基于语义的倾斜影像单体建模技术研究

Study of semantic modeling framework-based method for 3Dbuilding reconstruction using oblique aerial images

针对应用倾斜航空影像进行自动化建模所创建的建筑物建模存在诸如模型细节缺失,难以获取语义信息的问题,该文提出了一种采用基于语义的建模方式构建建筑物模型的方法;首先对密集图像匹配(DIM)点云进行建筑物类提取,实现场景中建筑和其它地物的分离;然后进行基于点云分割和建筑物验证的建筑物单体标注,以保证对单体建筑物的有效标记;最后利用倾斜航空影像和DIM点云的建筑物各组件的语义识别和提取算法,实现以单体建筑物为建模对象的模型重建;同时,设计适用于建筑物重建的语义建模框架,对单体语义模型进行自动化纹理映射,构建出语义化的建筑物模型。该文提出的建模关键技术方法将有助于减少城市建模成本,提高建模效率,为智慧城市基础数据的建设提供关键技术储备,并发挥相应的的社会和经济效益。

Aiming that the 3 dimension models generated by the oblique imagery have some limitations,such as the absence of model sharp features,the absence of semantic information.To solve these limitations,this project used a semantic based method to build 3 D models.Our research in the project was based on the building detection in density imagery matching point cloud,which can segment the buildings point cloud from ground point cloud.Based on this,to obtain the detached building,building segmentation labeling would be researched,then the algorithms about semantic feature recognition of building elements would be researched to modeling the 3 D building models.Meanwhile,a semantic modeling framework would be designed for 3 Dreconstruction,which can produce the 3 D models with rich semantic information.Aside of this,the research about automatic texture mapping was also involved in our project.In the project,apractical and advanced technology of 3 D modeling based on oblique imagery and software prototype would be maintained.This project would contribute to the application of oblique imagery in various fields,and had a large benefits in social and economic areas.