Revision history [back]

Hi.

What do you mean by image model (an image template ?) ?

If your database is composed of a list of images and you want to retrieve this image in a query image (a bigger image that contains or not an image stored in the database), you could do:

• After the matching, you select the image in the database that contains the greatest number of matched keypoints (possibly plus a threshold on the minimum number of keypoints matched)
• Once you have the guessed image, you could compute the homography to locate it in the current image: http://docs.opencv.org/doc/tutorials/features2d/feature_homography/feature_homography.html#feature-homography

But this supposes that you have in your database for each image the list of keypoints (with 2D coordinates) and the corresponding descriptors.