This paper describes a feature point correspondence-based technique for motion recovery applied to vehicle navigation. A common configuration of the camera on autonomous vehicles moving on the ground plane is used, and a transformation of the image plane is proposed that keeps the motion of the vehicle on a plane parallel to the defined virtual image plane. A method to solve the correspondence problem is presented. The parallelism between the virtual image plane and the plane of motion allows the tracking filters defined to estimate the real-world positions of the features to remain linear, and also allows a selection of the matches that accomplish the rigidity of the scene by a Hough transform-based technique. The correspondences between features are first selected by similarity and taking into account the smoothness of motion, further processing brings out the rest of correspondences. A real application is presented consisting of a vehicle navigating in a crop, here the features to be tracked are dominant points is chain-coded contours.