Research On Adaptability Evaluation Method Based On Target Area Building Change Detection

With the rapid development of aerospace technology,high-resolution satellite images have been applied to all aspects of the military field,not only in national security,military situation monitoring,inertial guidance and other aspects of prevention,but also in the identification,positioning,tracking of military targets,Military early warning and other informatized military confrontations also play a key role.This research is aimed at the detection of changes in high-resolution satellite image building target strikes in the military field.On the basis of comparing the characteristics of the current commonly used change detection methods,this paper proposes a project that takes engineering as the core and meets the actual application needs of the military field.The technical solution for building change detection and the adaptability of the reference map after the damage occurred in the target area.The thesis mainly completed the following work:In order to improve the adaptability of the change detection technology,a highresolution satellite image building recognition algorithm based on the combination of enhanced morphological building index and fully connected condition random field is proposed.Under the premise of using the enhanced morphological building index to detect,the method further improves the detection results by using the full connection random field,realizes the automatic detection and recognition of buildings based on the object-based highresolution satellite image,effectively solves the problem of difficult segmentation such as road area,and overcomes the problem of difficult detection of buildings with different shapes,greatly improving the accuracy of detection.In order to improve the accuracy of the change detection technology,a multi-temporal high-resolution satellite image change detection algorithm based on multi-feature fusion and Markov random field fusion is proposed.Based on the traditional multi-feature fusion change detection method based on threshold selection,this method implements adaptive threshold selection based on the analysis of feature heterogeneity histogram curvature,and uses the Markov random field model to introduce the space between building objects Context information enhances the spatial connection between adjacent buildings,and improves the accuracy and robustness of the detection results of building targets in complex scenes.In order to ensure the precise positioning guidance of the aircraft,aiming at the adaptability of the benchmark image after the strike,a series of adaptability evaluation indexes are proposed from the perspective of practical application,and a model of the adaptability evaluation method of the benchmark image based on the damage of the target area is established.The verification results of the model show that the adaptability evaluation method in this paper has a certain reference value in evaluating the adaptability of the reference map of the damaged target area.