Font Size: a A A

Research On Optimization System Of Solid Wood Based On Machine Vision

Posted on:2020-04-20Degree:MasterType:Thesis
Country:ChinaCandidate:Z K HuFull Text:PDF
GTID:2381330626951022Subject:Mechanical and electrical engineering
Abstract/Summary:PDF Full Text Request
In view of the less forest resources and low comprehensive utilization rate of wood in China,the application of intelligent and automated wood processing equipment for wood processing enterprises was strongly advocated to effectively improve the quantity and quality of solid wood board production.However,most domestic wood processing enterprises still adopt manual marking recognition and processing because of low technique or high cost of equipment developing.This situation inevitably leads to several problems e.g.low automation,high cost of manpower,strong subjectivity of defect identification and inaccuracy,and unintelligent automatic detection and sawing processing.In order to solve these problems above,this study proposed a machine vis ion-based optimization system for solid wood boards,which mainly completed the following researches:(1)Pine and Chinese fir were chosen as the research species.The selected panel specimens contained common natural defects such as tight knots,loose knots,cracks,decay,etc.The surface defect images of specimens were acquired on an image acquisition device.The causes of common wood natural defects and their effects on the quality of solid wood were analyzed referred to relevant literatures.(2)Defect detection of solid wood combined with Single Shot-multibox Detector(SSD)algorithm was proposed and adopted.As widely used target detection algorithm,SSD was excellently performed in target recognition accuracy and time.Based on the improved SSD algorithm,feature pyramid network structure to fuse the high-level semantics and low-level features were used in this study.Deep residual network was also adopted as the basic network part of the improved SSD model,to optimize the regression of the predict ion boundary box and the input characteristics of classification task.(3)Optimal processing system based on Twin CAT3 was designed.The improved SSD algorithm transferred the results of surface defect recognition to soft PLC system based on Twin CAT3 with candidate frame information(coordinates of candidate frame center point,length and width of candidate frame)and defect recognition type information.Meanwhile,the soft PLC system completed the optimal processing according to the processing list and mo de given by users,and control feeding of high-speed saw cutter.Afterwards,the sawing servo system completed the optimal sawing of solid wood.Finally,materials were classified and discharged through a multi-stage pushing device.A friendly man-machine interface was developed configured with the device for convenience.
Keywords/Search Tags:Solid wood, Machine vision, Defects recognition, Single Shot-multibox Detector algorithm, Optimization processing
PDF Full Text Request
Related items