Project Brief
General Competition (November 1993)Robust, Fast 3-D Image Processing and Feature Extraction Tools for Industrial Automation ApplicationsDevelop a standard set of mathematical tools for 3-D image processing and object feature analysis to enable the cost-effective development of commercial three-dimensional machine vision systems. Sponsor: Perceptron, Inc.23855 Research DriveFarmington Hills, MI 48335
For several years, researchers have struggled to develop accurate, useful machine vision systems, pairing digital cameras with computer processing to approximate human sight for a variety of industrial tasks. They have had the most success with two-dimensional images which show features such as intensity or color. Quite a few firms now offer standard 2-D machine vision products, and they are in reasonably widespread use in some niche markets such as printed circuit board inspection. Broader use of machine vision has been stalled, however, because many potential applications require three-dimensional vision. Sorting jumbled parts in bins is the classic example. Several new hardware systems suitable for 3-D machine vision have come on the market recently, based on techniques such as moiré interferometry, triangulation, and scanning laser radar. The principal obstacle now to wider use of machine vision in industry is the lack of a good software base for developing 3-D vision systems. There are a variety of known techniques for reducing noise and other distortions in 2-D images, libraries of standard algorithms for detecting and analyzing image features such as edges, and standard test environments for evaluating system performance. With the exception of some basic 2-D algorithms extended to 3-D cases, none of these exist for 3-D vision systems. Each new application is custom-designed and built, with all the expense that implies. Perceptron, a leading firm in 3-D measurement and imaging solutions, proposes to work with the School of Engineering at the University of Michigan at Dearborn to develop standard 3-D image pre-processing techniques for noise reduction and data smoothing, generic 3-D image feature extraction techniques, and a test environment for 3-D vision systems. Success will enable the machine vision community to develop cost-effective 3-D vision solutions to a wide variety of common industrial tasks.
|
|
ATP website comments: webmaster-atp@nist.gov
Privacy Statement / Security Notice • NIST Disclaimer • NIST Information Quality Standards NIST is an agency of the U.S. Commerce Department |