Design And Research Of Wood Surface Defect Detection System Based On Image Features

In view of the current situation that wood processing enterprises generally use artificial wood surface defect detection,this paper adopts machine vision technology to study the image characteristics of wood surface,and designs a set of automatic detection device including software and hardware structure to replace manual detection,thus improving the detection efficiency.The main contents of this paper are as follows:1)According to the production requirements of the enterprise and the characteristics of the board samples,the classification criteria are determined,and the board defects are divided into three categories: live joints,dead joints and oil lines.Determine the process flow of the automatic testing system,design the corresponding mechanical and electrical hardware structure,and make the software development plan for machine vision testing.2)Preprocessing and segmentation of board surface images.Aiming at the problems that general image segmentation methods(threshold segmentation,edge detection,etc.)cannot be applied to all types of wood boards,the segmentation effect is not ideal,and the optimal segmentation parameters need to be adjusted manually,the threshold segmentation method is improved by combining the image characteristics of the board surface,and the mean biased threshold segmentation method is obtained.This method can automatically segment all kinds of samples and ensure the accuracy of subsequent image processing.3)Extract the color features,texture features and shape features of the segmented image,and preliminarily use a total of 12 features including first-order color moment,second-order color moment,energy mean and entropy mean to represent the original image.Then,principal component analysis was used to reduce the image feature data to 8 dimensions as the final image feature extraction result.4)Classification and identification of surface defects of wood boards.Using BP neural network as the classifier,the average recognition accuracy of the three defect types was 92.12% and the average time was 216 ms.The performance of the testing device was analyzed comprehensively,and it was verified that the automatic testing system of wood surface defects could meet the actual production requirements in terms of recognition accuracy and speed.