基于双目立体视觉的乘员运动姿态测量方法研究

Research on test method of occupant motion attitude measurement based on binocular stereo vision

提出了基于双目立体视觉的汽车乘员运动姿态测量方法。首先使用自主搭建的双目立体视觉系统,按照张正友标定法标定相机并对图像进行立体校正;然后提出基于邻域的SGBM单参数优化算法,并依此建立乘员关键特征点与三维空间坐标间的映射关系,实现了对乘员运动姿态的非接触动态测量。实验结果表明,该方法可以有效地测量乘员运动姿态,且测量精度较高,对汽车乘员非标准坐姿下的保护策略制定具有重要的参考价值。

This paper presents a method based on binocular stereo vision to solve the dynamic measurement of occupant motion attitude in collision experiment.Using the binocular stereo vision system,the camera is calibrated according to Zhang Zhengyou calibration method and then conduct the stereo rectification.A neighborhood-based SGBM single parameter optimization algorithm is proposed to establish the mapping relationship between the occupant feature points and the three-dimensional space coordinates,and obtain the three-dimensional space coordinates of the occupant attitude.Experiments show that this method can effectively measure the occupant attitude position,and the measurement accuracy is high.It is of great significance to the development of the protection strategy for vehicle safety,especially for the non regulatory standard sitting posture of the occupant.