Study On SAR Image Fusion And Change Detection Based On Mathematical Frame Theory

Synthesis Aperture Radar(SAR)imaging technology can detect and image ground targets under severe climatic conditions,and is widely used in disaster monitoring and emergency rescue.However,in the SAR imaging process,due to the influence of external environmental interference,such as system equipment constraints and data processing,the edge texture of the acquired SAR image is blurred,and the target features are not obvious,which in turn affects and limits the interpretation and application of SAR images.The SAR image change detection technology can extract the change information of the ground scene by detecting the difference of different time-phase SAR images,and thus is widely used in the disaster area identification and infrastructure construction progress monitoring.However,in SAR image change detection,the traditional single type difference map has its own inherent disadvantages,which is impossible to accurately detect the change region while suppressing the noise in the background region.The frame is a set of vectors with upper and lower bounds of energy in the Hilbert space.In practice,many signal and image processing problems are inherently frame problems.In this paper,the mathematical frame theory is introduced into the SAR image sequence fusion reconstruction and SAR image change detection,and the fineness of the model representation is improved by the frame to improve the quality of image fusion.Aiming at the problem of SAR image blur degradation,a frame-based SAR image sequence fusion reconstruction method is proposed.This method mines the complementary information of low-resolution SAR image sequences by fusion preprocessing,and obtains high-resolution(referred to as grid precision)fuzzy SAR.The image is used as the initial image of the subsequent reconstruction.The prior knowledge of the point spread function of the SAR imaging system is used to construct the frame description of the fuzzy operator of the imaging system.Based on this,the initial estimation image is performed by the iterative regularization method.Iterative reconstruction provides high quality SAR images with information complementation and feature enhancement.The proposed method is used to reconstruct the actual SAR images.The subjective images and the objective evaluation index show that the proposed method effectively improves the sharpness of the target edge texture,improves the contrast between the target and the background,and enhances the target feature.Aiming at the problem of SAR image change detection,the single type difference image has its own inherent shortcomings.The SAR image change detection method based on wavelet frame fusion is studied.In this method,the wavelet frame decomposition decomposes the mean ratio difference image and the log ratio difference image into high-frequency and low-frequency sub-images without down-sampling,and then uses local energy minimum strategy and mean strategy to fuse high-frequency sub-images and low-frequency sub-images respectively.The fusion low-frequency sub-image and high-frequency sub-image are reconstructed by wavelet frame to generate a fusion difference map.The FLICM clustering method is applied to the fusion difference map for unsupervised classification,and the final change detection binary result image is obtained.The detection results of the measured SAR images show that the SAR image change detection method based on wavelet frame fusion can combine the advantages of the two difference images,and accurately detect the changes while suppressing the speckle noise in the background area.The method implemented in this thesis can be used for the safety hazard investigation of major infrastructure along the “Belt and Road” in China,macro construction progress monitoring,and natural disaster monitoring and emergency rescue.It is of great significance to ensure the safety of construction personnel along the “Belt and Road” and avoid major economic losses.