| In this thesis, an automated biometric identification system was developed. The system was successfully designed to acquire and compare the profile along the length of the finger. Prior to the actual design of the system an qualitative study of the finger profile was done using image processing methods, to determine if there was actually any information obtainable from the finger using a CCD camera.; With these results we proceeded to design a more compact system that exploits the same biometric information in the finger, extracted using the image processing techniques. A 1-D photo detector array was used to capture the finger profile. This data pattern was first filtered through an analog signal conditioning circuit and then converted to digital information. This digital pattern was verified against another reference pattern using correlation-based software techniques, implemented on a MC68HC11 microcontroller system.; A statistical evaluation was done to assess the quality of the system. False rejection rates and false acceptance rates were determined for various thresholds. Optimal results were seen for a threshold of .94 correlation value, for which the system achieved a 90% success rate. Further analysis of the data patterns was done to understand why the system could not achieve a higher success rate. |
