Multispectral inspection grades wood

Machine-vision and automated processing systems under development at Virginia Polytechnic and State University (VPSU; Blacksburg, VA) aim to help hardwood producers reduce costs and offer higher-value, more accurately graded products. Earl Kline and his colleagues at VPSU are developing a multisensor machine-vision prototype to scan full-size hardwood lum ber. The system can automatically detect knots, holes, stains, voids, and color of the wood. To do so, the system employs multiple sensors con

Multispectral inspection grades wood

Machine-vision and automated processing systems under development at Virginia Polytechnic and State University (VPSU; Blacksburg, VA) aim to help hardwood producers reduce costs and offer higher-value, more accurately graded products. Earl Kline and his colleagues at VPSU are developing a multisensor machine-vision prototype to scan full-size hardwood lum ber. The system can automatically detect knots, holes, stains, voids, and color of the wood. To do so, the system employs multiple sensors consisting of color imaging sensors, laser-based ranging sensors, and x-ray scanners.

Because each of these different sensing systems provides a unique type of information about the nature of the wood, research is being conducted to which sensing technology or combination of technology are best suited for a particular application. At present, the color imaging system uses an 864 ¥ 1 color line-scan camera configured for a 340-mm field of view.

According to Kline, this is wide enough to handle most lumber specimens. To calculate thickness, a polygon mirror sweeps laser light across the lumber where it is detected by high-speed array cameras. These view the laser line and through triangulation calculate thickness.

To detect knots and other internal features, an x-ray system is used. Configured with a linear x-ray detector, the system operates like a web inspection system, scanning boards as they move down the production line. Image data from the sensors is transferred to a PC using a custom-built PCI interface. This allows data to be collected for analysis as fast as it is generated, according to Kline. For more information on this project contact rsmith4@vt.edu.

More in Factory