Abstract:

Archived aerial images provide valuable information for cadastral studies and assessing boundary changes and surface features. However, the loss or damage of auxiliary information, such as the camera calibration file essential for processing these data, poses challenges in selecting an appropriate processing method. In this study, we compare the classical photogrammetry processing approach, which assumes the availability of internal camera calibration information, with a machine vision-based approach that uses self-calibration to perform processing without the camera information. The process of creating an orthomosaic from archived aerial images from the 1960s was conducted using both methods. The results show that the machine vision-based approach can achieve comparable geometric accuracy to classical methods and can effectively process images lacking camera calibration data.