基于图像识别的液压支架护帮板收回状态监测方法
Monitoring method of recovery state of hydraulic support guard plate based on image recognition
【索引】王渊,李红卫,郭卫,等.基于图像识别的液压支架护帮板收回状态监测方法[J].工矿qy288千亿国际,2019,45(2):47-53.
【Reference】WANG Yuan,LI Hongwei,GUO Wei,et al.Monitoring method of recovery state of hydraulic support guard plate based on image recognition[J].Industry and Mine Automation,2019,45(2):47-53.
【DOI】10.13272/j.issn.1671-251x.2018070037
【作者】王渊1,李红卫1,郭卫1,贺海涛2,郏高祥1
【Author】 WANG Yuan1,LI Hongwei1,GUO Wei1,HE Haitao2,JIA Gaoxiang1
【作者机构】1.西安科技大学 机械工程学院, 陕西 西安710054;2.中国神华能源股份有限公司 神东煤炭分公司, 陕西 神木719315
【Unit】1.College of Mechanical Engineering, Xi'an University of Science and Technology, Xi'an 710054, China;2.Shendong Coal Branch, China Shenhua Energy Company Limited, Shenmu 719315, China
【摘要】针对现有接触式液压支架护帮板状态监测方法在矿井雾尘环境下应用存在故障率高、测量结果容易受机身倾斜等因素影响等问题,提出了一种基于图像识别的液压支架护帮板收回状态监测方法。该方法利用雾尘图像清晰化算法与机器视觉测量方法对液压支架护帮板的收回角度进行监测,通过测量护帮板角度来确定液压支架护帮板的收回状态。首先采用改进的暗通道先验算法与导向滤波多尺度Retinex算法对采集的图像进行去雾处理,对经去雾处理的图像进行小波融合,着重恢复雾尘图像的边缘细节信息;然后利用机器视觉测量方法对融合图像的感兴趣区域(ROI)进行提取、二值化、水平和垂直投影处理,提取骨架、骨架像素点,拟合生成直线,由已标定好的CCD相机进行坐标变换,输出护帮板真实角度,进而判断护帮板是否收回。实验结果表明,该方法实现了煤矿井下雾尘图像的清晰化处理,保留了图像细节,且测量精确度高,综合误差小于2°,满足对护帮板的监测要求。
【Abstract】In view of problems of high failure rate and easy to be affected by incline of shearer in the application of contact-type monitoring method of hydraulic support in environment of mine fog and dust, a monitoring method of recovery stae of hydraulic support guard plate based on image recognition was proposed. The method uses fog dust image sharpening algorithm and machine vision measurement method to carry out monitoring of recovery angle of the guard plate of the hydraulic support, and determines the recovery state of the guard plate of the hydraulic support by measuring the angle of the guard plate. Firstly, an improved dark channel prior algorithm and a multi-scale Retinex algorithm with guided filtering are adopted to defog the captured image, and then wavelet fusion is carried out on the defogging image, focusing on restoring the edge details of the image of fog and dust. Then, the region of interest (ROI) of the fusion image is extracted, binarized and processed by horizontal and vertical projection with machine vision measurement method, the skeleton and skeleton pixel points are extracted and generated into straight lines by fitting, coordinate transformation is carried out by the calibrated CCD camera to output true angle of the guard plate, so as to judge whether the guard plate is recovered. The experimental results show that the method realizes sharpening process of images with fog and dust in underground coal mine, and keep the detail of the image, and has accurate measurement result, and the synthetic error is less than 2°, which meets monitoring requirements for the guard plate.
【关键词】 煤炭开采; 液压支架; 护帮板收回状态; 非接触监测; 图像识别; 图像融合; 图像去雾; 机器视觉; 多尺度Retinex
【Keywords】coal mining; hydraulic support; recovery state of guard plate; non-contact monitoring; image recognition; image fusion; image defogging; machine vision; multi-scale Retinex
【文献出处】工矿qy288千亿国际,2019年2期
【基金】国家重点研发计划子课题资助项目(2017YFC0804310)
【分类号】TD421
本网站仅提供本刊2009年之后的全文下载,其它年份的全文下载将自动转到中国知网。中国知网不支持迅雷等加速下载工具,请取消加速工具后下载。
【关 闭】