多激光扫描仪的大场景联合配准及精度评估

Overall Control and Registration Precision Analysis of Large-Scale Terrain with Multiple Laser Scanners

在复杂较大范围的扫描任务中,激光扫描技术需要兼顾扫描精度和扫描效率。传统的ICP配准方式由于缺少严格的平差机制以及累计误差问题,配准的整体精度得不到保障。针对上述问题,采用高精度相位式与长距离脉冲式扫描仪相互协作的扫描方式,并引入基于大场景整体控制网的配准方法。窟区点云通过整体控制网配准,消除误差积累,以提高配准精度。通过在长约1 km、宽约0.3 km的实验区里采集的多达20亿点云数据的配准实验,证实与传统控制测量单站配准精度(5 mm)相比,ICP配准方法的精度(11 mm)仍有待提高。

While scanning large complex scene, both scanning range and accuracy should be taken into account. As there is no rigid adjustment included to avoid accumulative error, traditional ICP method cannot guarantee high overall registration accuracy. To solve these problems, we proposed a scanning method combining the high accuracy phase-shift scanner and long-range pulse scanner, and a corresponding registration method based on overall control network for large scene. Through overall control network, the proposed method can eliminate accumulative error and improve overall accuracy. By registering data containing over 2 billion points collected along a region of 1 km length and 0.3 km width, the experiment demonstrates that comparing to accuracy of traditional method （5 mm） by registering single station to control network, the accuracy of ICP （11 mm） still needs to be improved.