I am wondering if https://github.com/Itseez/opencv/blob/master/modules/features2d/src/orb.cpp#L375 is the coding implementation hardcoded result of ORB rBRIEF greedy search ?
If the answer is yes, then why its size is 2564 instead of 2562 since we are taking pairs of two as mentioned in ORB rublee 2011 paper ? Besides, why is N = (31-5)^2 instead of (31-5+1)^2 ?
Thanks !