This study explores an indoor rescue navigation and positioning system based on the fusion of point and line features in binocular visual SLAM, aiming to address the navigation and positioning challenges of personnel in indoor environments where satellite signals are blocked. Pedestrians carry binocular camera to gather indoor environmental information and utilize SLAM algorithms to calculate their own positions, enabling information exchange with external command personnel to ensure operational rescue and personnel safety. The paper employs the LSD line feature extraction method and the LBD matching method, constructs a re-projection error model for line features in binocular vision, and compares the proposed SLAM algorithm with others using the EuRoC dataset and the corridor weak-textured dataset. Experimental results demonstrate that the system exhibits considerably high positioning accuracy and real-time performance capabilities, offering more precise and robust positioning results in indoor weak-texture environments, verifying the feasibility of the system.