Systematic, random and gross errors are considered the main problems facing those working in the photo-triangulation processes. The influence of systematic errors on photo-measurements may include lens distortion, film deformation, refraction and other distortions. Usually, these types of errors can be solved by the calibration process. Meanwhile, the traditional least-squares method was used to adjust photogrammetric data in order to solve the problems of random errors. In case observations contain gross errors, the reliability of least-squares estimates is strongly affected. In this paper, two independent mathematical models (photo-variant self-calibration and robust estimation) are combined for solving and processing the problems of systematic and gross errors in one step. Also, this paper investigated the effectiveness of robust estimation models on solving gross errors in close-range photogrammetric data sets that require photo bundle adjustment solution. The results of investigation indicate that all robust methods have the advantage of detecting and removing gross errors over the least squares method especially in cases of observation contains large-sized errors. Moreover, the Modified M-estimator (IGGIII) method has the best performance and accuracy. Furthermore, gross error was also revealed in the residuals |
