基于Adaboost与CNN的木材表面缺陷检测

doi:10.16182/j.issn1004731x.joss.17-0262

摘要/Abstract

摘要： 针对木材智能下料前期需要快速获取木材表面缺陷的位置以及类别信息的要求,基于数字图像处理技术,提出了一个组合算法用于木材表面缺陷的快速识别与定位。使用了Adaboost级联分类器在图像中提取出木材表面缺陷区域的候选框,有效解决了传统分割方法对于多目标难以处理的问题。使用了具有自学习特征能力的CNN(Convolutional Neural Networks)模型对输入的候选框进行分类,克服了传统分类方法中特征难以选择的不足。基于大量样本对模型进行训练,采用200张多缺陷样本进行测试。试验结果表明：检测的召回率为94%,检测的正确率为99%,分类的准确率为97.9%。试验验证了该算法可以满足木材表面缺陷的定位与分类要求。

关键词: 智能下料, 木材表面缺陷, Adaboost, 局部二值模式, 卷积神经网络

Abstract: In order to obtain the location information and category information of wood surface defects in the early stage of intelligent wood cutting, a combined algorithm for fast identification and location of wood surface defects is proposed based on digital image processing technology for fast recognition and location of wood surface defects. The Adaboost cascade classifier algorithm is used to extract the candidate frames of the wood surface defect region in the image in order to solve the problem that the traditional segmentation method is not effective for multi objects processing. A CNN model with self-learning ability is used to classify the input candidate boxes, so as to overcome the difficulty of selecting features in traditional classification methods. 200 multi-target samples are selected for testing, and the results show that the recall rate is 94%, the accuracy of detection is 99%, and the accuracy of classification is 97.9%. The experimental results show that the algorithm can meet the requirements of localization and classification of wood surface defects.

Key words: Intelligent cutting, Wood surface defect, Adaboost, LBP(Local Binary Patterns), CNN(Convolutional Neural Networks)

中图分类号:

TP391.41

王红军, 黎邹邹, 邹湘军. 基于Adaboost与CNN的木材表面缺陷检测[J]. 系统仿真学报, 2019, 31(8): 1636-1645.

Wang Hongjun, Li Zouzou, Zou Xiangjun. Wood Surface Defect Detection Based on Adaboost and CNN[J]. Journal of System Simulation, 2019, 31(8): 1636-1645.

参考文献

[1] 戚大伟, 牟洪波. 基于Hu不变矩和BP神经网络的木材缺陷检测[J]. 东南大学学报(自然科学版), 2013, 43(增1): 63-66.
Qi Dawei, Mu Hongbo.Detection of Wood Defects Types Based on Hu invariant Moments and BP neural network[J]. Journal of Southeast University (Natural Science Edition), 2013, 43(S1): 63-66.
[2] 杨慧敏, 王立海. 木材缺陷与超声检测参数相关性及影响因素[J]. 东北林业大学学报, 2015, 43(8): 114-116.
Yang Huimin, Wang Lihai.Correlation and Influencing Factors between Wood Defect and Ultrasonic Propagation Parameters[J]. Journal of Northeast Forestry University, 2015, 43(8): 114-116.
[3] 邵家鑫, 都东, 石涵, 等. 基于厚壁工件X射线实时成像的焊缝缺陷自动检测[J]. 清华大学学报(自然科学版), 2013, 53(2): 150-154.
Shao Jiaxin, Du Dong, Shi Han, et al.Aotumatic Weld Defect Detection Based on X-ray Images of Thick-wall Workpieces[J]. Journal of Tsinghua University(Science and Technology), 2013, 53(2): 150-154.
[4] 苏文浩, 刘贵珊, 何建国, 等. 高光谱图像技术结合图像处理方法检测马铃薯外部缺陷[J]. 浙江大学学报(农业与生命科学版), 2014, 40(2): 188-196.
Su wenhao, Liu Guishan, He Jianguo, et al. Detestion of External Defects on Potatoes by Hyperspectral Imaging Technology and Processing[J]. Journal of Zhejiang University (Agriculture and Life Sciences), 2014, 40(2): 188-196.
[5] 戴天虹, 邱筱斐. 基于形态学的木材缺陷检测[J]. 机电产品开发与创新, 2011, 24(5): 79-81.
Dai Tianhong, Qiu Xiaofei.Wood Defect Based on Mathematical Morpholog[J]. Development & Innovation of Machiner & Electrical Products, 2011, 24(5): 79-81.
[6] 吴思远, 张召, 邹洋. 基于2D Gabor小波和HSV空间的木材缺陷检测[J]. 郑州大学学报(理学版), 2010, 42(1): 93-97.
Wu Siyuan, Zhang Zhao, Zou Yang.Timber Defects Detection Dased on 2D Gabor Davelet and HSV Color Space[J]. Journal of Zhengzhou University(Natural Science Edition), 2010, 42(1): 93-97.
[7] Freund Y, Schapire R E.A Decision-Theoretic Generalization of On-Line Learning and an Application to Boosting[J]. Journal of Computer and System Sciences (S0022-0000), 1999, 55(1):119-139.
[8] 左登宇. 基于Adaboost算法的人脸检测研究[D]. 合肥: 中国科学技术大学, 2009.
Zuo Dengyu.A Study of Face Detection Based on Adaboost Algorithm[D]. Hefei: University of Science and Technology of China, 2009.
[9] 王海, 蔡英凤, 袁朝春. 基于多模式弱分类器的AdaBoost-Bagging车辆检测算法[J]. 交通运输工程学报, 2015, 15(2): 118-126.
Wang Hai, Cai Yingfeng, Yuan Chaochun.AdaBoost-Bagging vehicle detection algorithm based on multi-mode weak classifier[J]. Journal of Traffic and Transportation Engineering, 2015, 15(2): 118-126.
[10] Krizhevsky A, Sutskever I, Hinton G E.ImageNet Classification with Deep Convolutional Neural Networks[C]. International Conference on Neural Information Processing Systems. Lake Tahoe, Nevada: Curran Associates Inc, 2012(1): 1097-1105.
[11] 陈耀丹, 王连明. 基于卷积神经网络的人脸识别方法[J]. 东北师大学报(自然科学版), 2016, 48(2): 70-76.
Chen Yaodan, Wang Lianming.Convolutional Neural Network for Face Recognition[J]. Journal of Northeast Normal University (Natural Science Edition), 2016, 48(2): 70-76.
[12] 刘琮, 许维胜, 吴启迪. 时空域深度卷积神经网络及其在行为识别上的应用[J]. 计算机科学, 2015, 42(7): 245-249.
Liu Cong, Xu Weisheng Wu Qidi. Spatiotemporal Convolutional Neural Networks and Its Application in Action Recognition[J]. Computer Science, 2015, 42(7): 245-249.
[13] 陈永光. 遗传算法及模糊控制理论在木板材优化下料系统中的应用[D]. 北京: 北京林业大学, 2003.
Chen Yongguang.The Application of Genetic Algorithm and Fuzzy Control Theory in Control System of Wood Optimization Cut[D]. Beijing: Beijing Forestry University, 2003.
[14] 龙伶敏. 基于Adaboost的人脸检测方法及眼睛定位算法研究[D]. 成都: 电子科技大学, 2008.
Long Lingmin.Research on Adaboost Based Face Detection Method and Eye Location Algorithm[D]. Chengdu: University of Electronic Science and technology of China, 2008.
[15] Ojala T, Harwood I.A Comparative Study of Texture Measures with Classification Based on Feature Distributions[J]. Pattern Recognition (S0031-3203), 1996, 29(1): 51-59.
[16] Ojala T, Pietikainen M, Maenpaa T.Multiresolution Gray-Scale and Rotation Invariant Texture Classfication with Local Binary Patterns[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence (S0162-8828), 2002, 24(7): 971-987.
[17] 周志华. 机器学习[M]. 北京: 清华大学出版社, 2016: 30-31.
Zhou Zhihua.Machine Learning[M]. Beijing: Tsinghua University Press, 2016: 30-31.
[18] Viola P, Jones M. Rapid Object Detection using a Boosted Cascade of Simple Features[C]. Computer Vision and Pattern Recognition, 2001. CVPR 2001. Proceedings of the 2001 IEEE Computer Society Conference on. Kauai, HI, USA: IEEE, 2001(1): 511-518
[19] 宋晓琳, 邬紫阳, 张伟伟. 基于阴影和类Haar特征的动态车辆检测[J]. 电子测量与仪器学报, 2015, 29(9): 1340-1347.
Song Xiaolin, Wu Ziyang, Zhang Weiwei.Dynamic vehicle detection based on shadow and Haar-like feature[J]. Journal of Electronic Measurement and Instrumentation, 2015, 29(9): 1340-1347.
[20] Dalal N, Triggs B. Histograms of oriented gradients for human detection[C]. Computer Vision and Pattern Recognition, 2005. CVPR 2005. IEEE Computer Society Conference on. San Diego IEEE, 2005(1): 886-893.
[21] 孙锐, 陈军, 高隽. 基于显著性检测与HOG-NMF特征的快速行人检测方法[J]. 电子与信息学报, 2013, 35(8): 1921-1926.
Sun Rui, Chen Jun, Gao Jun.Fast Pedestrain Detection Based on Saliency Detection and HOG-NMF Features[J]. Journal of Electronics & Information Technology, 2013, 35(8): 1921-1926.
[22] Ioffe S, Szegedy C. Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift[C]. International Conference on Machine Learning,2015. ICML 2015. Lille, France, 2015(37):448-456.
[23] 高学, 王有旺. 基于CNN和随机弹性形变的相似手写汉字识别[J]. 华南理工大学学报, 2014, 42(1): 72-76,83.
Gao Xue, Wang Youwang.Recognition of Similar Handwritten Chinese Characters Based on CNN and Random Elastic Deformation[J]. Journal of South China University of Technology, 2014, 42(1): 72-76,83.
[24] 吴东洋, 业宁, 苏小青. 基于灰度共生矩阵和聚类方法的木材缺陷识别[J]. 计算机与数字工程, 2010, 38(11): 38-41.
Wu Dongyang, Ye Ning, Su Xiaoqing.Wood Defect Recognition Based on GLCM and Clustering Algorithm[J]. Computer & Digital Engineering, 2010, 38(11): 38-41.