Research On Key Technology Of Intelligent Online Inspection For Wood-based Panel

With the rapid development of manufacturing industry in China,the demand for wood-based panels is also increasing year by year.In the process of panel production,there will be different kinds of wood-based panel surface defects due to the influence of equipment,artificial,environment,material and other factors.This paper studies and designs an intelligent online detection system for wood-based panels by using machine vision technology and image processing technology,aiming at replacing manual inspection and automatic classification of wood-based panels.This paper studies the detection method of wood-based panel based on machine learning and image processing,and realizes the search,classification and information acquisition of the surface defects of wood-based panel:In the aspect of defect finding,an improved LTP algorithm was proposed to calculate the eigenvalue of plate samples by analyzing the difference between defect area and non-defect area.A large number of artificially cropped image samples are studied and trained by Adaboost classification algorithm,and the resulting recognition classifier is used to find defect regions.Finally,the detected defect area is identified and the location information of each defect area is output.On the aspect of defect classification,by analyzing the shape features and texture features of different kinds of defects,a method based on the combination of improved SURF algorithm and Tamura texture features is proposed for feature extraction and description of defect regions.The effectiveness and rapidity of the algorithm are verified by experiments.In order to reduce the redundancy of sample training information and improve the efficiency and speed of defect detection,the similar features of defect regions are clustered by Gaussian clustering.Finally,support vector machine classification algorithm to train the image feature vector and the trained classifier is used for type recognition of defect area.The defect information obtained in this paper mainly includes the category,quantity,area and width of defects,in which the effective area and actual length and width are difficult to obtain accurately.In view of this kind of problem,a defect size calculation method based on image morphology processing is proposed.In this method,the features and parameters of the defective parts are highlighted and calculated by image enhancement,image segmentation,noise elimination and contour extraction.Finally,more accurate defect information is obtained,which is used to screen wood-based panels against the evaluation criteria.Finally,in order to make the wood-based panel defect detection method better applied to the actual production,the software design and function realization of the on-line detection system are carried out through the Microsoft Visual Studio platform.The testing system designed in this paper can effectively detect and evaluate the defects of wood-based panels,and has important application and popularization value.