The research mainly aims to solve the application problems of UAV railway patrol video spatial absolute positioning and railway mileage positioning. Based on the collinear equation, the spatial positioning model of line patrol video is constructed by using the spatial position and attitude information of UAV during flight. Based on the line patrol video spatial positioning model, the algorithm is expanded, and the video railway mileage positioning, three-dimensional scene and video synchronization and two-phase video synchronization algorithms are completed. The accuracy is verified by the actual engineering project. The mean square error of mileage positioning is 3.586 m, the mean square error of scene video synchronization is 4.23 m, and the mean square error of two-phase video synchronization is 3.67 m. This method realizes the geographic information coding of line patrol video, and provides basic conditions for the accurate application of UAV video and multi-source data fusion application in the railway industry.