Accuracy and reliability are two terms that are vital in a
biometrics system, which must also tolerate the fuzziness of the
biometrics characteristics to a certain degree. In this paper, we
propose and implement fingerprint image enhancement as a
preliminary stage to increase the accuracy and reliability of
minutiae extraction process. In this pre-processing stage, we
attempt to recover and enhance the corrupted and noisy region by
employing filtering technique. The enhance image is finally
transformed to its skeleton equivalent, preserving the ridges and
valleys connectivity for minutiae extraction process. Rutovitz
Crossing Number (CN) algorithms is then applied to extract the
candidate minutiae which will then undergo a series of minutiae
filtering processes to determine the validity of the extracted raw
minutiae as true minutia. The implementations of the minutiae
filtering processes are able to identify and eliminate the
predefined spurious minutiae. As we are focusing on extracting
accurate minutiae for the purpose of fuzzy vault implementation,
we also take into consideration the quantization of the minutiae,
which is an important factor in fuzzy vault locking and unlocking
procedure. Experiment results indicate that our implementation
methods are able to successfully achieve promising outcomes in
terms of accuracy and reliability in minutiae extraction.