| Good face detection is an important part of a reliable face recognition system. We propose methods of face detection and normalization. Normalization requires the exact location of eyes to straighten any rotated face image. Eyes detection methods are also developed for this purpose. To improve the accuracy of the classical PCA technique for face recognition, we developed a system based on the method of least-squares which takes into account the class information. We further improve accuracy by using multiple scales of the same face image. Such scales are obtained using the discrete wavelet transform. We show that there exists a particular set of subband images that are essential for robust face recognition. We propose here a model to find these subbands for input images. Based on this, we develop a multiscale face recognition system which utilizes the model to find important subbands. These subbands are then used for improving recognition accuracy. |
