Modern local image descriptors

asked 2014-04-11 15:27:43 -0600

Hi,

For academic purposes, I'm composing a list of modern local image descriptors. by "modern", I means ones that were published after 2011.

Currently on my list are BRIEF, BRISK, ORB, FREAK, BinBoost, ABRIEF, LIOP and DAISY.

Can someone please suggest more descriptors I can add to the list?

Sorry for the somewhat bizarre question.

Thanks,

Gil

edit retag flag offensive close merge delete

add a comment

2 answers

Sort by » oldest newest most voted

answered 2014-04-12 07:02:52 -0600

Guanta
6736 ●6 ●25 ●79

Your question seems to be a little bit fuzzy, since at each computer vision conference you will find at least one new local feature descriptor. Here some from the last ICCV (2013):

"BOLD Features to Detect Texture-less Objects" http://vision.deis.unibo.it/BOLD/
"SYM-FISH: A Symmetry-Aware Flip Invariant Sketch Histogram Shape Descriptor" http://cs.tju.edu.cn/orgs/vision/~xguo/publication/FISH.pdf
"Nested Shape Descriptors" https://github.com/jebyrne/seedoflife
"An Adaptive Descriptor Design for ObjectRecognition in the Wild" http://ece.ubc.ca/~zhenyug/iccv_2013_camReady.pdf

So maybe you should constrain your question somehow (I don't know how, maybe number of citations). One feature which raised some attention is probably KAZE and its binary version AKAZE, s. www.robesafe.com/personal/pablo.alcantarilla/kaze.html (<-- seems to be down atm), if you search via github, you'll also find some attempts to integrate it in OpenCV.

edit flag offensive delete link

Comments

Thank you @Guanta. I'm looking for some local image descriptors that one should compare against when developing a novel local image descriptor.

GilLevi ( 2014-04-12 08:00:29 -0600 )edit

@GilLevi: Well, when writing a paper, I'd compare with some popular ones, e.g. SIFT/SURF and then w. one ore two recent methods like KAZE. However, it also depends on the task and the needs, SIFT e.g. wouldn't be used for tracking.