I am using vl_sift(from vl_feat site)function for finding SIFT descriptor, which is giving me descriptor of size 128*1 for one image and 128*2 for the second. I got the reason behind 128 but I am not getting why the columns keep changing(sometimes 1 and sometimes 2),it has to be same right? Can anyone explain me what does those columns represents? Thank you
The features are extracted from the key points. In SIFT case they are blob like structures. In other words the vl_sift function looks for the key point and extracts a 128 dimensional descriptor from EACH of the key points.
You have various dimensions (128*1 or 128*2 etc...) because in one image vl_sift finds one key point in the other one it finds 2 key points. That is why you have various dimensions.