From the Ground Up: Direct georeferencing in aerial photography.
Technology continues to affect the way mapping projects are conducted. These changes have been profound over the last 10 years. Some of the most significant technological improvements directly impact the way we control aerial photography. The impacts in this area can be measured in tens of thousands of dollars for a typical project. The changes can also shave weeks off a project’s schedule.
Direct georeferencing is certainly one of the most significant applications of new technology in the mapping arena. This approach combines the technologies of airborne GPS (ABGPS) for determining the position of the plane and camera with the abilities of an inertial measurement unit (IMU) for determining the rotation of the camera.
Inertial systems have been around for more than 50 years. However, the first system built for use in the collection of aerial photography wasn’t released until 1996 when Applanix Corporation of Toronto, Ontario, Canada, introduced its Position and Orientation System for Airborne Vehicles (POSAV). Applanix remains the dominant player worldwide for mapping applications. Inertial measurement only became useful in aerial applications once the systems could be made small enough and light enough in weight that they could be rigidly mounted on an aerial camera. Furthermore, the systems were not of value in this arena until the Global Positioning System became fully operational. The combination of ABGPS and inertial measurement provides great benefits.
Controlling Aerial PhotographySix different variables must be solved for each photographic exposure before mapping can take place. First, the 3D position of the camera must be determined—the x, y and z at the center of the photograph. Second, the 3D rotation of the camera must be resolved—the angular measurements omega, phi and kappa for the photograph. Think of these last three variables as the tip, tilt and swing of the camera. The camera could be tipped slightly to the front or rear of the aircraft. At the same time it could be tilted toward the left or right of the aircraft. The camera could also be swung off axis from the desired flight direction due to a crosswind.
The conventional way of determining these six variables for all frames of photography is done in a two-step process. First, lots of control points must be placed on the ground in and around the project area. GPS procedures are normally used to determine the precise position of each of these control points. Second, a complex mathematical process termed analytical aerotriangulation is used to tie each of the photographs together and relate these photographs to the position of each of the ground control points. Both the control and aerotriangulation phases can be significant cost items in a mapping project.
The mapping industry saw a great improvement in the traditional way of controlling photography around 10 years ago when ABGPS procedures were first used to provide the three-dimensional position of the aircraft at all times during a photographic flight. Significant savings in terms of both cost and time can result from ABGPS procedures, particularly when used on large projects. For example, my crew and I used ABGPS procedures to control the mapping of an 1,800 square mile area of Atlanta in 1995 for locating the transportation infrastructure for the Olympic games. Several weeks and more than $100,000 in project costs were saved on this project alone.
While ABGPS technology greatly enhances the way we control photography, analytical aerotriangulation must be carried out before any project mapping can take place. The reason here is simple: although the ABGPS solution provides the accurate position of the camera for every photograph, it does not provide any information about the rotation of the camera. The rotation of the camera can only be determined from aerotriangulation in this instance.
Potential UsesDirect georeferencing is ideally suited for many different types of projects, especially in remote or inaccessible areas. However, the limiting factors can generally be summarized as project size and required accuracy. For example, a very small project could actually be more expensive using the direct georeferencing approach, as there are some fixed costs to overcome when applying direct georeferencing techniques to controlling a mapping project. Moreover, while the technology provides very accurate results, large-scale mapping projects planned for 1"=50' scale or most 1"=100' scale projects will still require the accuracies gained from a traditional control approach due to the inherent accuracy limitations of ABGPS technology. Any large project, however, that requires mapping at scales of 1"=200' or smaller might benefit from this control approach.
The lead time for placing a plane in the air to acquire photography under the direct georeferencing approach can be as little as an hour or two. Compare this to the lead time required for a traditional approach when the planning and placement of targets can take a few weeks and the advantage should be clear. In extreme situations or environments, the advantages offered by the new approach can stand out. For example, in times of emergency, such as an act of terrorism or when natural disasters strike, the ability to put an aircraft in motion on a moment’s notice can have profound value.
Cost SavingsThe use of direct georeferencing in a mapping project can lead to substantial savings in both cost and time necessary to complete the project. The savings are derived from the elimination of aerotriangulation for the project as the substantial reduction of ground control points within the project area. Ground control points are not absolutely required in this case, but some quality control points should be established to verify the accuracy of the mapping performed using the ABGPS and inertial solution.
It is important to understand, however, that the aerial photography acquisition cost will increase substantially when using direct georeferencing. There are several reasons for the increase in up-front costs. At least one ground-based GPS receiver is required to allow the accurate post-processing of the flight data. Moreover, the actual time required to fly the project can increase substantially due to current restrictions on banking the aircraft in the turns to maintain GPS satellite lock, and further restrictions of the flight window due to the strength of the GPS satellite configuration. Six or more satellites are strongly preferred for the flight window. Furthermore, a maximum positional dilution of precision (PDOP, a measurement of the strength of the satellite constellation’s geometric configuration) of 4.0 is necessary for most projects. It should be noted though that improvements in the IMU and ABGPS processing are expected to eventually eliminate the need to fly flat turns, thus making this extra cost go away.
Another factor in the increase of the photo acquisition cost is the recovery of the cost of the IMU. An initial investment of over $200,000 is required to add a top-of-the-line IMU to an existing aerial camera. This cost must be recovered over some reasonable length of time for this investment to be a wise business decision on the part of the aerial firms. Finally, all the in-flight data must be carefully processed after the aerial mission is complete. This can add a day or more of office processing time to the cost of a typical project. Table 1 illustrates the increased cost of photo acquisition and the overall cost savings for an actual medium-sized project where the direct georeferencing approach was used to control the photography.
Some projects are better suited for this approach than others. Large block-type projects, like the mapping of entire cities or counties, can be the perfect project for direct georeferencing. Linear projects, like roadway or utility line mapping, may also work well for this approach. However, in a linear project you are always flying away from your ground-based GPS occupation station (in contrast to flying block coverage where you are always flying around the base station) that is used to post-process the flight data. The typical accuracy requirements for most projects therefore require multiple base stations along a linear project. This obviously makes the logistics more complex and increases the cost of the photo acquisition. It is expected, however, that new developments using satellite based Real-time Kinematic (RTK) differential GPS corrections and Continuously Operating Reference Station (CORS) data, which are free via the Internet, will prove to reduce this complexity and cost in the near future.
Finally, one more issue can come into play for selecting the most appropriate way to control a project. Many aerial firms fly projects every year during the spring flying season for clients on speculative projects. In this case, the client authorizes the photo flights before they are absolutely certain of a project’s status in order to capture leaf-off photography that is suitable for mapping later on. If the project’s notice to proceed comes in the middle of the summer they will have appropriate photography to work from. If the project is turned down, the film sits on the shelf unused. A traditionally controlled project enjoys an advantage in the acquisition of speculative photography as the ground control and aerotriangulation efforts can be postponed until after the exact nature of the project’s viability is known. The photography is the only thing that has to take place prior to the leaves coming out in late spring. Therefore, a lot less money is at risk depending on the project’s status.
Just Getting StartedTechnology is constantly changing the way the mapping industry tackles new projects. Direct georeferencing continues to play a very important role in the way mapping projects are controlled. For the right projects, this approach can provide the accuracies that you need while saving time and significant dollars along the way.
Joe Hutton, the director of airborne products for Applanix, summed up direct georeferencing well: “Although direct georeferencing has revolutionized the mapping industry over the last six years, in many ways it is still in its infancy. New inertial sensor technology, enhancements to GPS, better processing algorithms, miniaturization and direct integration with new digital camera systems and other sensors promise tremendous benefits and cost savings. These changes will provide the technology base to support the push toward near real-time or even real-time mapping products. It is quite an exciting time to be involved in this industry.”