Search code examples
performancealgorithmopencvsift

Reduce time needed for SIFT Brute-force Matcher Algorithm in OpenCV

I've been wondering how can SIFT Brute-force Matcher (for object identfication) time be reduced. My program iterates over some trainsets of images and returns the maximum number of matches found in each case. Is there a way (except for resizing the images to a smaller scale) to shorten this time interval because I think that the greater becomes my database it will be rendered useless if I have a huge identification time?

Thanks in advance

PS I use Python Bindings for OpenCV 2.4.9

Solution

  • why dont you try flann matcher...... thats much faster than bruteforce......

    http://docs.opencv.org/trunk/doc/py_tutorials/py_feature2d/py_feature_homography/py_feature_homography.html#py-feature-homography