Match descriptors with cluster centers

asked 2015-06-02 15:00:19 -0500

Kevvin gravatar image

Hi I am new to opencv, right now. i want to do the images comparison. For example, i have 4 images as my database. Firstly, i get all the descriptors and cluster them into several groups. After that, i have a query image which is one image, i also get the descriptors. Right now, i want to match the query image descriptors with the existing cluster centers and group them. But i can not find any functions for doing that. Thanks for helping

edit retag flag offensive close merge delete


I don't know if you can match your descriptor without writting your own code but if you use descriptor defined in opencv ( ORB, SIFT....) you can found matcher algorithm and a sample in samples\cpp\tutorial_code\features2D\AKAZE_match.cpp Blob features include center so may be you can use blob data for your descriptor.

LBerger gravatar imageLBerger ( 2015-06-03 01:05:57 -0500 )edit

@Kevvin: What you are doing is called bag of (visual) words. OpenCV provides you interfaces for that, see

Guanta gravatar imageGuanta ( 2015-06-03 04:53:12 -0500 )edit