Motion estimation of a vehicle using sequential recursive algorithm from images

Aiming at solving the problem of coupling and time-consuming in estimating the motion parameters of a vehicle based on computer vision, a sequential model for recursive algorithm is designed. It is proved that the estimation of rotation parameters can be separated from that of the translation parameters. With this fact, the rotation model is build based on lines, and then the proposed sequential model is build. A proof for the uniqueness of the motion parameters is given. The sequential model is tested using both simulated and real images, and the results show that the time-consuming problem is solved partly, estimation of rotation parameter is more accurate, and the estimator based on the sequential model is more robust.