Geomosaics from Digital Imagery

THE NEED

High resolution photography from low flying aircraft can produce imagery of the ground that has very good ground sample distances. In many cases better than any satellite image available. Aerial imagery has the added benefit of being available in near real time - if a user can get a camera in to the sky, they can have high resolution photography in their hands within minutes of landing, and in some cases even during flight. This allows land planners, aerial survey operators, border patrol agents, police, and any else who flies with cameras to get very high resolution, timely imagery of the area of earth they are surveying.

But a survey is not worth much if you have no way to correlate the imagery with a specific point on the earth, so some sort of positioning system is required. Fortunately, GPS systems are now ubiquitous, and even the casual person can acquire a GPS unit for less than $100. This means that aerial survey operators, with minimal expense, can acquire not only high resolution imagery, but also position that approximately imagery on the earth's surface.

Pretty good, unless you want to know precisely where that spot on the ground is, or if you want to measure the distance between two points that are not on the same image, or if you want to build a map of an area larger than any single image. To accomplish those tasks, you need some way to correlate the images you have taken, join them together into a single image, and then export the result into a format that can be utilized by other tools.

There certainly exist software packages on the market today that allow a casual user to create panoramas from a set of still images - many digital cameras have the capability built-in. But, those solutions have strict acquisition requirements - the user either has to ensure the images overlap perfectly when acquiring the photography, or the camera motion has to be sufficiently constrained to allow for a solution, such as placing the camera on a tripod. Neither of those methods is possible in a moving aircraft. 2d3 has solved the problem of correlating large sets of individual images that have little overlap between them - and it operates automatically.

THE SOLUTION

Release 2d3's AltiMap software on a set of images, and it will not only correlate the images in ‘image space' but if the EXIF header contains metadata about the camera GPS position during the shot, AltiMap will also geo-locate the resultant mosaic.

Through a process called Wide Baseline Matching, AltiMap automatically compares features in each image with other images to determine their match in what we call ‘image space'. This means that AltiMap can perform its tasks even if you have no other data about the imagery. Not sure if your image set has good overlap throughout? AltiMap will produce separate mosaics for cases where there is no overlap in the image sets.

Have a GPS connected to your camera? Great! Your camera will put that GPS data into the EXIF header for each image you capture. AltiMap will use that data, together with the results from the image matching process to put your resultant mosaic "on the map"!

FORMATS SUPPORTED

AltiMap supports the following input formats:

• Imagery - JPG, BMP, PNG, GIF
• Data - NMEA, XML, EXIF, TXT

AltiMap Outputs Mosaics in the following formats:

• Imagery - JPG, PNG
• Data - KML, Tile Map Specification (TMS), XML