Simultaneous localization and mapping (SLAM) is one of the main research topics in the field of robotics. In order to solve the problem of how to estimate the camera pose according to the image, an improved visual odomety method based on VINS (ORLK-VINS) is proposed in this paper. First of all, the image information is obtained by the stereo camera; secondly, the image information is processed by histogram equalization to improve the image contrast and brightness; then, the original image feature extraction algorithm is improved by introducing FAST corners with direction in the ORB algorithm. Finally, the extracted feature points are tracked and matched by forward and reverse LK optical flow to ensure the accuracy of the matching feature points. The experimental results show that the improved visual speedometer has better real-time performance and positioning accuracy in some scenes than the mainstream VINS-Fusion algorithm.