ing at the problem of poor robustness and accuracy in the initialization of monocular visual simultaneous localization and mapping (SLAM), this paper proposes a robust initialization method based on point and line features. First, the line features are extracted and matched between two frames. Then, the initial rotation matrix and translation matrix between the two frames are optimized by maximizing the overlap length of the projected line features. Finally, a sliding window is used to increase the number of initial image frames, and the initial map and the estimated keyframe pose are optimized based on information and constraint of multi-frame images and the global bundle adjustment method. The test results on the TUM and OpenLORIS datasets show that compared with traditional initialization methods, the method is more robust and accurate, and can quickly complete high-precision initialization in challenging scenarios.