基于监控视频的油田施工场景状态识别  

State Recognition of Oilfield Construction Scene Based on Monitoring Video

作　　者：段鸿杰[1] 计洪图 李晓宇 梁琦[1] DUAN Hong-jie;JI Hong-tu;LI Xiao-yu;LIANG Qi(Control Center of Informatization of Sinopec Shengli Oil Field,Dongying 257000 China;Beijing Key Lab of Petroleum Data Mining(Beijing),Beijing 102249 China;China University of Petroleum(Beijing),Beijing 102249 China)

机构地区：[1]中国石化胜利油田分公司信息化管理中心,山东东营257000 [2]石油数据挖掘北京市重点实验室,北京102249 [3]中国石油大学(北京),北京102249

出　　处：《自动化技术与应用》2020年第8期80-85,共6页Techniques of Automation and Applications

摘　　要：场景状态识别作为监控视频内容分析的重要内容,对保障油田监控区域的安全发挥极为重要的作用。本文以胜利油田的监控视频数据为例,对监控视频图像进行图像预处理后,运用灰度梯度共生矩阵和HOG特征提取的方法分类并建模,进而识别监控视频状态。经测试,异常场景识别效果良好。Scene state recognition, as an important content of monitoring video content analysis, plays an extremely important role in ensuring the safety of oilfield monitoring area. This paper takes the monitoring video data of Shengli Oilfield as an example. After image preprocessing, the monitoring video image is classified and modeled by gray gradient co-occurrence matrix and HOG feature extraction method, and then the monitoring video status is recognized. After testing, the effect of anomaly scene recognition is good.

关 键 词：监控视频 图像处理 特征提取 施工场景 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]