基于SURF算法的微小位移测量

Mini-displacement measurement based on SURF algorithm

针对导弹等高加速度运动目标发射初期微小初始位移检测的需要,提出采用SURF算法将位移变化转化为提取的特征点对欧式距离的变化的方法。该算法在SURF提取图像特征点的基础上,使用最邻近匹配进行快速的初匹配,然后采用提出的一种基于平行距离的匹配方法进一步去除错误的匹配点对,根据正确的匹配点对的距离计算出初始位移。实验结果表明,该算法精度在亚像素级别,运行时间在10 ms之内,能够实时可靠的给出目标的初始位移。

A mini-displacement measure method based on SURF algorithm is proposed for the initial state of high ac- celeration target such as missile. The algorithm converts the displacement to the coordinate change of the matching points. The algorithm detects the interested points by SURF, rapid-index of the interested points is established, and the nearest neighborhood method is used during the process of initial matching. A new matching method based on parallel distances is proposed to eliminate the mismatched points. The initial mini-displacement is calculated by the matched points. The results reveal that the algorithm can achieve sub-pixel accuracy, and the running time is within 10 ms. All these ensure the algorithm＇s reliability and real-time performance.