A Novel Classification based on Ellipse for Face Recognition
Journal of Information Hiding and Multimedia Signal Processing
This paper, we presented a novel classification based on the nearest and the center of ellipse (NCE) for face recognition. Any ellipse is generated by three points, which are obtained from the combination of m distinct points taken from the same class: m is a number of image per class. The classification criteria are as follows: 1) if the test image is inside or is on the ellipse, we calculate the distance between the test image and the center of the ellipse; 2) if the test image is outside the
... ellipse, we find the minimum distance. The distance between the test image and the z point such that the z point is any point in the three points in the ellipse then the test is in class C NCE when the distance that is the shortest of all of classes. A large number of experiments were investigated on the Faces94 and the Grimace database. Meanwhile, the comparison between the NCE and the other neighborhood-based classification methods; including the nearest feature line, the shortest feature line segment, and the restricted nearest feature line with ellipse.