Adverse conditions challenge image acquisition
Adverse conditions challenge image acquisition
George Kotelly Executive Editor
Various technical approaches to capturing images are available and in wide use. Each centers on picking up the largest amount of imaging information possible. Underwater, a newly developed hand-held sonar system can detect and identify foreign objects despite near-zero visibility. According to contributing editor John Haystead, US Navy divers are using this acoustic-beam, PC-based, face-mask-display unit to image covert mines attached to the hulls of ships (see p. 44).
On land, a truck-mounted laser-based imaging system is obtaining quantitative real-time data on paved roads at fast vehicle speeds. In this approach, reports Sandra Todd, director of marketing at Coherent Inc., and Liviu Bursanescu, vice president of research and development at GIE Technologies, 3-D laser imaging has been combined with a global positioning system to image road faults (see p. 24).
In the air, fuel mixtures and pressures inside jet engines can thrust missiles at Mach speeds. In fuel-injection design studies for a supersonic-combustion ramjet, researchers are implementing low-light-level sources to capture images of the fluorescence of pressure-sensitive paints applied to the injectors, says contributing editor Larry Curran (see p. 36).
To speed up image-pattern recognition and classification operations, researchers have developed advanced neural-network integrated circuits that more quickly classify imaging data against trained images. According to editor at large Andy Wilson, a novel RAM-based neural network establishes compressed binary image data that represent images as thousands of small shapes for detailed image comparisons (see p. 50).
To ease their system-development tasks, says Andy Wilson, system integrators are turning to graphics-friendly image-processing systems. These vendors are providing tools that free developers from programming tasks and speed development of imaging applications (see p. 56).
In the second part of a series, Peter Eggleston continues his discussion of software-segmentation techniques. He describes how threshold-based segmentation is one approach to extracting information about the structure of imaging features and to separating data parameters (see p. 20).