With the development of digital cities and smart cities, urban simulation has developed from two-dimensional to three-dimensional, and the requirements for urban three-dimensional real-world modeling have become higher and higher. Due to the high density and rapid change of urban features, traditional manual 3D modeling methods are inefficient and cannot adapt to the rapidly changing and changing urban environment. Therefore, UAV tilt photography technology is often used for fast 3D modeling, but UAV tilt photography method cannot solve the problems of urban object occlusion, canopy occlusion, eaves occlusion, glass light transmission, etc., resulting in defects such as local texture distortion of the 3D model, object pulling, and object cavity voids. There are many problems in model accuracy, quality and application. In order to solve the above problems, this article took the University of Chinese Academy of Sciences as the research area, and used the UAV tilt photography technology and ground lidar technology to perform fusion modeling. In the course of the experiment, a drone image acquisition strategy of "regular route automatic shooting-based, manual interest area supplemented by manual shooting" was used in the field; in interior work, "manual coarse registration, ICP algorithm fine registration" was used. The experimental results show that the method of multi-source data fusion can not only ensure the efficiency of 3D modeling, but also correct the problems of ground distortion and holes in the UAV alone modeling model, improve the accuracy of the model, and optimize the 3D model of the city.