Method for rectification of images and/or image points, camera-based system and vehicle
View Patent ↗A method for is disclosed for rectifying images and/or image points acquired by camera of a camera-based system of a vehicle with a windshield pane. A raw image of a scene is acquired with the camera. Raw image data is selected from the raw image. Intermediate image data is calculated based on the raw image data and camera parameters. The intermediate image data includes an intermediate image or intermediate image points. The intermediate image data resembles an image/image points of the scene obtained by a pinhole camera through the windshield pane. A set of points in space of the scene is calculated, the set of points corresponding to pixels of the intermediate image or to the intermediate image points. The calculation is performed using a parallel shift of the optical path induced by the windshield pane based on windshield pane parameters. Also disclosed is vehicle and a camera-based system therefor.
1. Method for rectification of images and/or image points acquired by at least one camera of a camera-based system of a vehicle with a windshield pane, the method comprising:
acquiring a raw image of a scene with the camera;
selecting raw image data from the raw image, wherein the raw image data is the full raw image, a part of the raw image or a plurality of raw image points of the raw image;
calculating intermediate image data based on the raw image data and on camera parameters, wherein the intermediate image data comprises an intermediate image or a plurality of intermediate image points and wherein the intermediate image data resembles an image or image points of the scene obtained by a pinhole camera through the windshield pane; and
calculating a set of points in space of the scene corresponding to pixels of the intermediate image or to the intermediate image points, using a parallel shift of an optical path induced by the windshield pane based on windshield pane parameters.
2. Method according to claim 1 , wherein
a sequence of raw images of a scene is acquired with the camera;
the raw image points and/or the intermediate image points are tracked; and
the set of points in space corresponds to the tracked raw image points or intermediate image points.
3. Method according to claim 1 , wherein the windshield pane parameters comprise at least one of a normal vector n of the windshield pane in a region close to the camera, a thickness t of the windshield pane or a refraction coefficient v of the windshield pane.
4. Method according to claim 3 , wherein the parallel shift equals to a slab shift σ times the normal vector n.
5. Method according to claim 4 , wherein the slab shift σ is approximated as a constant, in particular the slab shift σ is equal to t(v−1)/v.
6. Method according to claim 4 , wherein the slab shift σ is calculated as a root σ 0 of the quartic g(σ)=a 4 σ 4 +a 3 σ 3 +a 2 σ 2 +a 1 σ+a 0 , wherein 0≤σ 0 ≤t, a 4 =v 2 , a 3 =−2v 2 (w+t), a 2 =(v 2 −1)(u 2 +t 2 )+v 2 w(w+4t), a 1 =−2t(v 2 (w 2 +u 2 )+tw(v 2 −1)−u 2 ), a 0 =(v 2 −1)t 2 (w 2 +u 2 ), w=n·s, u=√(s·s−w 2 ) and s is the space point of a point in space of the scene.
7. Method according to claim 4 , wherein the slab shift σ is calculated as a fixed point of a fixed point equation σ=φ(σ), wherein φ(σ)=t(1−1/√((v 2 −1)(u 2 /(w−σ) 2 +1)+1)), w=n·s, u=√(s·s−w 2 ) and s is the space point of a point in space of the scene.
8. Method according to claim 7 , wherein only one or two iterations of the fixed point equation are computed.
9. Method according to claim 1 , wherein the camera parameters are determined for the camera or for another camera that is similar to the camera by
acquiring a calibration image of a known pattern that is placed, without the windshield pane, in a field of view of the camera or the other camera,
comparing the calibration image to the known pattern, and
determining the camera parameters from the comparison.
10. Method according to claim 1 , wherein the windshield pane parameters are determined for the windshield pane or for another windshield pane that is similar to the windshield pane by
acquiring a calibration image of a known pattern that is placed in a field of view of the camera or another camera that is similar to the camera with the windshield pane or other windshield pane in front of the camera or other camera,
calculating an intermediate calibration image based on the calibration image and on the camera parameters, wherein the intermediate calibration image resembles an image of the known pattern taken by a pinhole camera through the windshield pane,
comparing the intermediate calibration image to the known pattern and
determining the windshield pane parameters from the comparison.
11. Method according to claim 1 , wherein the windshield pane parameters are determined or improved based on an autocalibration of the camera-based system.
12. Method according to claim 1 , wherein the calculation of the points in space of the scene corresponding to the pixels of the intermediate image or to the intermediate image points is performed by bundle adjustment.
13. Method according to claim 1 , wherein the calculated set of points in space of the scene is used for at least one of object recognition, object tracking, or for advanced driver assistance systems.
14. Camera-based system for a vehicle, comprising at least one camera to be mounted behind a windshield pane of the vehicle and a computing unit, wherein the camera-based system is adapted to perform the method according to claim 1 .
15. Vehicle, comprising a windshield pane and a camera-based system according to claim 14 .