Measurement of cutter marks on planed wood surfaces with machine vision methods
thesisposted on 2014-02-03, 14:39 authored by D. Yang
Cutter marks on machmed wood surfaces are generated by the plamng and mouldmg wood machming process. Cutter mark defect is referred to as inconsistency of widths and heights of the cutter mark waves, which IS cntical m some sectors of the woodworkmg industry. Machining speeds m the woodworkmg mdustry are remarkably high In order to meet the demands of high efficiency and high quality, mprocess measurement of cutter marks on machmed wood surfaces is highly desirable. Machme visiOn technology IS bemg widely used m various quality control applications due to Its advantages of non-contact and high data rates. Clearly, machme visiOn IS also highly smtable for m-process measurement of wood surfaces This research focuses on usmg machme vision techniques to measure cutter marks on planed wood surfaces Before machme vision methods are mvestigated, a laser profilometer IS mvestigated for its feasibility of measunng cutter marks on wood surfaces Although the profilometer cannot be used for m-process applications, it provides a good reference for other methods. Three maJor machme vision methods and their vanatwns are investigated They are the Light Sectionmg method and the Differential Light Sectwmng method, the Shadow Analysis method and the Multi-Angle Shadow Analysis method, the twoImage Photometric Stereo method and the one-Image Shape From Shadmg method Nme samples, made of three species of wood - beech, oak and ramin, with cutter mark widths of l.Smm, 2mm and 2 Smm generated on the samples of each species, are tested. Surface profiles measured with all the machme visiOn methods are compared to the reference profiles measured with the laser profilometer. Expenments mdiCate that the Light Sectwmng method and the Shadow Analysis method both work to some extent, the Differential Light Sectwmng method and the Multi-Angle Shadow Analysis method are not practiCal; the two-Image Photometnc Stereo method IS the most reliable machine vision method among all the methods mvestigated; and the one-Image Shape From Shadmg method needs further studies.
- Mechanical, Electrical and Manufacturing Engineering