边缘扩展的皮带撕裂支持向量机视觉检测

摘要/Abstract

摘要： 提出了基于视觉的皮带撕裂监测方法，并构建了皮带撕裂视觉监控系统。针对皮带输送机运行过程中由于干扰导致的图像退化，采用维纳滤波方法实现了退化图像的复原。为实时识别高速运动的皮带裂纹，采用CamShift算法对快速移动的皮带裂纹序列目标图像进行跟踪与捕捉。采用Canny算子对皮带裂纹进行边缘提取，并通过增加一个δ值，使检测到的裂纹边缘向外扩张，从而增加检测到的皮带裂纹权重，获得鲁棒性更高的边缘检测效果。最后，构建了SVM皮带裂纹预报模型，以皮带裂纹图像的像素面积及长宽比几何特征量作为模型输入量，对皮带裂纹状态进行预报。实验表明，提出的皮带撕裂检测方法是有效的。

关键词: 皮带撕裂, 图像分割, Canny边缘提取, 支持向量机, 裂纹识别

Abstract: A belt safety monitoring method was proposed based on vision, and a visual monitoring system for belt tearing was constructed. Aiming at the image degradation from interference during operations of belt conveyor, Wiener filtering method was used to restore the degraded images. In order to recognize the belt cracks with high-speed moving in real time, CamShift algorithm was used to track and capture the targets of fast-moving sequence images of belt cracks. Canny operator was used to extract the edges of belt cracks, and the detected edges of belt cracks were expented outwards by adding a value δ, increasing the weights of the detected cracks, and more robust edge detection results were obtained. Finally, belt crack prediction model was constructed based on SVM, geometric characteristics, such as pixel area and length width ratio of the belt crack images were taken as the model inputs to predict belt crack states. Experimental results show the effectiveness of the method of belt tearing detection method proposed herein.

Key words: belt tearing, image segmentation, Canny edge extraction； support vector machine(SVM), crack recognization

中图分类号:

TP242.3

王福斌1, 2;孙海洋1;TU Paul2. 边缘扩展的皮带撕裂支持向量机视觉检测[J]. 中国机械工程.

WANG Fubin1,2;SUN Haiyang1;TU Paul 2. Visual Inspection for Extended Edge Belt Tearing Based on SVM[J]. China Mechanical Engineering.

参考文献

[1]卢金龙. 基于机器视觉的皮带撕裂检测系统设计与实现[D]. 秦皇岛：燕山大学, 2016.
LU Jinlong.Design and Implementation for Belt Tearing Detection System Based on Machine Vision[D]. Qinhuangdao：Yanshan University, 2016.
[2]张明敏. 基于机器视觉的矿用输送带纵向撕裂检测系统研究[D]. 北京：中国矿业大学(北京), 2014.
ZHANG Mingmin. Research for Longitudinal Tear Detection System of Mine Conveying Based on Machine Vision [D]. Beijing ：China University of Mining & Technology, Beijing, 2014.
[3]LEE J W, YOON J S. Visual Inspection System for Irregularly Formed Timing Belt with Low Reflection Ratio[J]. 2012, 13(5):1996-2001.
[4]ANDREJIOVA M, GRINCOVA A, MARASOVA D, et al. Using Logistic Regression in Tracing the Significance of Rubber-Textile Conveyor Belt Damage[J]. Wear, 2014, 318 (1/2): 145-152.
[5]MOLNR V, FEDORKO G, STEHLKOV B, et al. A Regression Model for Prediction of Pipe Conveyor Belt Contact Forces on Idler Rolls[J]. Measurement, 2013, 46(10):3910-3917.
[6]MOLNR V, FEDORKO G, STEHLKOV B, et al. Analysis of Asymmetrical Effect of Tension Forces in Conveyor Belt on the Idler Roll Contact Forces in the Idler Housing[J]. Measurement, 2014, 52(1):22-32.
[7]LI W, WANG Z W, ZHU Z C, et al. Design of Online Monitoring and Fault Diagnosis System for Belt Conveyors Based on Wavelet Packet Decomposition and Support Vector Machine[J]. Advances in Mechanical Engineering, 2013, 5: 1-10.
[8]LI M, DU B J, ZHU M Q, et al. Intelligent Detection System for Mine Belt Tearing Based on Machine Vision[C]// 2011 Chinese Control and Decision Conference (CCDC). Mianyang: 1250-1253.
[9]LI J, MIAO C Y. The Conveyor Belt Longitudinal Tear On-line Detection Based on Improved SSR Algorithm[J]. Optik, 2016, 127 (19):8002-8010.
[10]HUNG K W, SIU W C. Single-image Super-resolution Using Iterative Wiener Filter Based on Nonlocal Means[J]. Signal Processing: Image Communication, 2015,39 :26-45.
[11]BISWAS P, SARKAR A S, MYNUDDIN M. Deblurring Images Using a Wiener Filter[J]. International Journal of Computer Applications, 2015, 109(7): 36-38.
[12]PHAM T D. Estimating Parameters of Optimal Average and Adaptive Wiener Filters for Image Restoration with Sequential Gaussian Simulation[J]. IEEE Signal Processing Letters, 2015,22(11): 1950-1954.
[13]KIM G W, KANG D S. Improved CamShift Algorithm Based on Kalman Filter[J]. Advanced Science and Technology Letters, 2015,98: 135-137.
[14]HSIA K H, LIEN S F, SU J P. Moving Target Tracking Based on CamShift Approach and Kalman Filter[J]. Applied Mathematics & Information Sciences, 2013,7(1): 193-200.
[15]SALHI A, JAMMOUSSI A Y. Object Tracking System Using CamShift, Meanshift and Kalman Filter[J]. Electronics and Communication Engineering, 2012,6(4): 421-426.
[16]SHRIVAKSHAN G T, CHANDRASEKA C. A Comparison of Various Edge Detection Techniques Used in Image Processing[J]. International Journal of Computer Science Issues, 2012, 9(1): 269-276.
[17]CHEN Y, XU M, LIU H L, et al. An Improved Image Mosaic Based on Canny Edge and an 18-dimensional Descriptor[J]. Optik, 2014,125(17): 4745-4750.
[18]DENG C X, WANG G B, YANG X R. Image Edge Detection Algorithm Based on Improved Canny Operator[C]// Proceedings of the 2013 International Conference on Wavelet Analysis and Pattern Recognition. Tianjin, 2013: 168-172.
[19]DI H B, GAO D L. Gray-level Transformation and Canny Edge Detection for 3D Seismic Discontinuity Enhancement[J]. Computers & Geosciences, 2014,72: 192-200.
[20]何青，褚东亮，毛新华. 基于EEMD和MFFOA-SVM滚动轴承故障诊断[J].中国机械工程, 2016, 27(9):1191-1197.
HEQing, CHU Dongliang, MAO Xinhua. Study on Rolling Bearing Fault Diagnosis Based on EEMD and MFFOA SVM[J]. China Mechanical Engineering, 2016, 27(9):1191-1197.
[21]方一鸣, 胡春洋, 刘乐,等. 基于主动学习GA-SVM分类器的连铸漏钢预报[J]. 中国机械工程, 2016, 27(12):1609-1614.
FANG Yiming，HU Chunyang，LIU Le, et al. Breakout Prediction Classifier for Continuous Casting Based on Active Learning GA-SVM[J]. China Mechanical Engineering, 2016, 27(12):1609-1614.