Study On Mean Variance Detection Method For Surface Defects Of Particleboard

The localization of continuous flat press has promoted the automatic production of particleboard,however at the end of the production line the surface defect detection is still detected by human eyes which causes many problems.On the one hand,the misjudgment and and omission are common during the manual detection especially after a long working time.In this way,the quality of produced particleboard cannot fulfill the standard,and the manufacturer might have economic losses because of return or compensation requirement of customers.On the other hand,the speed of continuous press production line needs to be reduced from 1500mm/s to 0mm/s when using human eyes to detect defects,which reduces the efficiency of particleboard factory.In spite of the disadvantages of low accuracy and efficiency in manual detection,there is no online detection system for particleboards' surface defects in China.The only existed defect detection system for particle board surface is called Argos,whose identification rate is only about 65~70% and it has not reach the stage of online production and application.At the same time,the conflict between the market demand and the quality and efficiency of particleboard production is increasing.In order to adapt to the changes in the market,manufacturers must produce particleboard with high-efficiency and high-precision,by conducting comprehensively testing and controlling the surface quality.Therefore,it is imperative to develop an online surface defect detection system for particleboard.Because machine vision has the advantages of high stability,detection efficiency and accuracy,this study is focus on an online surface defect detection system based on machine vision to recognize defective boards(the surface with sand,shavings,glue spots,soft,oil stain,or sundry),normal boards and protective boards rapidly and accurately.The research contentsare as follows:1.According to the criterias and requirements of online detection of particleboard surface defects,we proposed a scheme of online detection system for particleboard surface defects based on machine vision technology,and elaborated the structure,process and functions of the system.2.A synchronous acquisition system is proposed to solve the problem of image motion blur which affects the identification accuracy.CAM can make camera repeat periodic reciprocating motion,capturing image until the camera speed is synthetic with particleboard.In this study,the hardware construction,control scheme and software implementation of the synchronous acquisition system were introduced in detail,and the experimental results showed that the control system has high stability and could obtain high-definition particleboard images.3.A fast edge cutting method is proposed to obtain the interested area of the image which only includes the target particleboard.Firstly,the barrel distortion caused by wide-angle lens is eliminated with the calibration method of Zhang Zhengyou,and the corrected image is obtained.Secondly,we eliminate the background in the image and reduce the amount of data by pre-cut.Finally,we get the image area of interest by using Canny algorithm,conditional filtering and Hough transform.The test results show that the barrel distortion of the collected image can be corrected and the image of particleboard with no distortion and complete information can be obtained.4.Study is explored on defect detection of particleboard image.Firstly,the image of particleboard is segmented by a local adaptive mean variance method.The segmented image is processed by morphological filtering operation of corrosion and expansion,and the outliers were deleted to recover the image so as to obtain the complete defect area.Finally,the defect area attribute is obtained and the defect discrimination is carried out.The local adaptive mean variance method proposed in this study can suppress noise and overcome inhomogeneity.The segmentation method can perform complete defect segmentation,and the defect detection algorithm has high stability,timeliness and accuracy,and can meet the requirements of online defect detection.5.Finally,this paper introduces the software implementation of the online defectdetection system for particleboard.and we conducted experiment in Fenglin Yachuang(Huizhou)Wood-based Panel Co.,Ltd.The parameters are optimized through the online detection and the system can meet the needs of the company.The designed online defect detection system based on machine vision for particleboard surface can realize the functions of online collection,rapid analysis and recognition result output of particleboard.The time for defect detection is 1.922 s,and the defect recognition rate reaches 97.3%.The self-developed system can make the production of particleboard more efficient with high quality.And at the same time it can fill the gap in domestic industry of particleboard surface defects,and promote the fully automated development of the particleboard production line.