Sensar_SecureTM Iris Identification System

The system uses an active vision paradigm. Customdesigned image pyramid-generatinghardware, developed in conjunction with Sarnoff Corporation, generates stereo disparity images in real-time, yielding the average distance to the subject’s head (figure 2). The depth map is also used to restrict the system’s attention to the closest person out of a possible queue of people. A face-template matching algorithm is then used to determine the (X,Y) position in image space of the subject’s eyes, even if the subject is wearing glasses (figures 3, 4). The eye position, along with the depth of the subject’s eye region (Z), is mapped to an (X,Y,Z) location in real 3-D space with calibration tables specific to each stereo camera pair. The eye’s 3-D location is used to control a pan/tilt mirror and a fixed focal-length lens to precisely center and focus a high-resolution image of the subject’s right or left eye [4]. The system locates the eye in the high-resolution image Figure 2. Stereo image and depth map