Historical Photograph Orthorectification Using SfM for Land Cover Change Analysis

Abstract

Historical photographs acquired early 1900s provide valuable baseline data for earth surface change analysis. However, orthorectifying such data faces great challenges due to relatively low radiometric quality, the difficulty in finding control points due to surface changes, and the lack of camera calibration information. This study developed a method to standardize the scanned photographs into a common camera model, and a workflow to perform block bundle adjustment in aerial triangulation using the structure from motion technology that calibrates cameras within the aerial triangulation solution. Orthophotograph and DSM were generated from the historical photographs. This could potentially push the quantitative analysis of land surface change several decades back before the modern remote sensing era. The stability of aerial triangulation solutions using structure from motion and the accuracy of the produced DSM and orthophotographs were evaluated.