Camera calibration method, recording medium, and camera calibration apparatus
A camera calibration method, which calculates camera parameters of two cameras using calibration points, includes: (a1) acquiring three-dimensional coordinate sets of the calibration points and image coordinate pairs of the calibration points in a camera image of each camera; (a2) acquiring multiple camera parameters of each camera; (a3) for each calibration point, calculating a view angle-corresponding length corresponding to a view angle of the two cameras viewing the calibration point; (a4) for each calibration point, calculating a three-dimensional position of a measurement point corresponding to a three-dimensional position of the calibration point using parallax of the calibration point between the two cameras; (a5) for each calibration point, weighting a difference between the three-dimensional coordinate set of the calibration point and the three-dimensional position of the measurement point corresponding to the calibration point using the view angle-corresponding length corresponding to the calibration point; and (a6) updating the camera parameters based on the weighted difference.
1. A camera calibration method which calculates camera parameters of two cameras by use of calibration points, the camera calibration method comprising:
(a1) acquiring calibration point data which is stored in a first memory and which includes three-dimensional coordinate sets of the calibration points and image coordinate pairs of the calibration points in a camera image of each of the cameras;
(a2) acquiring camera parameters of each camera which are stored in a second memory;
(a3) for each calibration point, calculating a first value which corresponds to a value of an angle of view of the two cameras viewing the calibration point based on the calibration point data and the camera parameters;
(a4) for each calibration point, calculating a three-dimensional position of a measurement point which corresponds to a three-dimensional position of the calibration point by use of parallax of the calibration point between the two cameras based on the image coordinate pair of the calibration point and the camera parameters;
(a5) for each calibration point, weighting a difference between the three-dimensional coordinate set of the calibration point and the three-dimensional position of the measurement point corresponding to the calibration point by use of the first value calculated using the calibration point;
(a6) updating the camera parameters based on the weighted difference; and
(a7) outputting the updated camera parameters,
wherein at least one of the process (a1) to the process (a7) is executed by a processor, and
wherein the first value is equal to 2×L×sin (α/2), L being a distance between a central position of a lens of any of the two cameras and the calibration point, and α being the value of the angle of view.
2. The camera calibration method according to claim 1 , wherein
when weighting the difference, the process (a5) normalizes the first value calculated using the calibration point to be used by a total sum of the first values calculated using the respective calibration points.
3. The camera calibration method according to claim 1 , wherein the process (a5) uses only the difference related to the calibration point used to calculate the first value, when the first value is larger than a threshold.
4. The camera calibration method according to claim 1 , wherein when weighting the difference, the process (a5) divides the first value calculated using the calibration point to be used by a distance between the calibration point and the cameras.
5. The camera calibration method according to claim 1 , wherein when weighting the difference, the process (a5) divides the calculated using the calibration point to be used by a square of a distance between the calibration point and the cameras.
6. The camera calibration method according to claim 1 , wherein
the process (a6) calculates an evaluation value using a total sum of the differences for the respective calibration points and updates the camera parameters such that the evaluation value is small.
7. A recording medium which is non-transitory and computer-readable and which includes a control program to cause a device provided with a processor to execute processing, the processing being executed by a computer and comprising:
(a1) acquiring calibration point data including three-dimensional coordinate sets and image coordinate pairs of calibration points from a first memory, the image coordinate pairs of the calibration points being image coordinate pairs of the calibration points in a camera image of each of two cameras;
(a2) acquiring camera parameters of each camera from a second memory;
(a3) for each calibration point, calculating a first value which corresponds to a value of an angle of view of the two cameras viewing the calibration point based on the calibration point data and the camera parameters;
(a4) for each calibration point, calculating a three-dimensional position of a measurement point which corresponds to a three-dimensional position of the calibration point by use of parallax of the calibration point between the two cameras based on the image coordinate pair of the calibration point and the camera parameters;
(a5) for each calibration point, weighting a difference between the three-dimensional coordinate set of the calibration point and the three-dimensional position of the measurement point corresponding to the calibration point by use of the first value calculated using the calibration point;
(a6) updating the camera parameters based on the weighted difference; and
(a7) outputting the updated camera parameters,
wherein the first value is equal to 2×L×sin (α/2), L being a distance between a central position of a lens of any of the two cameras and the calibration point, and α being the value of the angle of view.
8. The recording medium according to claim 7 , wherein
when weighting the difference, the process (a5) normalizes the first value calculated using the calibration point to be used by a total sum of the first values calculated using the respective calibration points.
9. The recording medium according to claim 7 , wherein the process (a5) uses only the difference related to the calibration point used to calculate the first value, when the first value is larger than a threshold.
10. The recording medium according to claim 7 , wherein when weighting the difference, the process (a5) divides the first value calculated using the calibration point to be used by a distance between the calibration point and the cameras.
11. The recording medium according to claim 7 , wherein when weighting the difference, the process (a5) divides the first value calculated using the calibration point to be used by a square of a distance between the calibration point and the cameras.
12. The recording medium according to claim 7 , wherein
the process (a6) calculates an evaluation value using a total sum of the differences for the respective calibration points and updates the camera parameters such that the evaluation value is small.
13. A camera calibration apparatus which includes a processing circuit to calculate camera parameters of two cameras, the processing circuit performing processing comprising:
(a1) acquiring calibration point data which includes three-dimensional coordinate sets of calibration points and image coordinate pairs of the calibration points in a camera image of each of the two cameras from a first memory;
(a2) acquiring camera parameters of each camera from a second memory;
(a3) for each calibration point, calculating a first value which corresponds to a value of an angle of view of the two cameras viewing the calibration point based on the calibration point data and the camera parameters;
(a4) for each calibration point, calculating a three-dimensional position of a measurement point which corresponds to a three-dimensional position of the calibration point by use of parallax of the calibration point between the two cameras based on the image coordinate pair of the calibration point and the camera parameters;
(a5) for each calibration point, weighting a difference between the three-dimensional coordinate set of the calibration point and the three-dimensional position of the measurement point corresponding to the calibration point by use of the first value calculated using the calibration point;
(a6) updating the camera parameters based on the weighted difference; and
(a7) outputting the updated camera parameters,
wherein the first value is equal to 2×L×sin (α/2), L being a distance between a central position of a lens of any of the two cameras and the calibration point, and α being the value of the angle of view.
14. The camera calibration apparatus according to claim 13 , wherein
when weighting the difference, the processing circuit normalizes the first value calculated using the calibration point to be used by a total sum of the first values calculated using the respective calibration points.
15. The camera calibration apparatus according to claim 13 , wherein the processing circuit uses only the difference related to the calibration point used to calculate the first value, when the first value is larger than a threshold.
16. The camera calibration apparatus according to claim 13 , wherein when weighting the difference, the processing circuit divides the first value calculated using the calibration point to be used by a distance between the calibration point and the cameras.
17. The camera calibration apparatus according to claim 13 , wherein when weighting the difference, the processing circuit divides the first value calculated using the calibration point to be used by a square of a distance between the calibration point and the cameras.
18. The camera calibration apparatus according to claim 13 , wherein
the processing circuit calculates an evaluation value using a total sum of the differences for the respective calibration points and updates the camera parameters such that the evaluation value is small.