基于非参数测量模型的摄影测量方法研究

Study on Close-Range Photogrammetry Based on Nonparameteric Measurement Model

由于近景摄影测量具有测量范围广、精度高和效率高等优点,其在大尺寸精密测量任务中承担越来越重要的角色。其中,基于平差优化的自标定测量模型是保证该方法实现高精度测量的重要原因。然而,随着越来越多的商业级单反相机应用到三维空间测量,发现其测量精度与专业相机相比有一定差距。经过大量分析发现,除了相机本身原因外,自标定模型过多地依赖相机内部参数,尤其是畸变参数,是导致测量精度降低的重要原因。为了降低参数化模型对测量结果的影响,提出一种不依赖相机内部参数的摄影测量方法。结合垂线法和Zeiss实验室标定方法,设计了一种针对大视场相机的非参数标定方法。经过不同图像间同名点匹配和平差初值确定后,便可建立基于角度信息的非参数测量模型。结合平差优化算法,完成对空间被测点三维坐标的精确解算。经过与传统摄影测量结果比对,证明该方法可以有效提高大视场单反相机的摄影测量精度。

Because of the advantage of wide measurement range, high measurement accuracy and high efficiency, the close-range photogrammetry plays more and more important role in large-size accurate measurement tasks. The self-calibration measurement model optimized via bundle adjustment is considered to be the most reliable technique to high-accuracy close-range photogrammetry. As more and more off-the-shelf single lens reflex （SLR） cameras are adopted to three-dimensional measurement applications, the measurement results are not ideal compared with that of professional cameras. After being analyzed carefully, the self-calibration parameterized model has some limitations to the improvement of measurement accuracy in addition to the issues inherent in the qualities of cameras. In order to solve the problem, the close-range photogrammetry without relying on camera internal parameters is studied. The nonparameteric calibration method is proposed, which is suitable to the calibration of large-field cameras. The nonparameteric measurement model based on orientation information is established after the image points is matched and the initial value are determined. The three-dimensional coordinates of target points can be solved accurately via the optimization of bundle adjustment. Compared with measurement results of traditional photogrammetry, it is proved that our method is effective to improve the three-dimensional measurement accuracy with large-field SLR cameras.