The invention provides an indoor building three-dimensional modeling method based on point cloud data. The method comprises the steps of firstly, obtaining and preprocessing the three-dimensional point cloud data; then carrying out the point cloud plane clustering to extract plane patches, calculating the neighborhoods of the plane patches according to a point set contained in the plane patches and used for plane semantic classification, merging processing, plane boundary capturing and topological information generation, and finally outputting a closed three-dimensional patch model which is composed of three-dimensional plane polygons and contains semantic and topological information of the indoor building. According to the invention, the problems such as the measurement errors existing inpoint cloud data, the density distribution unevenness, the data is incomplete due to indoor object shielding and the lack of semantic information, are overcome, a working process for automatically and quickly establishing the three-dimensional surface patch model of the indoor building by utilizing the three-dimensional point cloud is provided, the three-dimensional point cloud data of differentsources can be processed, a new method is provided for automatic and quick modeling of the indoor building, and a rich data basis is established for the indoor three-dimensional model in the fields ofmeasurement, building decoration, property management and the like.