hiro's profile - activity

I’d like to get the transformation between the camera and the end effector for the eye-in-hand robot without specifying the object (like chessboard)’s position in the world coordinate. Does OpenCV have the function for this?

In the documentation of cameraCalibrate, it says it can return the transformation between the object / chessboard and the world coordinate (“brings the calibration pattern from the model coordinate space (in which object points are specified) to the world coordinate space, that is, a real position of the calibration pattern in the k-th pattern view”), but it actually returns the transformation between the camera and the object. Because of this, it’s not possible to calculate the transformation between the camera and the end effector without specifying the world frame pose of the object.

http://docs.opencv.org/3.0-beta/modul...

My current understanding is that there are only two options to get what I want: Using CamOdoCal library (https://github.com/hengli/camodocal) or using ViSP’s hand2eye calibration (https://github.com/lagadic/vision_vis...).

I’d like to confirm this and if anyone knows how to get the transformation from hand to eye in OpenCV, please let me know.

I'm using a checkerboard or circles to use solvePnP. In this case, how many pictures do I need to prepare? Only one picture is fine because it includes several points? I saw some articles recommend to use at least 10 pictures for calibrateCamera, but no articles say anything about solvePnP.

Also, I noticed that the latest calibrateCamera in OpenCV3 accepts the object points in the object points' coordinate frame (= checkerboard coordinate frame), and not necessarily be in the world frame. Is it the same for solvePnP? Or do I need to have the exact 3d pose for the different checkerboard images if I change the angle?