Study On Detection Of Wood Surface Defect Based On 3D Structure Laser

The detection of wood surface defect is an important procedure in wood processing,which is of great significance to improve the quality of wood products and increase the economic benefits of wood processing industry.The detection accuracy of the wood surface defect detection based on color image or gray image is not high,when the defect height information is introduced,it can achieve higher accuracy of defect detection.In this paper,the wood surface defect detection system based on 3D structure laser is proposed The system can obtain the height map and gray scale of the wood surface at the same time,and the defect image preprocessing,image segmentation,feature extraction and defect recognition are studied in depth.Firstly,on the basis of constructing the detection system,300 defect images are collected to establish the defect sample library,and the characteristics of wood surface defect image are analyzed.The image enhancement is carried out by using the gray scale transformation and the histogram modification,and the mean filter,the Gaussian filter and the median Filtering are used for image smoothing.Then,the image segmentation and feature extraction of the pre-processed image are followed.Image segmentation is implemented by the iterative threshold method,the Otsu method and the region growing method.The main features include geometric features(6 dimension)and grayscale texture(7 dimension)are extracted,and feature dimension reduction optimization is realized by using PCA Finally,SVM,GMM and KNN classification algorithms are used for image recognition of wood surface defect,and the classification results of various algorithms are compared and analyzed.The average recognition rate of SVM algorithm based on Gauss radial basis function is the highest,reaching 94.67%.The experimental results show that the wood surface defect detection system based on 3D structure laser is effective,and it has the advantages of high precision,non-contact and high recognition rate.It also satisfies the basic requirement of automatic detection of wood surface defect,and has certain application and popularization value for the finishing of wood.