激光扫描视觉提高DGPS/IMU定位定姿可靠性方法研究

The position and attitude parameters from DGPS and IMU integration processing is very important for photogrammetry processing without ground control points. The accuracy,stability and reliability of DGPS/IMU integration result is critical to land based mobile mapping system for large scope 3D spatial data production and application. The GPS signal block or outage and the IMU drift error affect the direct georeference accuracy of land based mobile mapping system and it will cause some defects for during the practice engineering application with land based mobile mapping system. In this project, a laser scanning and imaging vision approach is presented for the DGPS/IMU integration processing as following aspects: .1) Research on laser scanning point cloud data aided feature extraction algorithm for binocular sequence CCD images;.2) Algorithm research of binocular sequence images motion analysis based on matched image feature; .3) Position and attitude determine method study with laser scanning vision analysis; 4) Extented Kalman filtering algorithm integrated Laser scanning vision, imaging vision and DGPS/IMU for position and attitude determine..With the above key technique breakthrough, the proposed approach achieve high accuracy DGPS and IMU integration processing with laser scanning and imaging vision to solve the problem of stability and reliability position and attitude determine, which can be used for land based mobile mapping system during the GPS signal block or outage and IMU drift. It can provide methodology and technique for large scale mapping without control point with land based mobile mapping system.

DGPS/IMU 集成处理对于无地面控制摄影测量至关重要，其定位定姿精度、稳定性和可靠性影响着车载移动测图系统进行大规模三维空间数据生产和应用。当前，GPS 遮挡或失锁、IMU漂移均使得DGPS/IMU 直接地理参考精度在实际工程应用、测量连续性方面存在欠缺。本项目将激光扫描视觉引入DGPS/IMU 集成处理，研究1）激光扫描数据辅助CCD 影像特征提取算法；2）基于序列双目影像运动分析的位置和姿态推算方法3）基于三维激光扫描视觉的相对测量位置姿态参数估计算法；4）集成影像和点云视觉运动分析载体位置姿态估计参数辅助DGPS/IMU 的扩展卡尔曼滤波定位定姿处理算法。通过对上述关键技术突破，为DGPS/IMU集成引入CCD 序列影像、三维点云视觉辅助定位定姿，解决GPS 遮挡失锁、IMU 漂移时连续可靠和高精度定位定姿问题，为无控制移动测图系统直接地理参考及大范围测图应用提供理论和技术支持。

DGPS/IMU 集成处理对于无地面控制摄影测量至关重要，其定位定姿精度、稳定性和可靠性影响着车载移动测图系统进行大规模三维空间数据生产和应用。当前，GPS 遮挡或失锁、IMU 漂移均使得DGPS/IMU 直接地理参考精度在实际工程应用、测量连续性方面存在欠缺。本项目将激光扫描视觉引入DGPS/IMU 集成处理，研究1）激光扫描数据辅助CCD 影像特征提取算法；2）基于序列双目影像运动分析的位置和姿态推算方法3）基于三维激光扫描视觉的相对测量位置姿态参数估计算法；4）集成影像和点云视觉运动分析载体位置姿态估计参数辅助DGPS/IMU 的扩展卡尔曼滤波定位定姿处理算法。课题研究突破了基于激光扫描视觉定位定姿的关键技术，为DGPS/IMU 集成引入CCD 序列影像、三维点云视觉辅助定位定姿，解决GPS 遮挡失锁、IMU 漂移时连续可靠和高精度定位定姿问题，实现无GNSS下激光扫描视觉定位精度优于2侧面，姿态测量精度优于0.01度，为无控制移动测图系统直接地理参考及大范围测图应用提供理论和技术支持，促进测绘学科中导航定位与摄影测量、激光扫描技术、计算机视觉技术的交叉融合，提高地上地下、室内室外全空间三维空间信息数据采集和处理的能力，实现无控制的面向应急测绘保障的快速测绘响应。课题关键算法成果已经在室内、地下空间移动测图，铁路和地铁轨道高精度测量与监测领域得到推广应用。