我有一个盒子的4个世界坐标点和它们的图像点。我试图计算相机的姿势,但我得到了一个错误
File c:\Users\nmorsi200\AppData\Local\Temp\Prog6_4.py", line 292, in <module>
cv.solvePnP(objectPoints,imagePoints,cameraMatrix,distCoeffs)
cv2.error: OpenCV(4.5.5) D:\a\opencv-python\opencv-python\opencv\modules\core\src\convert_c.cpp:113: error: (-215:Assertion failed) src.size == dst.size && src.channels() == dst.channels() in function 'cvConvertScale
下面是我的代码:
imagePoints = np.float32([[[ 544 ,337 ],
[ 886 ,337 ],
[886 ,781],
[ 544 ,781 ]]])
objectPoints = np.float32([[[ 430 ,-210,90 ],
[ 430 ,-110 ,90 ],
[580 ,-210,90],
[ 886 ,-110,90 ]]])
cameraMatrix = np.float32([ 1.5968554012182622e+04, 0., 5.9163145834154648e+02, 0.,
1.0919346214593939e+04, 8.3519145165493478e+02, 0., 0., 1. ])
distCoeffs = np.float32([ 3.1817705820413217e+01, -2.4334106040843017e+03,
-5.7690325903983741e-01, -8.4352650966664180e-02,
-1.9337299660588971e+04 ])
cv.solvePnP(objectPoints,imagePoints,cameraMatrix,distCoeffs)
1条答案
按热度按时间uinbv5nw1#
你的相机矩阵是一个一维数组,而它 * 应该 * 是一个
(3,3)
形状的二维数组。OpenCV需要这个形状。它不接受平面数组。