Face Recognition in Automated Teller Machines

摘要

As the popularity of the Automated Teller Machines (ATMs) has increased as a result the crimes related to it have also increased. In order to overcome this problem one of the most popular approach is it install a camera in ATMs so that faces can be identified for criminal follow-up. In real time situations there might be several occlusion situations in the facial images of the users. Hence, in this paper we are proposing a method which can identify the faces of the users even with the occlusions. Initially we developed a facial recognition system for whole face image based on Eigen vectors. The method is further improved by using the concept of Region Of Interest (ROI) by applying DWT and DBC algorithm. We capture an image and then Discrete Wavelet Transform (DWT) is applied which derives LL, LH, HL, HH sub bands of the images. This is done by concept of splitting the frequencies by using low pass and high pass filters. Then Directional Binary Code (DBC) is applied sub bands of images, then Euclidean Distance (ED) is found to compare the features. Once the image recognition is done then it will be verified and the authorized user can access the account.