基于多尺度形态学图像的屈肘角度检测方法与实现

针对当前采用经验分析方法对曲肘的动作检测不精准的问题,提出一种基于多尺度形态学图像分析的屈肘角度检测方法.首先在多媒体视觉下进行武术动作的曲肘行为的图像采集和特征分析,然后对采集的图像采用多尺度Retinex角点筛选方法进行图像增强处理,提取图像的多尺度形态学边缘轮廓特征,实现对人体屈肘角度的特征点标定和检测.仿真结果表明,采用该方法进行武术动作中的曲肘角度检测优于经验分析方法,准确检测率高达80%以上,且稳定性好,具有较高的动作矫正指导价值.

Method and Realization of Elbow Flexion Angle Detection Based on Multi-scale Morphological Image

The paper presents a method of elbow flexion angle detection based on multi-scale morphological image to overcome the shortcoming of inaccurate motion detection,which is existing in empirical analysis methods.First of all,the image acquisition and feature analysis of elbow movement of Wushu movement are carried out in the multimedia perspective.Then,the multi-scale Retinex corner selection method is used to enhance the image.Multi-scale morphological edge features are extracted from image information to incriminate the characteristic points of elbow flexion angle.The simulation results show that the method is better than the empirical analysis method in the measurement of the elbow angle in the martial arts movement,the accurate detection rate is more than 80%,and possesses the advantages of high accuracy and stability with.It is of high clinical value in guiding.It has great guiding significance for corrective action.