Portland State University. Department of Electrical Engineering.
George G. Lendaris
Date of Publication
Master of Science (M.S.) in Electrical Engineering
Quality control -- Optical methods -- Automation, Computer vision, Sawmills -- Electronic equipment, Wood -- Defects
1 online resource (xxii, 182 p.)
The biggest obstacle to robust color image processing of wood is in developing a color model that represents all possible defect colors. When the color model is too general or too specific, defect recognition fails because too many or too few non-defect pixels match the model, respectively. Because a color image of wood contains far more clear and clear-grain colored pixels than grain-knot and knot colored pixels, it is beneficial to first statistically identify and remove the clear and clear-grain colors and to use the accumulated data to simultaneously enhance and normalize the remaining grainknot and knot colored pixels. This process is here called adaptive color correlation. The normal image processing strategy is to search and test for defect features directly. The strategy proposed and developed here is to instead classify all wood pixels containing non-defect colors first, and then identify defect features. Once non-defect features are removed from an image, the task of finding candidate defects becomes easier and faster. This improvement is realized in a sigmoid-shaped color correlation implemented as an adaptive look-up table. As wood has become more expensive relative to manufacturing costs, more efficient methods of maximizing the recovery of clear wood in every board are sought. Optimization, in the present context, is a broad term for selecting products that are made from wood boards so the value of products produced is maximized for a given production requirement. Wood contains random defects which prohibit the production of some products. The normal optimization strategy is to mathematically change the value of under/over-produced products directly. The strategy proposed and developed here is to instead separate optimization into two steps: 1) determine all possible product solutions for a board; and 2) select the single best solution that satisfies value and production goals. Maximum utilization of clear wood is achieved because the solution is "frozen" before mathematically changing the value of products. Recovering long-lengths of clear wood is achieved because various length-based valuation strategies may be implemented as postsolution processes. Separating the product selection process from the solution generation process is shown by this work (simulation) to maximize value recovery.
Goulding, John Robert, "Adaptive Color Correlation of Knots in Wood Images and Weighted-value Product Selection Methods in a Machine Vision System" (1996). Dissertations and Theses. Paper 5189.