Abstract: The train positioning technology based on the Global Navigation Satellite System (GNSS) reduces the cost of railway construction and operation. As the basis of map matching method, digital orbit map is usually used as an auxiliary positioning means to correct satellite positioning deviation. Accurate description and reduction of track alignment are the premise of the construction and storage of track electronic map. This paper used the azimuth approximate estimation of curvature to preliminarily identify and classify the track plane alignment. In order to improve the fitting accuracy of the line shape, the paper introduced an adaptive weight factor to optimize the least squares fitting method, and eliminate the influence of large deviation observation data, thereby ensure the smooth fitting and accuracy of the overall trajectory data. The verification of measured vehicle trajectory data shows that the algorithm can effectively extract and accurately describe the characteristics of vehicle trajectory.