0

I have a model of a object and a image of the object, and tried to align the object to the image. The camera pose and fov are not known, as the image are collect from different users and taken with different camera.

What I do now is assuming the fov is 60 degree and solve this problem with solvPnp. However, the result are not good, unless use a right fov information.

Is there a function in OpenCV to solve this kind of problem?

heLomaN
  • 1,634
  • 2
  • 22
  • 33
  • Could you add your input images? If you have an anchor point in both images, you can use the model of the object as a template and perform template matching on the other image. If there is a match, you will be able to obtain the coordinates of the image – nathancy Jul 17 '19 at 03:22
  • @nathancy We only have one image from each user, so can just assume one image per camera. Though, the fov don't vary a lot. – heLomaN Jul 17 '19 at 04:03

0 Answers0