The number of SURF keypoints is different for each image. I'm storing the features for each image in different columns as
FV(N X K)=[FV_IMG1 FV_IMG2]
I tried limiting 10 best features. But ofcourse some image have below 10 features. How to deal with this variability in number of detected features? I came across normalization- I'm not sure how that is done for SURF keypoints?