I have an object that is moving parallel to a stationary camera. I would like to use Structure from Motion or Shape from Silhouette to obtain the 3D points of the object.

For that, I need to obtain the camera matrix P = K [R | t]. I know that I can calculate the intrinsic matrix K using the checkerboard technique. But how do I know the rotation and translation of the camera?

  1. I know that the object is moving in a constant speed. Exactly 10cm per second. And I am taking one frame per second. Does that relate to the camera moving 10cm? That would mean that the x value of the translation vector t is incremented by 10 for each image. The second image would therefore have the translation vector (10 0 0)

  2. As the camera is stationary and the object moves parallel to the camera I am guessing that the rotation matrix is the identity I for each image.

Is my thinking correct?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.