Development Of On-line Inspection System For Veneer Appearance Quality Based On Machine Vision

Plywood is made by gluing an odd number of veneers.The quality of plywood in China is uneven and the overall quality is low.The important influencing factor is the appearance quality of the veneer.Therefore,the evaluation of the appearance quality of veneers is the key to improving the quality of plywood in China.The high efficiency,accuracy and reasonable evaluation method of the on-line quality inspection system of veneer are the research hot issue in this field.Therefore,this topic has carried out corresponding research.This project developed a set of on-line quality inspection system for single-panel appearance based on machine vision.Firstly,a machine vision inspection system was built,and a large number of single-board images were collected to build a sample library.Secondly,the digital image is distorted,and then the background area is extracted by the background subtraction method and the mask method,and the actual size of the board is calculated.Then,the image of the single-board area is smoothed and the defect is segmented.In this paper,four defect segmentation methods are used,which are the Otsu method,the iterative method,the fractal theory method and the dynamic threshold method of different color channels.The segmentation accuracy and running time of different algorithms are compared.Then the color features and LBP texture feature extraction are performed on the segmentation defects,and the defect recognition experiments are carried out by using k-nearest neighbor learning and SVM classification algorithms.The classification accuracy of SVM based on Gaussian kernel is built based on one-to-one mode.The highest is 92.67%.Finally,the author designed the appearance quality evaluation model of the veneer based on the national standard and actual production requirements,and developed the PC software.The experimental results show that the on-line inspection system for the appearance quality of the board is effective and feasible,and the system runs stably.