针对以航空影像为数据源的建筑物三维重建中存在的自动化程度低、精度不高、计算量大的问题,提出一种针对任意建筑物的半自动建筑物三维建模方法。对传统的模型导向的方法进行改进,首先建立基本模型库,从模型库中选择模型组建建筑物,根据建筑物内部约束确定建筑物参数;然后提出依据影像上提取的建筑物边缘信息计算建筑物参数初始值的方法;最后使用广义点摄影测量理论进行平差运算计算建筑物的精确参数。针对几种不同类型的建筑物进行实验,结果表明,该方法切实可行,可以对不同形状的建筑物重建,且具有精度高、计算速度快、人工辅助少的优点。
3D building reconstruction based on aerial images has many problems,such as needing too much human-computer interaction,low precision,low calculation speed and so on.To solve these questions,this paper proposed a model-driven semiautomatic 3D building reconstruction method.Firstly,a model library that includes some primitives was build,and primitives chosen to establish the building and parameters of the building were confirmed.Secondly,the initial parameters of the model were calculated by building edges that extracted from the images.Thirdly,model-image fitting algorithm based on generalized point photogrammetry was carried out to calculate accurate model parameters.The experiments with several different kinds of buildings prove that the proposed method has efficiency,high precision,fast calculation speed,and less human-computer interaction.