Data collection and processing of human faces are hot spots of reverse engineering. This paper studies the data collection of human faces, the pre-treatment of point clouds, and analyzes the error of data collection at the same time. Data collection mostly uses proper system parameters to get desired point clouds. Pretreatment of point clouds deletes the desultory points first, then merge the point clouds. Finally, analyze the errors that are caused in the process of data collection, mostly analyze the effects of equipment, ambient light, and system parameters on accuracy of data collection, then find the best treatment to get a desirable experimental result.