How to combine a key point detector and a key point descriptor to match images in MATLAB?

Question

I would like to try combination of 3 different key points detectors(Harris, MSER, DOG) and 3 different feature descriptors(SIFT, PCA-SIFT, GLOH) to local feature match 2 images in MATLAB. For instance, when matching two images I would like to try the combination of

• match_points(Harris, SIFT)
• match_points(Harris, PCA-SIFT)
• match_points(Harris, GLOH)
• match_points(MSER, SIFT)
• ...
• match_points(DOG, SIFT)
• ...

so that I can compare matching scores. I am thinking of using ration test to get matching scores.

I know that we can extract HOG features in MATLAB using;

[featureVector,hogVisualization] = extractHOGFeatures (img);

For MSER key points;

points = detectMSERFeatures(I);

For Harris key points;

corners = detectHarrisFeatures(gr);

However, I don't know how can I combine a key point detector and a key point descriptor such as described above in MATLAB?

Thank you.

Answer 1

Use extractFeatures. It supports SURF, BRISK, and FREAK descriptors.