Visual Odometry is a system of estimating the egomotion of the moving camera which can solely rely on pure visual data. This study compares the performance of Visual Odometry with MEMS IMU/GPS integration. Visual Odometry estimates the motion of camera based on sequential input images. The fundamental of this approach is to track feature points from the images of the moving camera. First, detecting the feature points in each frame. Second, matching feature points between two sequential images. In the last, by tracking these feature points so that the pose of moving camera can be produced. In our experiments, we utilize the images from single moving camera as input data to proceed the Visual Odometry to estimate the motion of camera. The results of Visual Odometry have been evaluated by result of INS/GPS integration. The results show that the accuracy of different situations of Visual Odometry and drawbacks that need to be improved.