摘要
为了实现输送带纵向撕裂的准确检测,解决单一特征鲁棒性低的问题,提出了一种双模板纵向匹配特征,同时采用Dempster-Shafer(DS)证据理论进行多特征融合检测。首先提取图像的灰度匹配和几何特征,然后利用DS证据理论加以融合;融合后根据组合决策规则判断图片撕裂与否。实验表明该算法在提高了检测的准确性的同时,满足了矿井下输送带撕裂检测对实时性的要求,可为及时报警停机提供可靠依据。
To solve the problem of low robustness with single feature and realize high accuracy detection of longitudinal tearing detection of belt, the paper proposed a longitudinal matching feature with double templates and detected with multi -feature fusion detection using Dempster - Sharer （DS） evidence theory. The gray matching feature and geometrical feature were extracted, which were then combined according to DS combining decision rule; the images were classified into tearing or off- tearing according to combining decision rule. The results of the experiment show that, the accuracy is improved, and the real - time monitoring is realized for conveyor belt tearing, creating conditions for timely alarming and stopping.
出处
《煤炭工程》
北大核心
2015年第12期103-106,共4页
Coal Engineering
基金
山西省特色学科资助项目(晋教财[2012]145号)
山西省高等学校留学回国人员科研项目(晋教外[2011]63号)
关键词
纵向撕裂
DS证据理论
模板匹配
鲁棒性
longitudinal tearing
Dempster- Sharer evidence theory
template matching
robustness
