def test_regression_16040_2(self):
obj_points = np.array([[0, 0, 0], [0, 1, 0], [1, 1, 0], [1, 0, 0]], dtype=np.float32)
img_points = np.array(
[[[700, 400], [700, 600], [900, 600], [900, 400]]], dtype=np.float32
)
cameraMatrix = np.array(
[[712.0634, 0, 800], [0, 712.540, 500], [0, 0, 1]], dtype=np.float32
)
distCoeffs = np.array([[0, 0, 0, 0]], dtype=np.float32)
r = np.array([], dtype=np.float32)
x, r, t, e = cv.solvePnPGeneric(
obj_points, img_points, cameraMatrix, distCoeffs, reprojectionError=r
)
def test_regression_16049(self):
obj_points = np.array([[0, 0, 0], [0, 1, 0], [1, 1, 0], [1, 0, 0]],
dtype=np.float32)
img_points = np.array(
[[[700, 400], [700, 600], [900, 600], [900, 400]]],
dtype=np.float32)
cameraMatrix = np.array(
[[712.0634, 0, 800], [0, 712.540, 500], [0, 0, 1]],
dtype=np.float32)
distCoeffs = np.array([[0, 0, 0, 0]], dtype=np.float32)
x, r, t, e = cv.solvePnPGeneric(obj_points, img_points, cameraMatrix,
distCoeffs)
if e is None:
# noArray() is supported, see https://github.com/opencv/opencv/issues/16049
pass
else:
eDump = cv.utils.dumpInputArray(e)
self.assertEqual(
eDump,
"InputArray: empty()=false kind=0x00010000 flags=0x01010000 total(-1)=1 dims(-1)=2 size(-1)=1x1 type(-1)=CV_32FC1"
)
def _solve_pnp_generic(
object_points: np.ndarray,
image_points: np.ndarray,
intrinsic_matrix: np.ndarray,
distortion_coeffs: np.ndarray = None,
prior: np.ndarray = None,
method=None,
):
"""Find an object pose from 3D-2D point correspondences.
The return value is a transformation consisting of a fixed-axis ZYX rotation
followed by a translation. It is not always correct, as cv2.solvePnP has
a tendency to silently fail -- only in corner cases as far as I know however.
N.B. OpenCV requires the points to be Nx3 and Nx2. This function will
silently flip them around if they're 3xN or 2xN, but if they're 3x3 or 2x2
they must be one row per point.
Arguments:
object_points: vector of 3D points in object coordinates
image_points: vector of corresponding 2D points in image coordinates
intrinsic_matrix: 3x3 camera intrinsic matrix
Returns: rotation vector, (x, y, z) translation in metres
"""
# Check arguments
require(object_points is not None, "Object points are required")
object_points = as_float(object_points)
require(np.all(np.isfinite(object_points)), "Object points must be finite")
if object_points.shape[1] != 3:
require(object_points.shape[0] == 3,
"Object points must be 3xN or Nx3")
object_points = object_points.transpose()
require(image_points is not None, "Image points are required")
image_points = as_float(image_points)
require(np.all(np.isfinite(image_points)), "Image points must be finite")
if image_points.shape[1] != 2:
require(image_points.shape[0] == 2, "Image points must be 2xN or Nx2")
image_points = image_points.transpose()
require(intrinsic_matrix is not None, "Intrinsic matrix is required")
intrinsic_matrix = as_float(intrinsic_matrix)
if distortion_coeffs is not None:
distortion_coeffs = as_float(distortion_coeffs)
prior_rvec = None
prior_tvec = None
if prior:
require(cv2.SOLVEPNP_ITERATIVE,
"Prior only used with SOLVEPNP_ITERATIVE")
prior_rvec, prior_tvec = prior
solutions, rvecs, tvecs, reprojection_errors = cv2.solvePnPGeneric(
object_points,
image_points,
intrinsic_matrix,
distortion_coeffs,
flags=method,
useExtrinsicGuess=prior is not None,
rvec=prior_rvec,
tvec=prior_tvec,
)
if solutions == 0:
raise RuntimeError("No solutions")
rvecs = [np.squeeze(r) for r in rvecs]
tvecs = [np.squeeze(t) for t in tvecs]
return rvecs, tvecs, reprojection_errors.reshape(-1)