A multiple sensor machine vision prototype is being developed to scan full size hardwood lumber at industrial speeds for automatically detecting features such as knots, holes, wane, stain, splits, checks, and color. The prototype integrates a multiple sensor imaging system, a materials handling system, a computer system, and application software. The prototype provides a unique general purpose research facility so that a variety of different industry-scale problems can be addressed in both primary and secondary hardwood manufacturing. BACKGROUND There are three main categories into which features on hardwood lumber may be classified. These are: 1) visual surface features (e.g. knots holes, splits, decay, color, grain orientation), 2) board geometry features (e.g., warp, crook, wane, thickness variations, voids), and 3) internal features (e.g., honeycomb, voids, decay). Surface geometric, and internal features have been categorized in this way because they often require different detection methods that depend on where they appear. Most of these features are treated as defects to be removed in the manufacturing process. Current hardwood mill operations examine lumber manually to locate and identify these types of defects. 1The authors are, respectively, Assistant Professor, Dept. of Wood Science and Forest Products, Virginia Tech, Blacksburg, VA 24061-0323; Associate Professor, Spatial Data Analysis Laboratory, Virginia Tech, Blacksburg, VA 24061-0111; Research Forest Products Technologist and Project Leader, U.S. Forest Service, Brooks Forest Products Center, Virginia Tech, Blacksburg, VA 24061-0503; Project Leader, U.S. Forest Semite, Forest Sciences Laboratory, Princeton, WV 24740. The authors wish to acknowledge Virginia’s Center for Innovative Technology and the U.S. Forest Service for partial support of this project.
[1]
S. M. Cramer,et al.
The slope-of-grain indicator for defect detection in unplaned hardwood lumber
,
1991
.
[2]
Philip A. Araman,et al.
Enhancement of the Computer Lumber Grading Program to Support Polygonal Defects
,
1993
.
[3]
Kent A. McDonald,et al.
Defect detection in lumber: state of the art
,
1981
.
[4]
Richard W. Conners,et al.
Automated computer grading of hardwood lumber
,
1988
.
[5]
Philip A. Araman,et al.
Hardwood Lumber Edger and Trimmer Training System
,
1992
.
[6]
Richard W. Conners,et al.
Computer vision hardware system for automating rough mills of furniture plants
,
1990,
Defense, Security, and Sensing.
[7]
K. A. Mcdonald,et al.
Characterizing the roughness of southern pine veneer surfaces
,
1986
.
[8]
Marshall S. White,et al.
CORY: A Computer Program for Determining Dimension Stock Yields
,
1989
.
[9]
Daniel L. Schmoldt,et al.
Machine Vision Systems for Processing Hardwood Lumber and Logs
,
1992
.
[10]
Richard W. Conners,et al.
A machine vision system forautomatically grading hardwood lumber
,
1992
.
[11]
Richard W. Conners,et al.
ALPS- A potential new automated lumber processing system
,
1984
.