利用双目箭载图像测量目标位姿方法研究

摘要

This paper carries research on position and attitude measurement method using rocket image, which is regarded as a new approach of physical parameters measurement on rockets. Firstly, STK software is used to simulate rocket's stage separation process and output binocular simulated images in rocket's view. Afterwards, SURF and MSER methods are used to detect, match and filtrate feature points in simulated images. The coordinates of feature points are solved through binocular vision model. The central point is chosen to describe position of the target, and three coplanar points are chosen to describe attitude of the target. The result indicates that the method measures position and attitude of the target using rocket images effectively, which will strongly support engineering application in the future.%研究箭载图像中的目标位姿测量方法,可作为箭上物理量测量的一条新途径.首先,使用STK软件仿真运载火箭级间分离过程,输出箭载视角双目仿真图像.然后,应用SURF/MSER方法对双目图像中的特征点进行检测、匹配、筛选,基于双目视觉测量模型解算特征点的空间坐标.最后,选取目标中心点描述被测目标位置,选取与分离面平行平面上的三点解算被测目标姿态.实验结果表明,方法可有效地测量箭载图像中目标的位置、姿态,为未来工程应用实现打下基础.