Taking one camera and moving it around to take two images of the same object, from a different viewpoint, one should be able to compute a matrix that relates these two scenes. In OpenCV, how is this accomplished?
Share
Sign Up to our social questions and Answers Engine to ask questions, answer people’s questions, and connect with other people.
Login to our social questions & Answers Engine to ask questions answer people’s questions & connect with other people.
Lost your password? Please enter your email address. You will receive a link and will create a new password via email.
Please briefly explain why you feel this question should be reported.
Please briefly explain why you feel this answer should be reported.
Please briefly explain why you feel this user should be reported.
Taking one camera and moving it around to take two images of the same object, from a different viewpoint, one should be able to compute a matrix that relates these two scenes. In OpenCV, how is this accomplished?
If said object is a calibration pattern like the chessboard used by OpenCV, then the camera calibration routine mentioned by ChrisO would give you both the camera intrinsics (focal length, principal point, and lens distortion) as well as the camera extrinsics (where they are relatively in space).
If you have general object, then you need to establish a set of 2D correspondences which you can feed into cvFindFundamentalMat. This finds the fundamental matrix which relates the two perspectives. Namely, for each point x in camera 1 and corresponding point x’ in camera 2, x’Fx = 0. You can similarly find the epipoles, etc. This uses the 8 point algorithm which requires at least 8 point pairs of point correspondences.
You can get the correspondences either manually or with a robust feature extractor and matcher along the lines of MSER/Affine Harris + SIFT.