Face recognition base on a new design of classifier with SIFT keypoints
Tong Liu, Sung‐Hoon Kim, Hyon-Soo Lee, Hyung‐Ho Kim · 2009
This paper investigates a new face recognition system based on an efficient design of classifier using SIFT (scale invariant feature transform) feature keypoint. This proposed system takes the advantage of SIFT feature which possess strong robustness to the expression, accessory, pose and illumination variations. One MLP (multi layer perceptron) based network is adopted as classifier of SIFT keypoint feature. The proposed classifier classifies each keypoint into face ID then an ID index histogram counting method is applied as the identification method to recognize face images. Also a bootstrapping method is investigated to select training images during training MLP. The performance of face recognition in some challenging databases is improved efficiently. Experiments on ORL and Yale face database show that the best recognition rate reaches 98% and 98.6%.