Automated Georeferencing of Historic Aerial Photography

Historic aerial photos are typically very popular, sought-after collections in library archives. Their ability to be used in GIS is limited, however, by the laborious and painstaking georeferencing process. A method is presented for automating the georeferencing of historic aerial photography. In this method, the Harris Corner Detector is used to detect corner points in already-georeferenced and unreferenced aerial photos, which are then matched by cross correlation. Falsely-matched corner points are removed with the application of the Random Sample Consensus with a six parameter transformation. The pixel arrays of inlier corner point pairs are input to a Java program developed with ArcObjects, which converts their pixel coordinates to map coordinates. The unreferenced aerial photos are then rectified and georeferenced by a Warp() function with a first order polynomial option. The results of tests thus far indicate the method is satisfactory and promises to be an important component in an aerial photo library digitization workflow.