Iris Recognition

Block diagram of iris recognition system developed in the DMCS Biometric Laboratory is presented in Fig. 1. First the image is acquired using the Iris Station device (see the description of the Iris Station project). Next the image is segmented to find iris region and normalize it. Then features of iris structure are extracted. Finally obtained feature vector is compared with iris signatures of registered users. Based on the comparison result the person’s identity is determined.

iris4

Fig. 1. Block diagram of the iris recognition system developed in the DMCS Biometric Laboratory.

Image segmentation

Block diagram of the eye image segmentation algorithm developed in the DMCS Biometric Laboratory is presented in Fig. 2. The segmentation process starts with reflections localization. Next, found reflections areas are filled in based on values of neighbor pixels. In the reflections filled in image, the iris boundaries modeled as non-concentric circles are found. Then the eyelids are found. Lower eyelid is modeled as circular arc, upper eyelid is modeled as line segment. Example result of the segmentation process is depicted in Fig. 3.

iris5

Fig. 2. Block diagram of the eye image segmentation algorithm developed in the DMCS Biometric Laboratory.

iris1

Fig. 3. Result of the segmentation process.

Once the iris region is found, it is transformed to pseudo polar coordinate system. This process normalizes iris image to make it convenient for further processing. Example result of the normalization process is depicted in Fig. 4.


a)

iris2

b)

iris3

Fig. 4. Result of the normalization process: (a) normalized iris image, (b) corresponding binary map indicating ROI (Region Of Interest).