A Computer Vision System for Automated Grading of Rough Hardwood Lumber Using a Knowledge-Based Approach

Abstract

A sawmill cuts logs into lumber and sells this lumber to secondary remanufacturers. The price a sawmiller can charge for a volume of lumber depends on its grade. For a number of species the price of a given volume of material can double in going from one grade to the next higher grade. Thus, accurately establishing the grade of a volume of hardwood lumber is very important to both seller and purchaser of this material. Currently, the grading of hardwood lumber is done by human inspectors. Unfortunately, they often lack consistency in their grading of lumber. Hence, there is a strong motivation for wanting to develop an automatic lumber grading system. The research reported in this paper is aimed at developing a computer vision system that will be used to automatically grade rough hardwood lumber. The purpose of the computer vision system is to locate and identify grading defects in a species-independent manner. The current system can detect four of the most common types of grading defects: knots, holes, wane, and splits/checks. The system has been designed using a knowledge-based approach employing a Blackboard framework. The system has been tested on a number of boards from the most used hardwood species. The results indicate that the development of a fully automatic grading system is possible.

Citation

Cho, Tai-Hoon; Conners, Richard W.; Araman, Philip A. 1990. A Computer Vision System for Automated Grading of Rough Hardwood Lumber Using a Knowledge-Based Approach. Proceedings, 1990 IEEE International Conference on Systems, Man, and Cybernetics. pp. 345-350.