A Modification of Kernel-based Fisher Discriminant Analysis for Face Detection

This paper presents a modification of kernel-based Fisher Discriminant Analysis (FDA) for face detection. In face detection problem, it is important to design a twocategory classifier which can decide whether the given input sub-image is a face or not. There is a difficulty to train such tow-category classifiers because the “non face” class includes many images of different kinds of objects and it is difficult to treat them as a single class. Also the dimension of the discriminant space constructed by the usual FDA is limited to 1 for tow-category classification. To overcome these problems of the usual FDA, the discriminant criterion of the usual FDA is modifed such that the covariance of the ”face” class is minimized while the differences between the center of the ”face” class and each training sample of the ”non face” class are maximized. By this modification we can obtain a higher dimensional discriminant space which is suitable for “face” and “not face” classification. It is shown that the proposed method could outperform the support vector machine (SVM) by experiments of “face” and “non face” classification using the face images gathered from the available face database and the many face images on the Web.