基于Gocator的多传感器数据拼接方法研究

Research on Multi-sensor Data Registration Method Based on Gocator

点云数据拼接在众多科研领域有着十分广泛的应用;为完整、精确地得到复杂物体的点云数据,提出一种基于Gocator的多传感器数据拼接方法;该方法需要对多传感器系统进行两两校准以获取各传感器坐标系与基准坐标系之间的空间变换关系,进而将各传感器自身坐标系下的数据转换到基准坐标系下,实现多传感器数据的拼接;对于双传感器数据拼接,首先通过两只传感器同时拍摄单孔标定块,利用最大距离法提取标定块轮廓坡口特征点,根据坐标转换原理,初步确定了两传感器间的旋转平移关系;在此基础上采用迭代最近点(ICP)算法进一步优化确定两传感器之间的最优变换矩阵,以得到精确的拼接关系;实验室搭建双传感器钢轨廓形检测平台对该算法进行验证,实验结果表明,多次拼接得到的钢轨廓形与标准模板误差不超过0.2mm,完全符合钢轨廓形允许误差要求,该算法具有较高精度和稳定性。

Point-clouds registration has a very wide range of applications in many areas of scientific research.In order to get the complex object with more complete and precise point-clouds data,a new multi-sensor data registration method is proposed based on the Gocator.According to the method,the data in each sensor coordinate system can be converted into the base coordinate system to realize multisensor data registration.To complete the data registration of two sensors,firstly,is to take the contour of a single hole respectively by two sensors and to use the maximum distance method to extract the contour points.Then it will determine the rotation and the translation relationship between the two sensors according to the principle of coordinate conversion.On this basis,by using the iterative closest point method(ICP)algorithm,the optimal transformation matrix between the two sensors will be further optimized so that the result of the data registration will be more accurate.The algorithm is validated through building a dual sensor rail profile detection platform,and the experimental results indicate that the error of rail profile and standard template does not exceed 0.2 mm,which is in accordance with the allowable error of rail profile.The algorithm has the advantages of high precision and stability.