Revision history [back]

Hi,

I found this:

• if the rotation R between the cameras is known and if you have also the camera intrinsic parameters for both of the cameras,
• and if there is no translation between the camera frames, the homography matrix can be computed as: [1].

There is also a function called decomposeHomographyMat(based on [2]) in OpenCV 3.0 that extract from an homography matrix and a camera calibration matrix a list of possible solution for rotation R, translation t and plane normals. Maybe you can use it to test if the formula above is correct. You can also use the function findHomography to test.

If you find some results, feel free to share and update your post.

Hi,

I found this:

• if the rotation R between the cameras is known and if you have also the camera intrinsic parameters for both of the cameras,
• and if there is no translation between the camera frames, the homography matrix can be computed as: [1].

There is also a function called decomposeHomographyMat(based on [2]) in OpenCV 3.0 that extract from an homography matrix and a camera calibration matrix a list of possible solution for rotation R, translation t and plane normals. Maybe you can use it to test if the formula above is correct. You can also use the function findHomography to test.

If you find some results, feel free to share and update your post.