Application Research On Mining Area Change Detection With Multisource SAR Images

China is a large energy country with abundant mineral resources,and the exploitation of resources will cause the destruction of the surface.In emergency disaster treatment how to quickly and accurately determine the changing area has great significance.Synthetic Aperture Radar(SAR)can obtain images in all-time and all-weather which play a key role in the changes of global resources and environment,assessment of disaster loss,land use,agricultural production,urban expansion,and forest monitoring.In recent years,with the rapid development of SAR technology,there are many kinds of satellites equipped with SAR sensors,and the types and resolutions of images data are various.The change detection by fusing different resolution images has been paid much attention by scholars.On the basis of the characteristics of different resolution SAR images,this thesis research on mining area change detection application based on multisource SAR images fusion.The main work and results are as follows:(1)In order to reduce the influence of speckle noise,7 filtering methods are used to deal with multisource SAR images.It is found that the denoising effect of low resolution and high resolution SAR images is better by mean and Lee filtering respectively.(2)According to the filtered SAR images,differential and logarithmic ratio methods are used to construct differential images respectively.Using the Kilter-Illingworth(KI)threshold segmentation and mathematical morphological post-processing,the change detection results are obtained.In the low resolution images,combining the difference image constructed by the differential approach and the 3×3 square shape structure,we can get a more accurate change map.In the high resolution image,using the logarithmic ratio method combined with the 3×3 square shape structure work better.(3)In order to retain the texture features of the images,the original image texture features are extracted by the Gray-level Co-occurrence Matrix(GLCM),and the variation threshold is extracted with the maximum between-cluster variance algorithm(Otsu).The mean texture features have the highest accuracy in the use of low resolution and high resolution images..(4)In view of the optimal change results extracted by two methods.First of all,the results of the homologous data are two values and the operation is fused.Then,the advantages of the large area and the high resolution image change detection results with the results of the low resolution image change detection are adopted to the results of the multisource data change based on the geographical coordinates registration.The fusion strategy is fused to get the final change detection result.It is found that the pore size in the large change area is reduced,and the edge and fine structure of the image change are maintained,and a more accurate change detection image is obtained.