Experimental Verification of the Influence of the Quality of Input Images on the Processing of the Photogrammetric Model

Anotace: Photogrammetric acquisition and subsequent data processing has become increasingly common in recent years. However, this method of documentation is affected by many circumstances that can negatively influence the results. This paper is devoted to the experimental determination of the influence of the quality of input images on the processing of a photogrammetric model, because the quality of input images is essential for the accuracy of the final product and can be influenced by various factors. The effect of different image quality on the ability of software to align these images and on the accuracy of this alignment and subsequent processing of was tested practically in this case. Three specific interior surfaces and three specific exterior surfaces were used for the experiment, each of which was photographed in four series - one series was taken with high quality images, while the remaining three were specifically degraded in quality. Three photogrammetric software, Agisoft Metashape Professional, iTwin Capture Modeler and Pix4Dmapper, were selected for processing. In all software, all acquired series were processed and the ability to perform image orientation and the subsequent quality of this orientation was mainly monitored, where reprojection error was the main criterion. In the case of successfully aligned series, a dense point cloud was produced, which was also subjected to analysis. Based on these results, appropriate procedures for photogrammetric processing were determined when conditions that may negatively affect image quality are known in advance.