Performance analysis of the hardware and software components of an optical scanning system for monitoring lumber sizes

Graduation date: 1995In recent years, the forest products industry has been\ud forced to cope with a supply of raw material that is\ud declining in both quality and quantity. In addition, log\ud prices are escalating rapidly. Because log costs may\ud represent as much as 90 percent of the overall production\ud costs of lumber (Williston, 1985), it is crucial that a mill\ud make every attempt to maximize grade and recovery.\ud Quality control (QC) is a broad term used loosely to\ud indicate producing a product to meet the customer's\ud Specifications. In lumber production, quality control is\ud heavily influenced by meeting lumber grade requirements\ud while minimizing fiber use. One component of these grade\ud requirements is conformance to size standards.\ud Size control is one of the functions of a mill's QC\ud department. Size control for lumber manufacturing involves\ud minimizing the variations in the dimensions of lumber that\ud are introduced by the sawing process.\ud An adequate size control program enables a sawmill to\ud maximize both grade and recovery. Grade may be maximized by\ud reducing the amount of boards downgraded due to either\ud excess wane or from planer skip that can result from\ud excessive sawing variation. Minimization of sawing variation can also lead to\ud maximizing recovery. It should be made clear that the\ud reduction in target size that can result from minimizing\ud sawing variation will most often result in a slightly longer\ud or wider board from any one log rather than obtaining\ud additional boards (Darwin, 1989). There are however certain\ud limited log diameter breakpoints at which additional boards\ud may be obtained due to decreased sawing variation.\ud What is proposed in this project is an automated system\ud to collect and analyze board measurement information using\ud optical scanning techniques. The system is intended to\ud provide feedback to mill quality control personnel. This\ud thesis has built upon the image processing and quality\ud control system developed by Aslam (1990) at the Department\ud of Forest Products at Oregon State University. The system\ud is discussed in Aslam's M.S. thesis as well as in the\ud article by Funck et al. (1992).\ud The image processing system scans the edges of boards\ud while still held in cant form after leaving a gang edger.\ud The system can provide lumber thickness as well as kerf\ud width data. A size control program is included to analyze\ud measurement data and provide information regarding sawing\ud variation.\ud A far more detailed representation of sawing variation\ud can be obtained using this system than is possible with\ud standard size control practices providing measurement\ud accuracy is sufficient. For instance, at gang edgers\ud immediate feedback regarding sawing variation for both\ud boards and kerf is provided. Sawing variation information\ud is of sufficient detail for diagnosing which machine center,\ud and more specifically, which pocket of saws is causing\ud sawing problems.\ud This thesis continued the previous work by quantifying\ud measurement accuracy using various hardware and software\ud configurations, and by developing the software necessary to analyze and then display the data in the most meaningful\ud form. Software techniques were created to avoid measurement\ud errors.\ud The first step in the project was a pilot study\ud conducted using several different imaging configurations,\ud noise filtering techniques, and edge-detection algorithms to\ud quantify the system's precision in measuring static objects.\ud The second step was to develop display software to\ud convey the detailed information this system can provide.\ud The development of measurement error trapping\ud mechanisms was the third step. Measurement errors may occur\ud through incorrectly locating the edges of boards due to the\ud presence of wane, bark, pitch or bark pockets, or knots.\ud Measurement errors may also occur when physically blocked\ud saw kerfs cause a failure to detect a board edge.\ud Lastly, the system was tested on images representing a\ud range of measurement difficulty. This test was performed to\ud determine the measurement accuracy of the system when using\ud combinations of three different edge-detection algorithms\ud and three different noise filtering techniques.\ud This type of automated measurement collection, data\ud analysis and display system will have a financial benefit to\ud a mill by enabling the mill to quickly identify and correct\ud situations leading to sub-optimal lumber or grade recovery