障竞技体育比赛的公平性.对受伤动作的监测需要提取受伤动作图像特征点,结合距离监测公式监测是否为运动员受伤动作.采用当前方法组建运动员受伤三维视觉图像模型时,难以提取运动员受伤动作图像特征点,存在运动员受伤动作监测误差大的问题.提出基于支持向量机的三维视觉图像对运动员受伤动作的监测方法.结合中值滤波原理对给定模块大小的人体运动目标图像块进行降噪处理,利用灰色共生矩阵的二阶统计量预测三维运动员受伤动作图像属性特征,得到运动员受伤动作特征最优分类决策函数,组建三维运动员受伤动作特征判定模型,结合距离监测公式监测是否为运动员受伤动作.实验结果表明,所提方法能够有效提升三维视觉图像对运动员受伤动作的监测精度.