engleski

Iris print extracting from reduced and scrambled set of pixels

Iris recognition is a form of biometric identification based on unique patterns within the region that surrounds the pupil of the human eye. This technology uses pattern recognition algorithms in person identification procedure, and requires high-resolution images. The confidentiality of the stored data is an important issue, and the access to personal data should have only an authorized person. This paper focuses on iris recognition procedure using reduced and scrambled set of data. Namely, the number of available pixels from an iris image is reduced intentionally, according to the Compressive Sensing principles. Therefore, we deal with much less data than it is usually required. Scrambling is done according to the unique key, in order to provide secrecy of the data. When it comes to the identification, the set of received pixels is firstly reordered in original positions, and the rest of the pixels (the missing ones) are recovered by using the CS reconstruction algorithm. The theory is verified with the experimental results.

biometrics, Compressive Sensing, iris recognition

nije evidentirano