In the process of the evolution of urban-rural spatial patterns, traditional village forms have been seriously eroded and destroyed, resulting in the gradual loss of garden feature elements and the blurring of spatial features. Facing these problems, accurate and efficient acquisition of relevant data becomes the key to traditional village conservation. We take the typical Huizhou Shuikou garden Tangmo village as an example and conduct a study on how to use UAV aerial images to extract spatial feature data. Firstly, the UAV was used to acquire aerial images of the study area through a predefined mission. Then, the image data are processed by using structure from motion (SfM) software to obtain digital surface model (DSM), digital orthophoto map (DOM), point cloud, and 3D model. Finally, the produced spatial data are applied to the spatial feature extraction analysis study. The results show the following: (1) the data produced by UAV aerial images have a horizontal accuracy of 0.034 m and a vertical accuracy of 0.039 m, which meet the requirements of traditional village spatial data collection. (2) The results of spatial feature elements and terrain feature extraction show that the 3D model and DSM can accurately extract and analyze the micro and macro spatial feature of traditional villages. (3) By using the cloth simulation filtering (CSF) to extract point clouds of buildings and streets, and after statistical analysis, we could quickly obtain their spatial features. Based on the spatial data obtained by using UAV, the study achieves the accurate collection of Tangmo DSM, DOM, 3D model, and spatial feature data and forms a method of data collection, processing and spatial feature extraction. The results of the study can provide a scientific basis for data collection on the spatial features of other cultural heritages and the conservation of traditional villages.