Research On Automatic Grading System Of Veneer Based On Machine Vision

The veneer is wood processed by CNC rotary cutting machine wood raw materials,veneer surface defects directly affect the appearance and quality of use,so it is very important to veneer accurate surface defects detection and efficient standard quality grading.So far,the identification,detection and quality classification of veneer surface defects have not realized intelligence and automation,so this paper studies and designs the automatic grading system of veneer based on machine vision.The main contents of this paper are as follows:(1)The hardware equipment of the veneer automatic grading system is designed,which is mainly composed of three parts: image acquisition module,grading system and transportation module.The image acquisition module is composed of an industrial CCD camera,a light source,a photoelectric sensor and a computer.Its function is to collect the surface image of the veneer and transmit the image to the computer.The grading system is composed of photoelectric sensor,PLC,relays,solenoid valves,air pump,air cylinder and push rod.The transportation module consists of a motor,a driving roller and a transportation belt,and its function is to transport the veneer.(2)A veneer defect detection model based on the YOLOv3 target detection network is constructed.The veneer surface defect dataset in VOC format was independently collected and produced,including the collection of sample images,the annotation of the dataset,and the augmentation of the dataset by means of data enhancement.The dataset divides defects into two categories: scars and holes.It independently designed a set of grading algorithms,mainly including defect feature extraction and threshold grading algorithm.The target frame of different defects is extracted through color threshold segmentation,and then the maximum area and maximum area of the target frame for scar defects and hole defects in the sample image are calculated.Six defect feature data such as hypotenuse length and total area;the threshold grading algorithm is based on the maximum area,maximum hypotenuse length,total area and other defects of the scar and hole target frame of the veneer in the sample image extracted by the defect feature extraction algorithm The feature data is used as the grading standard,and the threshold is set for grading.In this paper,the quality grade of veneer is divided into five grades.The grading standard can be changed by setting the six parameter thresholds,and the number of grades can also be changed.The control program of PLC to hardware equipment is designed.(3)A set of veneer automatic grading software is designed.The software consists of a login interface and an operation main interface: the login interface is responsible for authority management;the main interface is designed with image acquisition functions,defect detection and quality classification functions,parameters Setting function and detection amount statistics function.Finally,the trained defect detection model was tested,and the accuracy in the test set reached 84.20%;in the system integration experiment,the system can run stably according to the design goal;the automatic grading software for veneer was tested,and each The function behaves normally,achieves the design goal of this software.In summary,the research in this paper provides a set of automatic grading equipment,an efficient defect detection model,a unified standard grading algorithm,and a visual automatic grading software for veneer processing.This research can greatly improve the level of automation and intelligence in the field of veneer processing.