Building Object-level 3D Change Detection Method Based On Oblique Photogrammetry

With the development of urbanization in China,the shape of urban buildings and spatial pattern of city are becoming more complex,and 3-D change detection has become an important part of geographic conditions monitoring and geographic information updating.At the same time,it can also analyze the urban planning with the help of larger scenes.The 3-D change detection and update of urban buildings are playing an increasingly important role in smart cities.In recent years,the aerial tilt photogrammetry system increasing mature,it can produce TIN model products with image resolution up to centimeters-level efficiently and rapidly.More cities have accumulated such high precision real-scene TIN model.The real scene TIN model can provide new data sources for urban3-D precise model updating and illegal buildings detection.However,in 3-D urban scenes,there are many kinds of ground objects,and change detection is only important for buildings.Therefore,this paper will focus on high-frequency,high-aging dynamic monitoring of urban buildings,use height-analysis method to extract the contour information of buildings,then cut total factor of the 3-D city model reconstructed by oblique images to obtain the contour features of ground objects at different heights.Since the contour features of ordinary buildings change little at the vertical height,while ground objects such as vegetation change greatly,it's possible to extract buildings by contour change rate,and preserve the multi-height information of the building,furthermore,we can detect the object-level change of 3-d model directly based on the building contour.The main researches in this paper are listed as follow:(1)Building contour extraction based on multi-height analysis.Using multi-height analysis method to extract total factor of TIN model under the 3-D scene.Then use constraints such as contour closure,area operator,aspect ratio operator and multi-height area difference to extract the contour of buildings from the contour set and fit out the minimum external rectangle(2)Building object level change detection considering contour features.Using overlay analysis method to the minimum enclosing rectangle from same target area but different periods,and detect the change of the area size and the displacement of the center point,different change values correspond to different change types,combine with theirs' heights to classifing and visualizing their change types.(3)Design and implement change detection prototype system.Design and develop a3-D urban real-scene model change detection platform prototype system based on theaforementioned algorithm,we can do experiment systematically and systematically through this platform.This paper uses an oblique photographic model with 10 cm resolution in LongHua district,ShenZhen city collected in 2017 and 2018 to carry out experiments in order to verify the feasibility,accuracy and operational efficiency of the method for buildings contours extraction based on multi-height analysis and the 3-D change detection of buildings considering contour features.Using the algorithm proposed in this paper to extract the contours of buildings,then detect the 3-D models change.Finally,calculate the accuracy of contour extraction and change detection to evaluate the advantages and disadvantages of the algorithm.The experiment results show that the building extraction method and change detection method of this paper have the advantages of high efficiency,good effect and good accuracy.