TY  - THES
T1  - An artificial neural network based human fingerprint classification system a comparative study
A1  - Galicia, Eric Paolo T.
LA  - English
YR  - 1996
UL  - https://tuklas.up.edu.ph/Record/UP-99796217602381186
AB  - This thesis seeks to compare the performance of three artificial neural network architectures with respect to the problem of fingerprint pattern-level classification. This comparison is limited by an observation based on 30 sample points. The observation domain was limited to 30 sample points due to the penalty of long training times needed by the artificial neural networks. Pattern-level classification for fingerprints is the first step for full classification according to the Henry System classification method which is currently being used by most law-enforcement agencies. The different artificial neural network architectures were compared according to accuracy and recall times. The image preprocessor included in the Pattern Level Classification Automation System for Fingerprints (PCASYS), a project by the National Institute of Standards and Technology (NIST) of the United States, was used to provide uniform inputs for the three architectures. Also, the Probabilistic Neural Network used in PCASYS was chosen as the benchmark upon which the other two architectures will be tested upon. As such, the two network architectures studied and simulated were the Backpropagation network with a Sine activation function and the Neural Tree Network.
NO  - Computer printout.
CN  - LG 995 1996 E6 G35
KW  - Artificial intelligence.
KW  - Fingerprints.
KW  - Neural networks (Compouter science).
KW  - Pattern recognition systems.
ER  -