Face recognition using feature extraction based on descriptive statistics of a face image

This paper proposes a new method of feature extraction for face recognition based on descriptive statistics of a face image. Our method works by first converting the face image with all the corresponding face components such as eyes, nose, and mouth to a grayscale images. The features are then extracted from the grayscale image, based on a descriptive statistics of the image and its corresponding face components. The edges of a face image and its corresponding face components are detected by using the canny algorithm. In the recognition step, different classifiers such as Multi Layer Perceptron (MLP), Support Vector Machine (SVM), k -Nearest Neighbors (k-NN) and Pairwise Opposite Class-Nearest Neighbor (POC-NN) can be used for face recognition. We evaluated our method with more conventional Eigenface method based upon the AT&T and Yale face databases. The evaluation clearly confirm that for both databases our proposed method yields a higher recognition rate and requires less computational time than the Eigenface method.