Technical Info
Throughout the development of the Automated Airborne Remote Information System (AARIS), we have been relying upon automation as the means to reduce costs. By automating as much as possible and minimizing those parts that require human involvement, we have been able to reduce the cost to acquire GIS-ready images to a point that many who could not afford such images in the past can now use them.
Automation is not the answer to everything though. There are still steps in the process that require a human to review the data. For example, after each flight is processed, we review the output of the IMU to make sure the data makes sense. We also plot out the ground track and various satellite parameters to ensure that the GPS data is accurate. It is also important to have a human look at each image and make sure that it is exposed properly and that adjoining images line up reasonably well. Further, when a large site requires multiple images, they must be combined in a mosaic that at this time requires a human to select reasonable cut lines.
We will continue to work on all aspects of AARIS to optimize the system to provide the images required at the lowest possible cost.
Our technology
The methods we use to acquire images are based on the integration of an inertial measurement unit (IMU) and a dual-frequency global positioning system (GPS) receiver. With post-processing, the integrated IMU/GPS provides a very accurate location and orientation of the camera at the moment when the image was acquired. This information (X, Y, and Z for position and omega, phi, and kappa for orientation) allows us to orthorectify an image without ground control points. A more detailed description of this technique can be found in the paper, "An Integrated INS/GPS Approach to the Georeferencing of Remotely Sensed Data", Schwarz, et al, Photogrammetric Engineering & Remote Sensing, Vol. 59, No. 11, November 1993, pp. 1667-1674.
Our approach builds on this technique by adding flight-planning software, moving-map flight software, an in-flight video aiming system, post-processing software, and significant automation. As our system evolved, we have continued to seek methods that will automate as much of the image-acquisition process as possible to keep the cost for the images low. We have also delayed certain steps in the system's evolution until technological advancements allowed us to provide a capability within our cost structure.
In 2003 we made the final step towards our goal of direct georeferencing by adding the Litton LN-200 IMU to the system. At this time we also added a NovAtel OEM4-G2 GPS and the Canon 1Ds 11-Mpixel digital camera. Since then, we have been delivering low-cost GIS-ready images to a variety of customers.
Our system takes four bands of color at the same time; blue, green, red, and near infrared. These four bands are combined to create a variety of image types. The green, red, and near infrared are combined to create the color infrared image while the red and near infrared are combined to create the NDVI image.
Flight planning
As we modify AARIS to meet the needs of various applications, we also modify the flight planning software so that we can optimize the time it takes to acquire the images. Many of our applications require acquiring images of many sites that are not adjacent to each other. Therefore, the flight planning software must be able to allow us to create a very complex flight plan very easily. MAARIS, the map-based flight planning software for AARIS, presents a graphical interface that shows all of the sites in the image-acquisition database.
To plan a flight, all that need be done is for the pilot to connect the dots. Each site is color coded to show the flight altitude required. The data provided for the site is also shown in the lower right side of the screen. The upper right corner shows a real-time calculation of the cost of the flight so that the pilot can ensure that the planned flight fits within the cost structure.
Once the flight plan is determined, the flight data is automatically formatted for use in AARIS during the flight. No other interaction is required of the pilot. All that now need be done is to follow the prompts on the flight display and acquire the images.
Image acquisition
The time it takes to acquire each image is the most time-consuming aspect of AARIS image acquisition. Most of the rest of the process has been automated, but the laws of physics limit how efficient we can become at acquiring images. The moving-map display that TALON, the in-flight software for AARIS, presents to the pilot is intended to make image acquisition as easy as possible. In most cases, the pilot merely presses the IMAGE button and then the NEXT button to set the system up for the next site. If needed, the pilot can review images taken and make adjustments while in the air.
AARIS includes a video aiming system along with the digital camera that acquires the final images. This aiming system allows the pilot to easily (and quickly) verify that the image of the site will be acquired properly. The video aiming system shows an area that is larger than what the digital camera will acquire. This allows the pilot to make last-minute changes so that the entire site will be properly acquired.
The video aiming system also includes a GPS-based intervalometer (the blue line on the screen shot) that shows the percent overlap of each image and can also automatically acquire a preprogrammed number of images with a preprogrammed overlap.
The advantage of the video aiming system for the types of applications we serve is that in many cases we only need a single coordinate near the center of a site to acquire the entire site. This is important when acquiring images of sites that change size, such as mine sites, especially when you consider that the reason for the image is to find out how large the site has become.
TALON accepts information from MAARIS to determine the flight path. When images are acquired, they are stored on a removable hard disk along with the camera, GPS, and IMU data. After the flight, the hard disk is removed and the data is then transferred to the post-processing computer and is also backed up onto a DVD.
Post-processing
The process of combining all of the GPS and IMU data with the image to produce the final orthorectified and georeferenced image is very complex. We have automated much of the process to minimize the time a human is involved with our post-processing AARIS software, PAARIS. There are still some areas that require humans, such as quality control (where we ensure that each image meets our high-quality standards), but we are continuing to work towards as much automation as possible.
To orthorectify the images, we are using Leica Geosystems ERDAS Orthobase. This is the leading software in the field and has an excellent track record.
The post-processed images and data are then separated into subdirectories based on customers or other requirements. These groups of images are then written to CD, DVD, or external hard disk for delivery.
Direct Georeferencing
The standard method for creating an orthophoto is to first set up ground control points. These are targets placed on the ground by surveyors to supply a known location on the surface of the earth. These targets show up on the image and are used to create the orthophoto. This is a labor intensive process where the images to be taken have to be determined well in advance so that the survey teams have the time to place the targets. It then takes considerable time to identify the targets on each image to orthorectify the image. Not only does this take additional time, it makes it very difficult to provide a quick response as so much of the project must be planned and prepared ahead of time.
With Direct Georeferencing, additional equipment is added to the aircraft to eliminate the need for the ground control points. Along with the added computers on the aircraft are an inertial measurement unit and a dual frequency carrier phase GPS. These sophisticated sensors are used to determine the external orientation of the airborne camera and provide the ability to create GIS-ready images very quickly. As our system does not require the ground control point targets, we do not have the same delay waiting for a survey team to place targets thereby allowing us to provide quick response to your needs.
Our system is much more than just a camera. The estimating and flight planning software allows us to respond very quickly to new project needs. The airborne system includes both color and infrared cameras so that we can provide a number of image types with a single flight. The post-processing software is a semi-automated system that allows us to create GIS-ready images very quickly. Other software, such as EagleVIew, allow us to provide the answers that you need that go beyond just a GIS-ready image.
We created this system so that we have the capability to modify it as needed for new applications. Further, by providing the entire process in a single company, we have control of the schedule and quality to make sure you get the answers you need on time. It also simplifies your life as you have a single company to deal with to go from your requirements all the way to the answer. If you have a question about the delivered answer, you can take care of it with a single phone call.
The images we supply are GIS-ready. When you receive the images, you simply load them into your GIS (Geographic Information System) and start using them. Each image is georeferenced so your GIS knows where to put it. Each image is also orthorectified so that distortions caused by ground elevation changes are removed. By placing an image at the lowest layer, you can add layers such as streets and property boundaries on top of the image for easy interpretation. You can also change the transparency of the image to lay it on other images to see how a site has changed over time.
The most important aspect of a GIS-ready image is that you can load it in as a layer along with the other layers in your GIS database. You don't have to perform a reprojection, realignment, orthorectification, registration, or any other procedure. You just load it and get to work.
Orthorectification is important to remove distortions caused by differences in ground elevation. Any image taken from above will have these distortions. A camera converts the 3-dimensional world into a 2-dimensional image. In doing this, the 3-dimensional reality is distorted. Consider an airborne camera taking an image of the ground that has a few equal-sized objects, some on a mountaintop and others in a valley.
If an image of these objects is taken from above, the objects on the mountain top would appear larger than the ones down in the valley. The resulting image could not be used to make realistic measurements, as shown in the upper left of the following figure. Further, the image does not include any geographic reference information to enable the location of each object.
The process of orthorectification distorts the distortion out of the image by using models of the earth's elevation called digital elevation models (DEMs). The process of georeferencing orients the image and creates geographic reference information for it. The result is an image that you can load into a GIS and combine with other layers such as various ground-cover types, roads, and property boundaries.
The amount an image may be distorted depends upon many factors. The following example shows a raw image taken of a stream in a rather mountainous area, along with an orthorectified image of the same area.
Color Code:
Blue- Roads
Green- Railroads
Red- River
Orange- Stream
Yellow- Recent Clear Cuts
Typical GIS software allows you to very easily load GIS-ready images and data into a graphic display. You can turn layers on and off, compare data, make measurements, create new data, and easily compare huge amounts of data.
There is no way for a human to interpret all of this data as a pile of numbers. But, as an image, our mind is very good at interpreting that entire 26 foot high stack of numbers. Humans are also very good at identifying when something is not as it should be. There is a good chance this is because not that long ago we had to quickly identify something looking through the grass with big teeth that might want to eat us.