|
1940 Greater Everglades and South Florida Aerial PhotosetMethodologyPhotographs were cataloged individually into Everglades Historical Aerial Photographic Database. Attribute information about each photograph was recorded, including identity, content, condition as well as general information about the flight and the photography. Photographs were scanned using UMAX Mirage II Scanner: Scan resolution of optical 800 dpi; final radiometric resolution of 8-bit per channel; true color (RGB) compatibility; TIFF format output. Two images which were inadvertently scanned as grayscale are noted in the 'oddities' section below. Scanned images were rectified using Erdas Imagine 8.7. Rectification process followed protocol developed in house and described in the report 'Guidelines for quality checking of digital ortho imagery, Issue 2.0.' Images were rectified with image to image rectification, using basemaps derived from digital orthophoto quarter quadrangles (DOQQs). The DOQQs were from a statewide set flown in 1995 and available for download from the Florida Department of Environmental Protection Land Boundary Information System (LABINS) website. These images were mosaicked into basemaps of four quadrangles, resampled to 2 meter pixel resolution and reprojected as UTM Zone 17 North. For a few images, the 1995 basemap images were unavailable or unusable and were rectified to a 1999 DOQQ. These are listed in the 'oddities' section below. To register the historic images, a first-order polynomial geometric model was used. Due to a variety of possible factors, such as the age and warping of the original media and irregularities of the camera (about which we had no information) or the flight, we found that a second-order polynomial model improved overall edge matching in some cases. Those images that used a second order polynomial geometric transformation model are listed in the section called 'oddities' below. Operators manually selected 10 to 15 tie points per image scattered evenly throughout the image. In some cases additional tie points were subsequently added automatically using the prediction function of Imagine. In all cases the total RMSE for the transformation model was well below a threshold of 0.25. Following the selection of tie points, images were resampled to 1 meter pixel resolution using a bilinear interpolation method. The resulting image was checked with respect to edge matching in a geographic data viewer (ESRI ArcMap 8.3). Once the operator was satisfied the image was properly registered and appeared to match up with its neighbors, the edges of the image were cropped using a subset function. All images, with the exception of 3 (listed in the 'oddities' section below) were subset to a square of 7500 pixels and saved in their final file format as geotiff files. There are 1401 images that were known to have existed in the 1940s set of aerial photography. Of these, we were able to locate and scan 920 distinct images. Of these images, we were able to rectify 793 images. The following tables list the 127 images that were not rectified and the 79 rectified images that violated our protocol in one way or another. We also give a brief description of the nature of the problem.
|
U.S. Department of the Interior, U.S. Geological Survey
This page is: http://sofia.usgs.gov/exchange/aerial-photos/40s_method.html
Comments and suggestions? Contact:Heather Henkel - Webmaster
Last updated:01 August, 2006 @ 01:21 PM(KP)