Improving Open-Source Photogrammetric Workflows for Processing Big Datasets

Photogrammetric techniques are used for the generation of point cloud data. Multiple images acquired with photo cameras are processed into dense point clouds using photogrammetric workflows. These are normally based on a basic photogrammetric workflow with three steps: tiepoints detection; estimation of camera positions and orientations and of calibration parameters; and dense-matching point cloud generation. When processing large image sets, the execution of the workflows is time-consuming and can become limiting due to the hardware requirements of the processing. This report presents two algorithms that decrease the memory requirements and the processing time of the step that performs the estimation of camera positions and orientations and of calibration parameters. The algorithms are implemented as Free and OpenSource Software (FOSS) tools within the MicMac photogrammetry suite. They are tested with three different datasets to assess the effect on the memory usage as well as on the quality of the generated point clouds. The algorithms presented can be combined with other existing solutions that target at speeding up the processing. Moreover, they allow running photogrammetric workflows in hardware systems that could not be used before. This work has been done as part of the eScience project “Improving Open-Source Photogrammetric Workflows for Processing Big Datasets”.