Method for estimating ego motion of an object

1. A method for estimating ego motion of an object moving on a surface,

the method comprising:

(a) generating, by a generation unit of an apparatus onboard the object, at least two composite top view images of the surface on the basis of video frames provided by at least one onboard video camera of said object moving on said surface;

(b) performing, by a region matching unit of the apparatus onboard the object, a region matching between consecutive top view images to extract global motion parameters of said moving object;

(c) calculating, by a calculation unit of the apparatus onboard the object, the ego motion of said moving object from the extracted global motion parameters of said moving object.

2. The method according to claim 1 , wherein said region matching between consecutive top view images comprises feature matching or block matching by the region matching unit of the apparatus onboard the object.

3. The method according to claim 1 , wherein the top view images are processed, by the generation unit of the apparatus onboard the object, by performing a 1D integral transformation to provide 1D integral images used for the region matching by the region matching unit of the apparatus onboard the object.

4. The method according to claim 1 , wherein the extracted global motion parameters of said moving object, extracted by the region matching unit of the apparatus onboard the object, comprise a translation and a rotation of the moving object.

5. The method according to claim 1 , wherein a motion of a feature on a top view image is decomposed, by the region matching unit of the apparatus onboard the object, into a sum of a global translation vector estimated from a large top view region and a local translation vector estimated from a small region around the respective feature.

6. The method according to claim 1 , wherein the top view image is formed, by the generation unit of the apparatus onboard the object, by a metric top view image of the surface constructed by a regular grid with a predetermined distance between samples on the surface and by projecting each sample into a camera image to obtain a pixel value.

7. The method according to claim 1 , wherein the global motion parameters are computed, by the region matching unit of the apparatus onboard the object, from a set of inliers provided by said region matching and selected by an estimator.

8. The method according to claim 1 , wherein the object is a vehicle moving on a flat surface.

9. An apparatus for estimation of ego motion of an object moving on a surface comprising

at least one onboard video camera attached to said object adapted to provide video frames of the surface;

a generation unit adapted to generate composite top view images of the surface on the basis of video frames;

a region matching unit adapted to perform a region matching between consecutive top view images to extract global motion parameters of said moving object; and

a calculation unit adapted to calculate the ego motion of said moving object from the extracted global motion parameters of said moving object.

10. The apparatus according to claim 9 further comprising a transformation unit adapted to perform a 1D integral transformation of the top view images to provide 1D integral images stored in a buffer to be used for the region matching by said region matching unit.

11. The apparatus according to claim 9 , wherein the global motion parameters are computed from a set of inliers provided by said region matching performed by said region matching unit and selected by an estimator.

12. The apparatus according to claim 11 , wherein the estimator is a RANSAC estimator.

13. The apparatus according to claim 9 , wherein the onboard video cameras are monocular cameras.

14. A vehicle moveable on a surface comprising an apparatus according to claim 9 .

15. The vehicle according to claim 14 , wherein the vehicle comprises a car moving on a road surface or a robot moving on a flat surface.