A New Bit-Plane Extraction Approach for Fingerprint Recognition Using Phase-Only Correlation Function
Fingerprint patterns are virtually unique and make human identifiable. Based on this fact, many researches have been conducted on fingerprint to create new or improved recognition systems. Due to the rapid growth of human population, larger capacity of databases are needed for the storage of fingerp...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English |
| Published: |
2016
|
| Subjects: | |
| Online Access: | http://ir.unimas.my/id/eprint/30526/1/A%20New%20Bit%20Plane%20Extraction%20For%20Fingerprint%20Recognition%20using%20Phase%20%2024pgs.pdf http://ir.unimas.my/id/eprint/30526/4/Florence.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Fingerprint patterns are virtually unique and make human identifiable. Based on this fact, many researches have been conducted on fingerprint to create new or improved recognition systems. Due to the rapid growth of human population, larger capacity of databases are needed for the storage of fingerprints. With larger sizes of fingerprint images, bigger memory storage is needed and the cost for such memory is also higher. The focus of this thesis is on bit-plane extraction of a fingerprint image, which could reduce the file size of a grayscale image. Literature studies reveal that this method is commonly used in image compression and retrieval with limited research in fingerprint recognition. Combination of this method with Phase-Only Correlation (POC) function reduces computation time for a fingerprint recognition while improve the recognition rate. POC based recognition method is adopted because of its simplicity and ability to achieve high accuracy compared to other recognition algorithms, especially for low quality fingerprint image. There are two fingerprint databases used for evaluations, which are FingerDOS and a benchmark database, FVC2002-Db1a. A number of analyses are conducted to examine factors that might reduce the performance of the recognition system. Different rotation angles and cropping dimension are applied on the fingerprint to investigate the effects of misaligned fingerprints. The best bit-plane is selected based on the highest recognition rate produced by the system. From the experiment results, bit-plane 7 has the highest performance compared to other bit-planes. Since fingerprint recognition by using bit-plane image could perform as good as grayscale image, this system has the potential to reduce the storage requirement for fingerprint database. |
|---|