Real-Time Template Based Face and Iris Detection on Rotated Faces

Abstract

Real-time iris and face detection on video sequences is important in applications such as study of the eye function, drowsiness detection, man-machine interfaces, face recognition security and multimedia retrieval. In this work, we present a real-time template based method for iris detection in faces with wide coronal (− 40°, + 40°) and transversal (− 45°, + 45°) axis rotations. This method is based on anthropometric templates that were constructed off-line for face coronal and transversal rotation, using face features such as elliptical shape, location of the eyebrows, nose and lips. A line integral is computed using these templates over the fine directional image to find the actual face location, face size and rotation angle. This information provides a region to search for the eyes and the iris boundary is detected. Results computed on five video sequences including coronal and transversal rotations with over 1,700 frames show correct face detection rate of 98.5% and iris detection rate of 94.4%. The method was compared with a “weighting mask method” on two video sequences showing an improved performance. The method was also compared for eye detection to a method using combined binary edge and intensity information in two subsets of the AR face database (63 and 564 images). Different disparity errors were considered and for the smallest error, a 100% correct detection was reached in the AR-63 subset and 99.8% was obtained in the AR-564 subset.