Research On Building Extraction Algorithm For Oblique Image Based On Point Feature Map

In urban environmental protection,the "digital city" concept is the main solution for urban scene data management.By digitizing urban objects and intelligentizing urban management,it assists in decision-making on environmental protection,population management,social services,and urban planning.As an important data source in the construction of "digital city",Oblique image data and 3D point cloud data can be obtained based on oblique photogrammetry,The research on the extraction of building objects in oblique image data and the detection of building objects in 3D point clouds has great application value.The traditional method of obtaining urban scene data is mainly through traditional remote sensing photogrammetry,by analyzing scene image data,and evaluating and predicting environmental changes based on information such as building density,vegetation coverage,and water distribution in the image data.Compared with the traditional method of acquiring image data,the oblique photogrammetry technology uses multi-angle shooting methods to obtain multi-view images of the scene containing a vertical perspective,which has richer feature information and can obtain highly accurate true 3D models.However,due to algorithmic reasons,the ground objects in this 3D model are inseparable,making it difficult to perform more advanced analysis applications.This paper analyzes the principle and data characteristics of oblique photography technology,and combines common algorithms in the fields of point cloud filtering and image processing.This paper proposes a point of interest extraction algorithm for buildings based on point cloud filtering,and obtains the 3D object detection frame of the building and point feature map,and completes the extraction of building objects in oblique images.The main contents of this article include:(1)The research background and significance of the thesis are briefly explained,and the research status of building extraction algorithms in different data sources is summarized.The data acquisition scheme of oblique photogrammetry technology and the 3D modeling process based on oblique photo technology are introduced,and the data results of tilt photography technology and the application status in the field of environmental engineering are briefly analyzed;(2)Aiming at a large range of urban scenes,a building object detection algorithm for densely matching point clouds in oblique images is proposed.In view of the inseparability of buildings in the scene,according to the unique color,normal and elevation information of different ground objects,Remove non-building objects from the scene,then obtain the points of interest of the building based on the multi-scale filtering method,and finally obtain the detection frame of the building object according to the spatial position of the points of interest;(3)This paper proposes a building extraction method based on oblique images.According to the experience of high-dimensional scenes having a good interpretation of low-dimensional scenes.In the virtual 3D scene,the corresponding view image is obtained by tilting the unique azimuth angle of the image.Then,according to the points of interest of the building,a point feature map corresponding to the oblique image is obtained.Finally,the building mask is obtained through image processing to complete the extraction of building objects.