The fusion algorithm based on multi-state constrained Kalman filtering solely uses a single frame image for pose estimation. If the initialization is incorrect, then it can cause severe divergence in visual pose estimation.Furthermore, each visual feature point in the system state vector can easily lead to computational burden to the system. Given the aforementioned problems, an improved keyframe selection algorithm is proposed, which uses multiple visual keyframes to constrain the same feature points for reducing visual measurement errors and improving positioning accuracy. Simultaneously, only the camera pose calculated from keyframes is integrated into the system state vector, which can effectively reduce system computation. The experiment shows that the improved algorithm enhances positioning accuracy and computational efficiency by 29.09% and 32.2%, respectively, when compared to EKF. Additionally, the proposed algorithm increases computational efficiency by 35.48% when compared to that of Orb-slam2.