iaslab-unipd / rgbd_calibration

RGBD Calibration
142 stars 38 forks source link

Data Collection Procedure #3

Open aseyfi opened 7 years ago

aseyfi commented 7 years ago

Hello,

Would you please let me know the steps for setting up the physical environment and collecting data?

Thanks, Ahmad

ghost commented 7 years ago

I guess you are facing the same issue as in #4

aseyfi commented 7 years ago

Hi @Sahloul,

4 is not my problem (at least yet). I am trying to understand what should I do after the software is running. For example do I need to print a checker board? How big it needs to be? Where should it be installed? How should I move the camera setup? Is there a specific pattern to be followed? ...

Thanks, Ahmad

ghost commented 7 years ago

@aseyfi I see! I wonder then if you are aware of this page. There are tutorials indicated in the second section that might help you.

aseyfi commented 7 years ago

Thanks @Sahloul , I know about RGB camera calibration. My question is specifically about this library for RGB-D camera calibration.

ghost commented 7 years ago

I though you was asking about intrinsic calibration! Extrinsic calibration is no different (you need to print a checkerboard to fit well at your desired distance, and you do not really need to fix it or make any special setup). However, in order to perform extrinsic calibration between RGB and Depth sensors, you need to capture the checkerboard by the two cameras at the same time (i.e. the camera and the board should not move during taking each pair). Of course you would need many pairs in different positions in order to have a reliable extrinsic calibration.

Now how to take such pairs? That is what #4 is talking about. At worst case, you would need to capture the two frames yourself while fixing both the board and the camera for each pair and then move to a new pose and fix again ...etc.

More information on extrinsic calibration can be found here and here.

BTW: I guess you might need to read some of their papers. Although they have not provided good documentation here, they should have explained well their method in the above papers. As for GitHub issues, they are mainly used for reporting and discussing software glitches!