Palmprint Recognition with Fixed Number of SURF Keypoints

In this paper the palmprint recognition problem was studied using SURF (Speeded Up Robust Features) keypoints. The main setback when using SURF is that it computes, for the same set of parameters, very different number of keypoints for each image in the dataset, making the matching procedure inefficient. To solve this impediment, we use an algorithm that tunes a SURF parameter for approximately computing the same number of keypoints for all the images in the dataset. This idea was tested on five palmprint datasets. The palmprint recognition results using this approach are very good, better than those obtained in two recent papers.