The pro and con of BRIEF and ORB compared to SIFT

Question

I am doing some research in Local Feature representation, so SIFT, SURF and such.

Now, has anybody here ever tried BRIEF and ORB? If so, can you discuss what are some of the pro and con with respective to SIFT?

Solution 1

here is one comparison I have found helpful. Essentially BRIEF and ORB are much faster. There is not a good comparison of scale invariance there but personally I have found SURF/SIFT to be more scale invariant than BRIEF and ORB. I recommend if you are going to use these for a specific use case you try both to see which meets your needs best.

Solution 2

SURF/SIFT uses patents which need to be payed somehow. I'm not up to date on this but the costs can be significant. So i would go with ORB if possible - except of course if you don't care about money :)

SIFT: The algorithm is patented in the US; the owner is the University of British Columbia. (http://en.wikipedia.org/wiki/Scale-invariant_feature_transform)

SURF: An application of the algorithm is patented in the US. (http://en.wikipedia.org/wiki/SURF)

Comments

[Admin]

I recommend going with SURF/SIFT to start with, and then, only if feature extraction is a bottleneck, then explore alternatives, like GPU implementations or BRIEF/ORB.

[rbaleksandar]

I find the theory behind BRIEF and ORB much easier to understand compared to SIFT and SURF, so I would actually recommend the opposite. Binary descriptors such as BRIEF are proven to be better then SURF descriptors due to their...well...binary nature. Computing Hamming distance is lightning fast compared to Euclidean not to mention the storage advantages. In addition to all that I also recommend reading that paper on each. Since SURF for example appeared before ORB the original paper will not provide comparison with ORB so you have to look somewhere else. Paper on ORB also provides it.

[amb]

ORB is not scale-invariant: vision.cs.chubu.ac.jp/CV-R/pdf/Rublee_iccv2011.pdf

[honk]

Please don't post vague assumptions as answer. If you cannot provide some reference, then your answer is only of limited help.

[Tim Kuipers]

@honk The fact that the answer is vague is no reason to not post it. I think it is important information that SIFT is patented. It is however a good idea to make your answer more concrete.

[honk]

I think it is important information that SIFT is patented: Yes that's right. But in your answer you only state that in your opinion it is patented, without providing any reference. This sound as if you are not sure about it. Why don't you add at least this and this link to your answer in order to provide a minimal reference that SURF/SIFT is patented in the US? No offense intended: I just want to motivate you to improve your answer.

[Admin]

Hope your happy with the reduced vagueness ;)

[TommyAutoMagically]

2020 update: The SIFT patent expired in March 🎉

[petervanya]

Given that the patent on SIFT has expired (and it was always known that it would at some point expire), the answer should have primarily focused on performance comparison of the algorithms. The bureaucratic aspects should have been only briefly mentioned as a PS.