使用来自单应性结果的 cv2.solvePnP 获取相机位姿

Question

我使用以下代码对图像进行地理配准

有输入

grid    = "for example a utm grid"
img_raw = cv2.imread(filename)
mtx, dist = "intrinsic camera matrix and 
             distortion coefficient from calibration matrix"
src_pts = "camera location of gcp on undistorted image"
dst_pts = "world location of gcp in the grid coordinate"

我校正相机失真并应用单应性

img = cv2.undistort(img_raw, mtx, dist, None, None)

H, mask = cv2.findHomography(src_pts, dst_pts, cv2.RANSAC,5.0)
img_geo = cv2.warpPerspective(img,(grid.shape[0],grid.shape[1]),\
                              flags=cv2.INTER_NEAREST,borderValue=0)

然后我想获取相机的位置。我尝试使用在 cv2.solvePnP 中计算的旋转和平移矩阵，例如 here 所示。如果我是对的，我需要至少一组 4 个共面点的相机和世界坐标。

flag, rvec, tvec = cv2.solvePnP(world, cam, mtx, dist)

再次，如果我是对的，在solvePnP中，相机坐标需要来自原始图像帧，而不是像src_pts中的未失真帧。

所以我的问题是，我怎样才能在原始图像帧中获得 src_pts 的像素位置？或者有没有其他方法可以获取 rvec 和 tvec？

Answer 1

也许函数projectPoints 是你需要的。链接在这里：http://docs.opencv.org/2.4/modules/calib3d/doc/camera_calibration_and_3d_reconstruction.html#projectpoints

Answer 2

这是我找到的解决方案

grid    = "for example a utm grid"
img_raw = cv2.imread(filename)
mtx, dist = "intrinsic camera matrix and 
         distortion coefficient from calibration matrix"
src_pts = "camera location of gcp on raw image"
dst_pts = "world location of gcp in the grid coordinate"

注意src_pts 现在是原始扭曲图像中的点

src_pts_undistorted = cv2.undistortPoints(src_pts, K, D, P=K)
img = cv2.undistort(img_raw, mtx, dist, None, None)

H, mask = cv2.findHomography(src_pts_undistorted, dst_pts, cv2.RANSAC,5.0)
img_geo = cv2.warpPerspective(img,(grid.shape[0],grid.shape[1]),\
                          flags=cv2.INTER_NEAREST,borderValue=0)

然后我可以从solvePnP得到姿势

flag, rvec, tvec = cv2.solvePnP(dst_pts, src_pts, mtx, dist)