description:
|
This mosaicked image covers approximately 140 square miles of the Los Alamos National Laboratory's occupied area. The image depicts color digital aerial photography acquired in the fall of 2008 during leaf-off conditions. Each orthorectified pixel is exactly one half foot square. Accuracy of this imagery is 1"=200'. This high resolution imagery is to be used for internal purposes only.
Lens and terrain distortion have been removed and tonal differences between adjacent images have been minimized to form a mosaic of the entire project area. Orthorectified imagery is assumed to depict features at ground level, so above ground features may show some distortion from elevation. The only exceptions to this are bridges and elevated roads, which have been checked for distortion and manually replaced by their previous unrectified versions. Each orthorectified pixel is exactly one half foot square. Accuracy of this orthophotography was tested using National Spatial Data Infrastructure (NSDI) Chapter 3 process and conforms to American Society for Photogrammetry and Remote Sensing (ASPRS) Class I Standards for 1"=200' horizontal mapping. Horizontal accuracy of all ground control visible in imagery shows a displacement of not more than two pixels (one foot) 95% of the time as tested. Color balance is consistent across the project.
Aerotriangulation results for all raw imagery was checked for horizontal accuracy. All visible control points were checked in orthorectified imagery for horizontal accuracy. All bridges were checked for warping during the orthorectification process and corrected. All imagery was checked for consistency of color balance across the entire project area. All frames were checked with total visual inspection at full resolution to determine quality and corrections of misplaced seamlines and other potential quality issues.
Source images are collected with a digital sensor with 4096 brightness values (12 bit) per band. Each image's histogram is clipped at high and low values to increase contrast, and then a digital dodge is performed in order to reduce vignetting of brightness and darkness trends across the frame.
The horizontal positional accuracy and the assurance of that accuracy depend, in part, on the accuracy of the data inputs to the rectification process (elevation data, aerotriangulation control and methods, camera calibration, and source imagery parameters). These parameters are reviewed post orthorectification against the output imagery and initially validate the product is in compilance with ASPRS standards. Additionally, NSDI Part 3 Chapter 3 - (http://www.fgdc.gov/standards/projects/FGDC-standards-projects/accuracy/part3/chapter3) outlines process for determining horizontal RMSE and 95% confidence level for accuracy. Data deemed useful for this validation include independent higher accuracy geodetic control, GPS ground surveys, photogrammetric methods, and databases of hight accuracy point coordinates. Check point locations are generally distributed more densely in the vicinity of important features and more sparsely in areas that are of little or not interest. This data set has uniform positional accuracy thus the points are spaced so 20% of the check points are located in each quadrant of the dataset. A total of 20 points were selected and combined with project control data to achieve a 62 point sample of control visible within the final products. Bohannan Huston, Inc. selected the points used for the independent accuracy check. This data set has tested to a horizontal RMSE of 0.66 feet and a 95% confidence of 1.14 feet. Basic process steps included test point location determination of well defined features, field GPS real time kinematic measurement of selected points, comparison of the vector displacement from source control to location coordinates pulled from image interpretation, and finally an RMSE calculation of the check point sample to determine accuracy and confidence. Results for the NSDI Part 3 Chapter 3 horizontal accuracy test can be obtained from the Bohannan Huston, Inc. Product data tested 1.14 feet horizontal accuracy at 95% confidence level. Testing indicates that final product conforms to ASPRS Class 1 horizontal 1"=100' specifications.
ASPRS Class 1 Mapping Standards for 1"= 200' horizontal require an RMSE of 2.0' is achieved. Control points that were used to aerotriangulate the positions of the images at time of exposure were identified in the orthorectified images and found to be displaced not more than three pixels. All other ground features therefore should not be displaced from their true locations by more than an average of three pixels. Features not on the ground, such as building rooftops or trees, may be displaced. This accuracy specification was validated with NSDI processes defined in Part 3 Chapter 3 - National Standard for Spatial Data Accuracy.
Images were mosaicked into large aggregate tiles and automatic seamlines were used to combine the most nadir components of individual images. OrthoVista software was used in this process.
Raw image data was captured using an Intergraph DMC having a focal length of 120mm, an effective charge coupled device (CCD) array size of 165.888mm across track (left to right) by 92.16mm along track (top to bottom), and a pixel size in its panchromatic band of 12 by 12 microns. Output images were created using post-processing software which removes lens distortion and pan-sharpens color data with a panchromatic image taken at the same instant. Each image had its histogram clipped to optimize contrast, and was sliced to reduce it from a 12 bit image (4096 values per band) to an 8 bit image (256 values per band).
Ground Control targets (panels) were laid down at predetermined locations and their positions recorded with static GPS. Positions were then corrected and network adjusted via the federal government Online Positioning User System (OPUS) to the project coordinate system.
Images were color balanced and color corrected individually to ensure color consistent between flight lines, times of day, and dates of capture.
Aggregate tiles were again tonally balanced and color corrected with entire project in view. A second global tilt was performed to minimize color changes across images. Tiles were cut from the mosaic for review.
Images were orthorectified to remove distortions caused by changes in the elevation of the terrain. The orthorectification process used an exponential interpolation method and a ground elevation sampling interval of thirty two pixels. Intergraph ISBR was employed for the orthorectification.
Images were aerotriangulated to finalize their external orientation. Intergraph ISAT and GeoMedia Pro were used to perform digital point matching/mensuration/block adjustments and quality control queries respectively. Final block adjustments were evaluated at sub pixel accuracy.
The rectified mosaicked imagery were cut into smaller based quarter tile sections approximately 3000ft by 2000ft tiles.
Images were checked for gaps, smears caused by missing surface, and these were fixed and re-created. Above ground features such as bridges warped by the orthorectification or excessive building lean were patched in with versions from unrectified photos. Obvious color differences that remained between source images were feathered. Twenty percent of images selected at random were reviewed at full resolution for any other issues.
Raw image data was captured using a Intergraph DMC camera having a focal length of 120mm, an effective charge coupled device (CCD) array size of 165.888mm across track (left to right) by 92.16mm along track (top to bottom), and a pixel size in its panchromatic band of 12 by 12 microns.
|