摘要
分析了矿井水灾视频图像特征,提出了基于视频图像的矿井水灾识别及趋势预测方法,包括水灾视频动态识别、区域分割、面积估算及趋势预测,并通过了试验验证,得出如下主要结论:①阈值像素灰度统计法和像素灰度值统计法均可监测和识别水灾,阈值像素灰度统计法不但可抑制低于灰度阈值的噪声,提高识别的准确性,还可减少像素灰度统计数,增强特定像素灰度范围的对比度。②阈值分割法和视频差分分割法均可分割水灾区域图像,前者整体性较好,后者细节刻画更强。③根据分割出的水灾区域图像可估算突水区域面积及进行趋势预测。
The characteristics of mine flood video images were analyzed. The mine flood identification and trend prediction methods based on video images were proposed, including flood video dynamic identification, region segmentation, area estimation and trend prediction. The results were verified by experiments. The main conclusions are as follows:① Both threshold pixel grayscale statistical method and pixel grayscale statistical method can monitor and identify floods. The threshold pixel grayscale statistical method not only can suppresses noise below the grayscale threshold and improve the accuracy of recognition, but also can reduce the pixel grayscale statistics, enhance contrast of a particular pixel grayscale range.② Both the threshold segmentation method and the video differential segmentation method can segment the image of the flood area, the former is better overall and the latter is more detailed.③ The area of the water inrush area can be estimated and the trend can be forecast based on the segmented flood area image.
作者
孙继平
靳春海
曹玉超
SUN Jiping;JIN Chunhai;CAO Yuchao(China University of Mining and Technology(Beijing),Beijing 100083,China)
出处
《工矿自动化》
北大核心
2019年第7期1-4,16,共5页
Journal Of Mine Automation
基金
国家重点研发计划资助项目(2016YFC0801800)
关键词
矿井水灾
图像识别
水灾区域分割
面积估算
趋势预测
mine flood
image recognition
flood area segmentation
area estimation
trend prediction
作者简介
孙继平(1958-),男,山西翼城人,教授,博士,博士研究生导师,中国矿业大学(北京)信息工程研究所所长;获国家科技进步二等奖3项(第1完成人2项);作为第1完成人获省部级科技进步特等奖和一等奖8项;作为第1完成人主持制定中华人民共和国煤炭行业和安全生产行业标准28项;主持制定《煤矿安全规程》第十一章“监控与通信”;作为第1作者或独立完成著作12部;被SCI和EI检索的第1作者或独立完成论文90余篇;作为第1发明人获国家授权发明专利60余项;作为国务院煤矿事故调查专家组组长参加了10起煤矿特别重大事故调查工作;E-mail:sjp@cumtb.edu.cn。
