There exists a algorithm for computing F from 7 points - but it is quite a bit more complex that the one for 8 points. Thus, the 8-point algorithm is a good and common baseline for computing F.
I didn't get why the F matrix has 7 dofs. I mean since F is a 3x3 matrix we would have 9 dofs but since F is in homogeneous coordinates, we have that the element in position (3,3) must be 1, hence we lose a dof. How do we pass from 8 to 7?
Great video! Really enjoying this whole series! What is final step in being able to recover scale? In the case of monocular vision where the camera has moved slightly from frame 1 to frame 2, what are the additional requirement(s) for recovering the exact translation and rotation of the camera between frame1 and frame2? Are additional (more than 2) views of the same set of image features required?
What is final step in being able to recover scale? => from just images without any further information, you cannot compute the scale. You need to have extra information, e.g., the length of the baseline vector or the size of pictured objects to recover the scale.
Great summary as always
Awsome!
Thank You so much , i've been following your instructions and so far it works ok but that's because i have some old ps3 cameras
At 3:08, I can understand that dof(E) = 5, but why dof(F) = 7? What is more, since dof(F) = 7, why we need at least 8 points rather than 7?
There exists a algorithm for computing F from 7 points - but it is quite a bit more complex that the one for 8 points. Thus, the 8-point algorithm is a good and common baseline for computing F.
Thank you professor..👍🏼
I didn't get why the F matrix has 7 dofs. I mean since F is a 3x3 matrix we would have 9 dofs but since F is in homogeneous coordinates, we have that the element in position (3,3) must be 1, hence we lose a dof. How do we pass from 8 to 7?
It has a rank deficiency
Great video! Really enjoying this whole series! What is final step in being able to recover scale? In the case of monocular vision where the camera has moved slightly from frame 1 to frame 2, what are the additional requirement(s) for recovering the exact translation and rotation of the camera between frame1 and frame2? Are additional (more than 2) views of the same set of image features required?
What is final step in being able to recover scale? => from just images without any further information, you cannot compute the scale. You need to have extra information, e.g., the length of the baseline vector or the size of pictured objects to recover the scale.