篮球运动员大力扣球上肢动作轨迹预测仿真

摘要

A feature extraction method of upper limb movement in hard driven spike of basketball match is proposed based on Harris comer detection.Firstly,the attribute of space distribution of pixel grayscale in region of upper limb movements is mapped,and Gaussian mixed model standard is used to normalize athletes' contour of upper limb movements under hard driven spike.Then Harris comer detection method is used to carry out the enhancement in closed area of affine invariant for continuous action image of athletes.Finally,the comer detection for athlete upper limb movements edge is carried out,so as to complete the extraction of upper limb movement characteristics in hard driven spike.Simulation result shows that the proposed method has high precision of feature extraction and greatly improves the quality of basketball match.%对篮球大力扣球动作下上肢轨迹预测问题的研究,能够有效提高篮球比赛中运动员的安全性.对大力扣球动作运动员上肢动作轨迹的预测,需要对上肢动作特征进行提取,对上肢动作动态图像边缘轮廓进行角点检测,完成对上肢动作轨迹的分析.传统方法提取运动员上肢连续动作目标,归一化连续动作下上肢动作轮廓图像,但忽略了对上肢动作图像边缘轮廓的角点检测,完成在大力扣球下上肢动作轨迹的影响.提出基于Harris角点检测的篮球比赛大力扣球下上肢动作特征提取方法.该方法首先映射图像上肢动作区域中像素灰度级的空间分布的属性,采用高斯混合模型标准和归一化运动员大力扣球下上肢动作轮廓,利用Harris角点检测方法对运动员的连续动作图像进行仿射不变闭合区域增强处理,并对运动员上肢动作边缘轮廓进行角点检测,完成对篮球大力扣球下上肢动作特征提取.仿真证明,所提方法特征提取精度高,大幅度的提升了篮球比赛的质量.