如何使用OpenCV计算2帧之间的旋转平移矩阵

Question

I want to compute the rotation-translation matrix [R|t] matrix between 2 frames with OpenCV (see http://docs.opencv.org/modules/calib3d/doc/camera_calibration_and_3d_reconstruction.html?highlight=fundamentalmat#camera-calibration-and-3d-reconstruction )

I know I have to :

1) Detect features on both frames (with SURF , for example),
2) Matche features (with ORB and BFMatcher , for example),
3) Compute the [R|t] matrix. The intrinsic parameters are known.

However, I don't know how to complete the third step with OpenCV . Is there a regular/easy way to do this ?

My aim is to compute the trajectory of a camera.

Answer 1

You will need to use RANSAC to compute either the Fundamental or Essential matrices

OpenCV nicely provides the cv::findFundamentalMat function to do so.

Then it is a matter of getting [R|t] from both A , the intrinsic parameter matrix and F , the fundamental matrix. I would refer you to Extract Translation and Rotation from Fundamental Matrix for more information on that.

Answer 2

You can simply use stereoCalibrate where your rotation & translation matrices are R & T respectively. object_points, l_image_points, r_image_points coordinates of matched points/features in object and left/right image frames. stereoCalibrate(object_points, l_image_points, r_image_points, LcameraMatrix, LdistCoeffs, RcameraMatrix, RdistCoeffs, imageSize, R, T, E, F);