Study on Key Technology of Earthquake Emergency Using Multi-Mode Remote Sensing Data

Author:Li Jiang

Supervisor:Zhang Jingfa

Database:Doctor

Degree Year:2018

Download:302

Pages:153

Size:17434K

Keyword:

Earthquake has the characteristics of wide scope of damage,strong suddenness and large deformation.The rapid assessment and the initiation of emergency response after the earthquake can effectively reduce the damage caused by the earthquake.The features of macroscopic,rapid and extensive coverage of space-to-Earth observation technology provide an effective way for disaster information acquisition and play an important role in earthquake emergency.With the continuous development of earth observation space-to-earth observation technology,the types of images acquired after the earthquake have been increasing,showing obvious "big data" characteristics.However,although image acquisition capability has been continuously improved,seismic remote sensing information processing capacity is insufficient,and collaborative analysis methods for multi-source image data are lacking,which results in low degree of useful information mining from massive data.There has been a serious imbalance between the capability of image acquisition and information processing capacity.At present,the earthquake damage extraction technology based on remote sensing images can no longer meet the timeliness and accuracy requirements of earthquake emergency.This thesis focuses on the seismic information identification and disaster dynamic change detection using the multi-modal remote sensing image before and after an earthquake,and explores the intelligent and automated multi-model remote sensing data collaboration and earthquake emergency applications.The main contents of this paper are as follows:(1)In view of the situation that post-earthquake remote sensing images exists in the earthquake area,a damage information extraction method based on data mining and object-oriented is proposed.The method can realize the automatic construction of feature rules and avoid the lack of feature deletion caused by artificial selection.The CNN deep learning process and technical method is proposed.The method fills in the deficiency of human intervention in feature selection in data mining algorithm,and significantly improves the accuracy of seismic damage extraction.(2)A multi-source heterogeneous remote sensing image change detection method is proposed,which is suitable for multi-temporal remote sensing images before and after an earthquake.When the images before and after the earthquakes are homologous,a technical process based on the correlation change detection of principal components of multi-texture features is proposed.It can make full use of the rich texture features in the SAR image while avoiding redundancy of the feature information.When the data obtained before and after the earthquake is heterogeneous,a post-classification change detection technique based on object-oriented and CNN model is proposed.The method overcomes the requirement data type and time consistency,and realizes multi-sensor data assimilation and information collaborative processing.(3)A mathematical model for the dynamic monitoring and trend analysis of earthquake secondary disasters was constructed and applied to the coexistence of multi-mode data after earthquakes.The spatial distribution of secondary disasters in different time periods after the earthquake was extracted,and a mathematical model of the area,amount,and time of disasters was constructed to realize the dynamic evolution prediction of disasters.PSO-BP Neural network prediction model is constructed.Combined with the impact factors such as pregnancy disaster,catastrophe,and topography in the secondary disasters in the earthquake area,the prediction of landslide risk in the earthquake area is realized.(4)Aiming at the chaos and low efficiency of remote sensing emergency mode after earthquake,a framework of remote sensing earthquake was proposed.The tasks and requirements of different phases of the earthquake were analyzed in depth.In addition,the method of identifying earthquake damage information and the requirement of timeliness are planned after obtaining different modes of remote sensing images at different stages.The platform of spatial data collaboration for earthquake emergency was built to improve the efficiency of earthquake emergency response.The framework can standardize the application mode of remote sensing emergency workflow in future earthquakes.