Study Of Damage Effect Assessment Based On Multi-temporal Image

Damage effect assessment technology is widely used and plays a decisive role in informationizing modern war.In modern war,accurate acquisition of battlefield damage information plays an important role in the selection of military targets,battlefield situation monitoring,inertial navigation guidance and even strategic decision-making;the results of damage effect assessment have become an important reference basis for the military to make various military decisions.In this paper,the damage effect assessment technology is studied relatively at home and abroad,and the damage effect assessment process based on multitemporal image and the relative algorithms are established.The main contents of this paper are as follows:A set of image preprocessing related methods,including relative radiation correction algorithm based on wavelet transform and image registration method based on SIFT and RANSAC feature,are used to correct and register the subsequent change detection part.A change detection algorithm based on the combination of coarse-grained and redetection of image change is proposed.The purpose of the coarse-grained stage is to provide accurate samples for the fine-grained stage.In the coarse-grained stage,Gray Level Cooccurrence Matrix,which is short for GLCM,and its eigenvector are calculated for the input image,texture feature difference image is obtained by the eigenvector,then three types of regions including changed,unchanged and unknown regions are obtained by a kind of improved FCM clustering.In the fine-grained stage,in order to get the changes of the unknown regions in the three types of regions,a change detection algorithm based on CNN model is designed.The positive and negative samples are made by using the changed regions and the unchanged regions obtained in the coarse-grained stage,and the positive and negative samples are used to train the CNN model.After that,the images before and after the attack are processed by using the trained CNN model for image binary classification,then the binary result image of change detection is obtained.Compared with the direct FCM classification method,it avoids the false detection and miss detection caused by threshold selection.Compared with PCA and other algorithms,it can effectively filter out the pseudo-change areas in traditional methods and improve the accuracy of change detection results.To ensure the accuracy of the military’s decision-making on strike,a damage evaluation method based on image change detection is proposed.According to the different methods of damage assessment of various targets,including ships,buildings and airports,the expert knowledge base is established by using the prior model base.Using the Analytic Hierarchy Process(AHP)to weight different characteristic parameters of various targets in the knowledge base according to certain rules.The damage evaluation is based on geometry,texture and functional damage evaluation criteria.The results of the experiments show that the damage assessment method proposed in this paper has great accuracy.Meanwhile,compared with the expert opinions,the evaluation results obtained by this method have higher consistency with the expert opinions.