我正在尝试使用cv2.projectPoints(),但出现错误

时间:2020-08-28 10:25:56

标签: python opencv pose-estimation

这是错误:

imgpts, jac = cv2.projectPoints(axis, rvecs, tvecs, mtx, dist)
cv2.error: OpenCV(4.2.0) C:\projects\opencv-python\opencv\modules\calib3d\src\calibration.cpp:603: 
error: (-5:Bad argument) Rotation must be represented by 1x3 or 3x1 floating-point rotation vector, or 
3x3 rotation matrix in function 'cvProjectPoints2Internal'

这是我的代码:

axis = np.float32([[3, 0, 0], [0, 10, 0], [0, 0, -50]]).reshape(-1, 3)#axis of coordinates

# PnP calculates the rotation vector and translation vector
rvecs, tvecs, inliers = cv2.solvePnP(obj_points, image_points, mtx, dist)

print(f"mtx shape: {mtx.shape}, tvecs shape: {tvecs.shape}")
print(f"mtx:\n {mtx}")
print(f"tvecs:\n {tvecs}")

# Calculate the coordinates of 3D points projected on 2D image plane
imgpts, jac = cv2.projectPoints(axis, rvecs, tvecs, mtx, dist)

这是工厂:

mtx shape: (3, 3), tvecs shape: (3, 1)

mtx:
 [[1.71223579e+03 0.00000000e+00 1.02990683e+03]
 [ 0.00000000e+00 1.70818827e+03 7.83446773e+02]
 [ 0.00000000e+00 0.00000000e+00 1.00000000e+00]]

tvecs:
 [[-0.09038089]
 [ -0.05386737]
 [ -0.01652085]]

我不知道如何解决它。问题似乎在于参数的形式。但是,当我检查它们时,它们的形状似乎正确。所以我不知道问题出在哪里,但是如果您这样做,将会很有帮助。

1 个答案:

答案 0 :(得分:1)

也许您正试图从旧教程中获取一些代码,或者您误读了有关返回值的函数文档。尝试像这样从solvePnP分配返回值:

result, rvecs, tvecs = cv2.solvePnP(obj_points, image_points, mtx, dist)

solvePnP函数(至少从opencv4开始)不应有“内含物”。