# Extrinsics matrix after cv::StereoCalibrate

Hi all,

I have two cameras (2 kinects specifically) and I am using cv::StereoCalibrate to compute the position and orientation of Kinect2 relative to Kinect1. The function returns (among others), the rotation matrix R and the translation vector T.

Is the final transformation matrix given by E = [R | T] (meaning, I just use the rotation and the translation as is to compose the matrix -and then transform it to homogeneous 4x4-) or I should do something else? (Like E = [R | -R^(-1)*T] which I saw somewhere).

What I want to do is multiply the Kinect2 point cloud with this matrix to register the two point clouds.

Thanks so much