基于三维图像视觉特征的运动员训练错误动作图像自动识别方法

Automatic recognition method for athlete training error action images based on 3D image visual features

为了降低运动员训练动作的出错概率,提高训练水平,提出基于图像视觉特征的运动员训练错误动作自动识别方法。采用三维图像视觉特征点检测方法,提取运动员训练动作图像,通过约束点和末端点的运动学参数融合方法,分析运动员训练错误动作的视觉特征信息,利用几何空间辨识度分析和图像视觉参数特征重建实现对运动员训练动作的三维重构,在不规则三角网的空间融合机制下通过对图像视觉特征的三维结构重组和模糊度降噪处理,提取运动员训练错误动作的Harris角点信息和边缘轮廓特征信息,通过建立断层切片的轮廓树模型结果,根据图像视觉特征采集和信息融合结果,实现对运动员训练错误动作自动识别。仿真结果表明,采用本方法进行运动员训练错误动作自动识别准确率在95%以上,特征点重复率最低为0.012,特征辨识度较高,对错误动作的动态定位和纠正能力较强。

In order to reduce the error probability of athletes’training movements and improve their training level,an automatic recognition method of athletes’training errors based on visual features of images was proposed.The visual feature point detection method of 3D image is used to extract the athletes’training action images.The visual feature information of the athletes’training wrong action is analyzed by the kinematic parameter fusion method of constraint point and end point.The 3D reconstruction of the athletes’training action is realized by using geometric space identification analysis and image visual parameter feature reconstruction.Under the spatial fusion mechanism of Triangulation irregular network,Harris corner information and edge contour feature information of athletes’training errors are extracted through three-dimensional structure recombination of image visual features and blur denoising processing.Contour tree model results of sectional sections are established,and according to the results of image visual feature acquisition and information fusion,To realize the automatic recognition of athletes training wrong movements.The simulation results show that the automatic recognition accuracy of the wrong movements of athletes training using this method is more than 95%,the repetition rate of feature points is the lowest 0.012,the feature recognition is high,and the dynamic positioning and correcting ability of the wrong movements is strong.