Mobile Emergency Notification System Based on Geographical Information System

摘要

Multi-temporal synthetic aperture radar (SAR)images have been successfully used for the detection of differenttypes of terrain changes. However, SAR image change detectionbased on wavelet transform is still restrained from the existenceof speckle noise and the nature of wavelet transform. In thispaper, an unsupervised SAR image change detection fusionframework based on shearlet transform is proposed. In theproposed method, The Gauss filtering is combined with log-ratio to impair speckle. Then the difference map (DM) ofGauss-log ratio and the difference map of ratio based on Gaborfeature are fused with shearlet transform. Meanwhile, DM isdecomposed to low frequency image and four high frequencyimages, different fusion rules are used in multi-scales imagesrespectively, the work of noise reduction is operated with meanfiltering. After an inverse shearlet transformation, the finalchange map can be obtained via a simple OSTU segmentation.The real SAR image pairs in Bern area are used to verifyproposed change detection method. The experimental resultsdemonstrate the robustness of the proposed method.