Method and apparatus with vanishing point estimation

1. A method with vanishing point estimation, comprising:

detecting objects in an image of a current time point of the objects comprising a target vehicle;

tracking positions of the objects in a world coordinate system by associating the objects with current position coordinates of the objects determined from images of previous time points that precede the current time point;

determining a vanishing point for each of the objects including the target vehicle based on the positions of the objects; and

outputting the vanishing point determined for each of the objects.

2. The method of claim 1 , wherein the detecting of the objects comprises:

classifying a type of the target vehicle; and

detecting the target vehicle in the image of the current time point based on the classified type.

3. The method of claim 2 , wherein the detecting of the target vehicle comprises:

determining a bounding box corresponding to a rear of the target vehicle based on the classified type; and

detecting the target vehicle using the bounding box.

4. The method of claim 3 , wherein the determining of the bounding box comprises:

retrieving prestored width information corresponding to the classified type; and

generating the bounding box corresponding to a width of the target vehicle based on the width information.

5. The method of claim 1 , wherein the tracking of the positions of the objects comprises:

predicting positions of second bounding boxes corresponding to the determined current position coordinates using a Kalman filter;

matching first bounding boxes corresponding to the objects and the second bounding boxes; and

tracking the positions of the objects in the world coordinate system based on a result of the matching.

6. The method of claim 5 , wherein the Kalman filter is used to determine a position of a second bounding box corresponding to current position coordinates of the target vehicle by predicting the current position coordinates of the target vehicle and a speed in x and y directions of the target vehicle based on a measurement value comprising a width of the target vehicle in the image of the current time point and a distance in a horizontal direction from a center of the width.

7. The method of claim 1 , wherein the determining of the vanishing point comprises:

determining a vanishing point of an object among the objects based on a relationship between a position of a first bounding box corresponding to the object, a height of a camera capturing the image of the objects, and a position of a third bounding box corresponding to the object in an image plane.

8. The method of claim 1 , wherein the outputting of the vanishing point further comprises:

projecting and displaying the objects onto the image of the current time point by the vanishing point determined for each of the objects.

9. The method of claim 1 , further comprising:

calculating a distance between the vanishing point for each of the objects based on the vanishing point determined for each of the objects; and

outputting the calculated distance.

10. The method of claim 1 , further comprising:

generating driving assistance information corresponding to each of the objects based on the vanishing point determined for each of the objects.

11. The method of claim 10 , wherein the generating of the driving assistance information comprises:

calculating a distance between the vanishing point for each of the objects and each of the objects, based on the vanishing point determined for each of the objects; and

generating the driving assistance information comprising a driving line and a driving lane of each of the objects based on the calculated distance.

12. A non-transitory computer-readable storage medium storing instructions that, when executed by a processor, configure the processor to perform the method of claim 1 .

13. An apparatus with vanishing point estimation, comprising:

a processor configured to:

detect objects in an image of a current time point of the objects comprising a target vehicle;

track positions of the objects in a world coordinate system by associating the objects with current position coordinates of the objects determined from images of previous time points that precede the current time point; and

determine a vanishing point for each of the objects including the target vehicle based on the positions of the objects; and

a display configured to output the vanishing point determined for each of the objects.

14. The apparatus of claim 13 , wherein, for the detecting of the objects, the processor is configured to:

classify a type of the target vehicle; and

detect the target vehicle in the image of the current time point based on the classified type.

15. The apparatus of claim 14 , wherein, for the detecting of the target vehicle, the processor is configured to:

determine a bounding box corresponding to a rear of the target vehicle based on the classified type; and

detect the target vehicle using the bounding box.

16. The apparatus of claim 15 , wherein, for the determining of the bounding box, the processor is configured to:

retrieve prestored width information corresponding to the classified type; and

generate the bounding box corresponding to a width of the target vehicle based on the width information.

17. The apparatus of claim 13 , wherein, for the tracking of the positions of the objects, the processor is configured to:

predict positions of second bounding boxes corresponding to the determined current position coordinates using a Kalman filter;

match first bounding boxes corresponding to the objects and the second bounding boxes; and

track the positions of the objects in the world coordinate system based on a result of the matching.

18. The apparatus of claim 17 , wherein the Kalman filter is used to determine a position of a second bounding box corresponding to current position coordinates of the target vehicle by predicting the current position coordinates of the target vehicle and a speed in x and y directions of the target vehicle based on a measurement value comprising a width of the target vehicle in the image of the current time point and a distance in a horizontal direction from a center of the width.

19. The apparatus of claim 13 , wherein, for the determining of the vanishing point, the processor is configured to:

determine a vanishing point of an object among the objects based on a relationship between a position of a first bounding box corresponding to the object, a height of a camera capturing the image of the objects, and a position of a third bounding box corresponding to the object in an image plane.

20. The apparatus of claim 13 , wherein, for the outputting of the vanishing point,

the processor is configured to project the objects onto the image of the current time point by the vanishing point determined for each of the objects, and

the display is configured to display the projected objects.

21. A method with vanishing point estimation, comprising:

determining a lower edge of bounding box of an object in an image;

determining a position of the object in a world coordinate system based on the lower edge of the bounding box;

determining a vanishing point of the object based on the lower edge of the bounding box and the position of the object in the world coordinate system; and

outputting the vanishing point of the object.

22. The method of claim 21 , wherein

the determining of the lower edge comprises determining coordinates of a lower center of the bounding box, and

the determining of the vanishing point of the object comprises determining a position of the vanishing point in a vertical direction based on the coordinates of the lower center of the bounding box, the position of the object in the world coordinate system, and one or more intrinsic parameters of a camera used to obtain the image.

23. The method of claim 21 , wherein the determining of the position of the object in the world coordinate system comprises:

matching the bounding box and a second bonding box; and

determining the position of the object in the world coordinate system based on a result of the matching.

24. The method of claim 23 , wherein

the bounding box is generated by classifying a type of the object, and

the second bounding box is generated using a Kalmann filter.