Height Extraction Of Urban Buildings Based On High Resolution SAR Images

The city is an important place for human beings to engage in production activities.The acquisition and renewal of building information in urban areas by remote sensing technology will provide basic information for urban development planning and natural disaster monitoring.With the rapid development of remote sensing technology,there are more and more ways to obtain the height information of urban buildings.However,all kinds of technologies have limitations in practical application and productivity.Interferometric Synthetic Aperture Radar(InSAR)is an active earth observation technology developed in the middle and late 20 th century.It uses the phase information contained in SAR images to obtain the elevation of ground targets.Because of its advantages of all-weather and all-time work,InSAR is widely used.With the launch and operation of a new generation of high resolution SAR satellites,InSAR technology provides new support for urban services.Therefore,it is of great significance to study how to use high resolution SAR images to acquire elevation information of building targets.This paper mainly carries out the following work:1.For the problem of image factors such as uneven gray value and noise in SAR images,this paper uses median filtering and Gaussian filtering to preprocess the image respectively,which makes the gray value of some buildings in SAR images worthwhile to be enhanced and smoothed in some extent.Based on the difference of the grayscale value between the building target and the background,the detection of the building is realized by a target detection method based on the gray value information.Experiments show that compared with the median filtering technique,the integrity of the building detection is better than the image is processed by Gaussian filtering.2.For the problem that the branchcut is easy to produce more closed residual trees by connecting the branch tangent with the residues in the areas where the residues are more dense,which leads to the problem that the branch cut can not be unwrapped.This paper improves a method of residual identification optimization,and combined with the nearest neighbor connection method,to optimize the identification method of original branch cut,which reduces the total number of residues and unbalanced residues.The problem that the generated branch cuts is easy to form a residues tree in the dense residues point region and can not be unwrapped is solved.Finally,the experimental results show that the unwrapping method has certain advantages over the original algorithm.3.Using the edge of the detected building and generating the digital elevation model to extract the elevation information of the building,and taking the height value of the actual building as the reference standard.Compare the height values of the buildings obtained by the original branchcut method and the unwrapping method proposed in this paper.The results show that the accuracy of building elevation obtained by method in this paperis higher than that obtained by the original method.Finally,the elevation results obtained by the two methods are evaluated,and the factors affecting the accuracy of building elevation extraction are analyzed.