基于SIFT特征与SVM分类的地震灾情图像信息异常检测方法*

(1.四川省地震局,四川 成都 610041; 2.武警警官学院,四川 成都 610213)

地震灾情图像信息; 异常检测; SIFT; SVM; 芦山地震

Detection Method of Earthquake Disaster Image Anomaly Based on SIFT Feature and SVM Classification
ZHANG Ying1,GUO Hongmei1,YIN Wengang2,ZHAO Zhen1,RAN Qing1

(1. Sichuan Earthquake Agency,Chengdu 610041,Sichuan,China)(2. College of Armed Police Officer,Chengdu 610213,Sichuan,China)

earthquake disaster image information; anomaly detection; SIFT; SVM; Lushan M7.0 earthquake as an example. The results show that the algorithm model has better detection effect on image information anomaly,it can further supply and improve the disaster in

备注

通过分析以往震后获取的图像信息发现,部分信息存在与地震发生时间不吻合、不属于地震影响范围或与地震灾情无关等异常。通过将图像分类算法运用到震后灾情图像信息的异常检测中,提出了一种基于SIFT特征与SVM分类的地震灾情图像信息异常检测模型,以2013年芦山7.0级地震建筑物破坏灾情图像为例对模型进行验证。结果 表明:该模型对图像信息异常的检测效果较好,可进一步补充和完善地震应急救援的灾情信息源,为政府抗震救灾科学决策提供灾情信息支撑。

Through analyzing the image information obtained after the earthquake,we found that some of the information does not match the time of the earthquake,does not belong to the earthquake influence field or has nothing to do with the earthquake disaster. In this paper,we applied the image classification algorithm to the detection of post-earthquake disaster image information anomaly,and proposed an earthquake disaster image information anomaly detection model based on SIFT feature and SVM classification,and verified the model by taking the damage image of the Lushan M7.0 earthquake as an example. The results show that the algorithm model has better detection effect on image information anomaly,it can further supply and improve the disaster information source for earthquake emergency rescue,and provide disaster information support for the government's scientific decision-making for earthquake relief. Key words:earthquake disaster image information; anomaly detection; SIFT; SVM; Lushan M7.0 earthquake as an example. The results show that the algorithm model has better detection effect on image information anomaly,it can further supply and improve the disaster information source for earthquake emergency rescue,and provide disaster information support for the government's scientific decision-making for earthquake relief.