The invention discloses a multi-strategy rail transit obstacle recognition method based on a laser radar. The method comprises the steps: scanning, by a laser radar, the train operation environment data acquisition in real time in an actual train operation environment, wherein the obtained point cloud samples are used for establishing a point cloud data set of obstacles in front of a train; performing sampling filtering on the point cloud by using a VoxelGrid voxel filtering method; creating a new point cloud container, traversing the filtered point clouds, and storing the point clouds according to the distance; performing ground segmentation on indoor and outdoor rail transit train operation scenes by using different strategies, performing preliminary extraction on a rail region in the segmentation process, and finally performing accurate limitation on the extracted rail region and performing fitting by using a space curve; and performing multi-strategy obstacle detection according to different distances, and performing regional identification and screening based on multiple strategies of the distances, thereby obtaining a screening result. According to the invention, real-time detection of obstacles in front of the train can be realized, high-precision detection is carried out in a real scene in the whole experiment, and debugging is easy.