Computer Vision Systems for Hardwood Logs and Lumber

Abstract

Computer vision systems being developed at Virginia Tech University with the support and cooperation from the U.S. Forest Service are presented. Researchers at Michigan State University, West Virginia University, and Mississippi State University are also members of the research team working on various parts of this research. Our goals are to help U.S. hardwood producers automate, reduce costs, increase product volume and value recovery, and market more accurately graded and described products. The first system is being developed to recognize board defects, clear wood, and board outlines of rough hardwood lumber and to label and define these areas. This information will be fed into two computer programs. The first program will grade the board by National Hardwood Lumber Association grading rules. The second program will simulate the processing of the board into standard or specific cutting or part sizes by two different cut-up methods. A pre-prototype vision system is described. A system goal is to analyze images of rough lumber in a species independent manner. Illustrations of original board images, results of segmentation, and the results of defect recognition are shown. We also discuss some of our current problems. The second computer vision system deals with log scanning. This system is being developed to recognize log defects, clear wood, and log outlines and to label and define defect areas. This information will be used to sort logs as veneer or sawlogs; to buck long length logs; to determine how to flitch a veneer log; and how to process a sawlog. Some of our current recognition progress is illustrated.

Citation

Araman, Philip A.; Cho, Tai-Hoon; Zhu, D.; Conners, R. 1991. Computer Vision Systems for Hardwood Logs and Lumber. Artificial Intelligence Applications in Wood Manufacturing Group, XIX IUFROWorld Congress. 9pp.