Localization and ego-velocity estimation for vehiclebased on binocular image sequences

A real-time ranging location and ego-velocity estimation method based on binocular image sequences was presented to obtain the relative location and velocity of the vehicle in the running process. This method used the vehicle-borne binocular vision sensor to collect the image sequence of the surrounding environment. Then the depth of field of environmental feature points was obtained through matching feature points of left and right images at the same time based on speeded up robust features(SURF) in order to achieve vehicle location. Meanwhile, the feature points of two adjacent frames were tracked and matched based on SURF.The transformation parameter of the camera coordinate system before and after vehicle movement was computed through the 3D coordinates of the corresponding matching points in the two adjacent frames. And the velocity of the vehicle was estimated according to the transformation parameter. The simulation experiment results show that the method is feasible, and the speed calculation results are relatively stable, the average error is less than 6%.