Video data system to transform images to 3-D models
Under a recent Defense Advanced Research Projects Agency (DARPA) contract, Vexcel Corp. (Boulder, CO) is exploring an imaging system that will automatically generate three-dimensional (3-D) models from video images. This system is slated for use in military applications, where 3-D scenery information could be captured and then used in virtual-reality, visualization, and mission-rehearsal systems. In addition, 3-D object modeling of physical objects and environments has possible applications in machine design and industrial engineering.
Although a fully automated video-based modeling system has yet to be achieved, the benefits of this DARPA program are already being realized. Companies such as Ford Motor Co. and Mobil Corp. are using the technology to generate images for engineering applications. The first phase of the project focuses on automating image correlations and extracting edges and surfaces. These techniques are being incorporated into Vexcel`s FotoG-FMS software to allow developers to generate 3-D models from digital images. In the next phase, structures from motion algorithms, will be developed. For information, contact Lanny Mack at [email protected].
Personal recognition system can distinguish identical twins
Obtaining security access to an automated teller machine, a personal computer (PC), or a protected entry is now possible by just gazing into a one-way mirrored camera. Developed by Identification Technologies International (ITI; Columbia, MD), the One-on-One facial-identification system uses off-the-shelf hardware components, proprietary software, frame-grabber boards, and PCs. This system, claims ITI, is so precise it can accurately distinguish identical twins.
The system initially captures the facial image of the person requesting access with a 1/3-i. charge-coupled-device board camera. Then, it digitizes the images using a CX100 PC-based frame grabber from Imagenation (Beaverton, OR). Next, the system extracts and compresses the digitized images of the areas around the person`s eyes as 96 bytes of identification information. Lastly, this information is stored on a smart card, which precisely registers the person`s facial topography. "Because the area around a person`s eyes changes very little with time," declares David Hertz, chairman and CEO of ITI, "it can be used as an effective means of image identification." The downloaded image data on the smart card become an electronic key that can verify the person`s identity for a range of applications, such as cashing checks and granting admission.
For personal recognition, a smart card reader unit extracts the stored facial features list while the person`s facial image is being acquired. These features, known as personal identification vectors, are compared to the acquired image data; if they match, access is granted.
The company developed the algorithms to process images and ensure consistent image illumination when data are both captured and verified. According to Hertz, because the system uses infrared illumination, a person can be isolated from the background even in dark space.
