Aerial Perspective: Innovate Solutions
Aerial Mapping Fall '13 - Aerial Mapping Fall '13
By Lewis Graham
I have been associated with aerial imaging for quite a few years. Strangely, my career in this field began with digital national resource imagery, moved to analog systems, and moved back to digital with the advent of the Z/I Imaging DMC aerial camera system. Each year I find myself saying, “There has never been a more exciting time to be in this industry!” This year is no exception.
Remember how worried many of us were about the advent of commercial satellite imagery? What would happen to our huge investments in air breather technology? Of course, the lower resolution data from these space-borne platforms had very little impact on the rapidly accelerating airborne digital business. By the time 50 cm data were available from space, clients were demanding 15 cm color data. Next came the free imagery of advertising companies such as Google. Again, this served only to whet the appetite of our professional clients for more-precise, higher-resolution data. It made an orthophoto backdrop a necessary component of a vector street map.
Now we are on the brink of another major evolution in the digital imaging business. Professionals and amateurs alike have taken to the skies with lightweight remote controlled aircraft (what we might call “micro-Unmanned Aerial Systems,” µUAS) carrying non-metric digital cameras. Overall, we are seeing a fundamental change in digital imaging. Traditional (dare I say staid?) photogrammetric processes are rapidly being superseded by exciting new algorithms from the fields of robotics and computer vision. Applications such as Structure from Motion (SfM) are enabling the generation of high density, metrically accurate point clouds from low-cost platforms without ancillary positioning information—in plain language, creating dense 3D point clouds without GPS or inertial measurement unit. New comparison algorithms such as Scale Invariant Feature Transform (SIFT) have made comparisons between images of overlapping object space but with significantly different image properties possible. New feature vector search algorithms such as Fast Library for Approximate Nearest Neighbors (FLANN) have made searches of huge image databases for similar images practical. It is astonishing that an enterprising young chap can generate a useable colorized 3D point cloud of the exterior of a building with a $300 Parrot quadcopter and open source software.
Does this mean the end of high-end aerial imaging? Of course not. What it means is a new thirst for innovative products. The digital orthophoto will soon be replaced by 3D colorized point clouds. When viewed from the nadir, they are the perfect “true” ortho. But when rotated, the world explodes into 3D visualization. All of the major large-format camera manufacturers are already offering a path to these models (what Michael Gruber and I dubbed, “Photo-Correlated Digital Surface Models” at a recent conference—does it get any geekier?) through various post-processing software applications. All we need now is a native viewer for common mapping platforms.
On a more traditional front, aerial oblique imaging (first introduced for useful purposes in the American Civil War) has been making a strong comeback during the past 10 years. Made viable by answering simple viewing questions for emergency dispatch (how many floors in that burning building?), it is now on to metric questions such as the exact location of a wire-insulator attachment point. Our insatiable need for quantitative information will ensure the future expansion of precision imagery.
Now the innovation focus needs to be on solutions. Pictometry/EagleView have proven that a technologically simple product such as roof measurements for insurance companies can drive considerable demand. The rapidly expanding need for pipelines to move the fruits of fracking is crying for new products to ensure public and environmental safety. Further out will be the need for data at the 4 cm accuracy level for autonomous vehicles. However, the need is for information, not just informative content. Roofers need measurements, not pretty images. Pipeline managers need encroachment metrics, not fly-through movies. And, of course, they want all this stuff to be in a hosted solution!
We are always on the sharp curve up of the adoption hockey stick in this digital imaging business. We just have to keep moving as the stick superimposes over the next evolution.
Lewis Graham is president and CTO of GeoCue Group and a member of PSM’s editorial board.
» Back to our Aerial Mapping Fall '13 Issue