The Technology Study Of Buildings’ Change Detection Based On Multi-temporal Remote Sensing Images

Change detection is according to comparation and analysis multi-tempotal remotesensing images. It’s the process of finding、indrtify and obtain the changed information.It’s a hot topic in the field of the remote sensing in recent years. It is widely used inmedicine、military and civil fields etc. Buildings are important man-made featureselements in the general images. Fetching building information from remote sensingimages is a hot topic in the field of remote sensing in recent years. It would be wise tostudy buildings’ change detection based on multi-temporal remote sensing images. Inthe aspect of practice, the feature points in images of the buildings from changedetection could be control points for revision of topographic map. At the same time, itcould be meeting the need of GIS data updating. In the aspect of theoretical, it couldprovide theoretical guidance an methodology guidance for detection of changedinformation from other objects in the images.First, the man-made features is the main source of information among thechanges in the geographical information. The buildings and roads are typical man-madeobjects, they have wide varieties and diffirent forms. In the urban area, the man-madefeatures is compact and large area distribution. Among them buildings occupy a largeproportion, and the region is flat. The change detection in order to get the changeinformation of buildings is particularly important; Generally speaking, The features ofbuildings in images are obvious, the feature points could be survey control points, It’lldirectly affects its level of automation; At last, the basic shapes of buildings arerectangular and the materials of buildings are relative stability. Therefore, they havehigher distinguishing degree in images and lower difficulty for study. The success inchange detection of buildings would be theoretical guidance for change detection ofother complex surface features.Based on the background aforesaid, The mainly studies of this paper is using themethod of change detection to get the change information of buildings based onmultitemporal DMC images. Because of the images of the roof on buildings arerectangular or combination of rectangles,therefore a extraction method is proposedwhich based on linear features for the information of the roof on buildings.Throughcomparing the detection results of building information in the two images before andafter,it used the method of change detection to get the change information ofbuildings.Through the experimental results,the results of the method are not ideal.Inthe extraction process of building information,only consider the information of rectangular edge can not effectively eliminate information of rectangular green in theimages.As a result,the results of information for the green change to buildings is veryfuzzy,and it can not accurately detect.In order to effectively eliminate the rectangular non building information, takinginto account the building materials of the roof on buildings are relatively simple,and inaddition to the obvious spatial characteristic of building image,the gray texture andother characteristic are also unique.Comprehensive consider the image features ofbuildings,finally proposed on extraction method based on edge of the zone of buildinginformation.This method first uses Mean Shift algorithm to multi-scale segmentationthe information of original images,then using the Otsu algorithm to thresholdsegmentation based on the results.This will remove the non buildinginformation.Next,it extract the information of the roof on buildings using the extractiontechnology of building information based on linear feature.At last,it detect the changeinformation of buildings according to compare the two kinds of results.Theexperimental results show that the method can detect the change information ofbuildings relatively accurate.