A Visual Inertial SLAM Method for Fusing Point and Line Features

The current SLAM methods generally have some drawbacks: such as poor stability and long time-consuming SLAM tasks; in order to solve the problem of poor positioning accuracy of SLAM tasks due to the drawbacks of these SLAM methods, the quality and speed of line feature extraction are improved by improving the traditional line feature extraction method LSD, and the point-line feature fusion with IMU information is fused into the visual inertial SLAM system, which can overcome the difficulties of some previous SLAM systems in facing special environments for SLAM tasks. The experimental validation of this paper's method is carried out by using data from the publicly available dataset EuRoC, and the experimental results show that this paper's visual inertial SLAM method of fusing point and line features has a high positioning accuracy.