robust camera calibration

Question

I will try and be specific about my question, firstly i am doing a project for 3d image reconstruction using a camera. I have completed camera calibration but it is not robust occlussions, such as placing object on chessboard it won't detect the corners. Is there a way of mopdifying the program, or can i use the corners detected before placing object and how can i use those corners to detect the corners without occlussion.

i was thinking of using the corners representing the 4 corners of the chessboard, display it and then i can place object on chessboard.

help would be appreciated in how i can modify or use the current camera calibration program to deal with occlusions i.e. placing object on chessboard.

Answer 1

For robust object detection, write a program that can detect a larger number of features, and then test whether a sufficiently high percentage of the features is present.

For a chessboard, you could try to find the 112 edges between squares, and see if they line up. It won't matter if you miss a few of those 112.