机载LiDAR点云与光学影像的整体空三方法研究

The integrated system of airborne LiDAR and frame camera can obtain 3D point cloud and optical image simultaneously, which greatly improves the efficiency of data acquisition. Integrated processing of point cloud and image can make full use of the advantage of geometry precision of laser scanning and the advantage of texture details of image, showing great potentials in a variety of applications. However, owing to the various system error sources, systematic position biases and misalignment of the original point cloud and image are difficult to be prevented. How to eliminate systematic biases and improve the positioning accuracy of the two data source has become a principal issue for further application. Since the differences in resampling methods, data characteristics and location model, existing methods adopt two separate models to adjust the point cloud and image respectively. The main problem of existing methods is that the individual processing of each data ignores the essential relationship between the two data sources. The adjustment models are not rigorous so that the systemic bias cannot eliminate completely. We propose an integrated aerial triangulation in which the errors of the point cloud and image are adjusted simultaneously. In our solution, a rigorous model of the integrated aerial triangulation will be established and the systematic errors can be compensated as complete as possible for utilizing the essential relationship of the two data source. With this new method, positioning accuracy can be improved and the two kinds of data can be registered better.

机载LiDAR与相机的集成实现了三维点云和光学影像的同步获取，提高了数据采集效率。点云和影像的集成处理可以充分发挥激光扫描的几何精度优势和影像的细节优势，应用潜力显著。由于各种系统误差的影响，原始点云和影像的定位不可避免地存在系统性偏差，两种数据之间也无法直接对准。如何消除系统误差的影响，提高两者的定位精度是进一步应用面临的首要问题。由于点云和影像在采样方式、数据特征、定位模型等方面差异较大，现有的处理方法一般将两种数据分开，分别采用不同的模型进行平差处理，问题是忽视了两者之间的内在联系，处理方法不够严格，处理的精度有进一步提高的空间和必要。针对该问题，本项目提出一种点云与影像整体空三的联合平差方法。在对两者的系统误差特性进行深入分析的基础上，依据两者系统误差之间的关联关系，建立整体空三的严格平差模型。通过对系统误差的整体补偿，不仅可以提高点云和影像的定位精度，也可以更好地实现两者的配准。

激光点云与光学影像是两种重要的遥感数据源，点云数据能够直接获取目标的三维信息，而影像纹理细节丰富，两者的融合能够实现优势互补，在地形制图、灾害评估、数字城市等领域极具应用价值。目前Leica、Optech、Riegl等公司相继推出了机载LiDAR与相机的集成系统，实现了三维点云和光学影像的同步获取，提高了数据采集效率。..由于各种系统误差的影响，原始点云和影像的定位不可避免地存在系统性偏差，两种数据之间也无法直接对准。如何消除系统误差的影响，提高两者的定位精度是进一步应用面临的首要问题。由于点云和影像在采样方式、数据特征、定位模型等方面差异较大，现有的处理方法一般将两种数据分开处理，分别采用不同的模型进行平差，这种处理方法忽视了两者之间的内在联系，算法模型不够严格，处理的精度和可靠性还有待提高。针对上述问题，本项目从以下三个方面开展研究：.（1）激光点云与光学影像的联合误差建模.（2）激光点云与光学影像之间同名观测值的稳健提取方法.（3）激光点云与影像的整体空三联合平差以及系统误差的优化求解方法..通过理论推导和实际数据验证，建立了激光点云与光学影像整体空三的联合平差方法框架。利用该套方法对登封实验区、安阳实验区获取的多套激光点云和光学影像数据进行测试，结果表明该方法不仅可以改善点云和影像数据之间的配准精度，还能够实现对两者系统误差的整体补偿，有效提高点云和影像的定位精度。..本项目所取得的成果可以直接作为传统空三方法的有效补充和扩展，项目中的相关理论成果和研究方法，还能够应用于组合导航领域，解决光学传感器和距离传感器的联合平差问题。