A new SLAM system based on tightly coupled binocular camera and IMU is developed in this paper. In this system, line and point features are extracted from the front end of the system. Meanwhile, the camera poses and the positions of Key-Points are calculated with the optimization method at the back end of this system. In this paper, Key-Points and descriptor are extracted from point features for system tracking, relocation and loop-detection. For line features, we only extract points from line features for calculating the line direction in the image. Optical flow method is adopted for tracking line features and line features are only used in the tracking parts. In the back end optimization of the system, tightly coupled method is used for optimizing vision and IMU constrain. The tightly coupled binocular camera and IMU SLAM system in this paper improves the accuracy and robustness compared with traditional SLAM systems.