Problems implementing LSH with SIFT feature

Question

I have a problem in ANN Search when using LSH with SIFT Feature. With some feature detector tool (Sift Demo) or some available dataset, I received descriptors with 128 dimension for an image. But I don't know how to store it into a .mat file (database) and query vector to received k-closest images from query image.

Please help me. Thanks a lot.

Answer 1

Have a look at the Matlab demo provided by David Lowe You can save the descriptors in .mat file. match.m shows how matching is done.