Abstract
Fingerprint recognition system is one of the more popularand reliable applications used security access and E-commerce. Minutiae-based fingerprint matching is the most well-known and widely used method. Most two prominent features of minutia are ridge ending and ridge bifurcation in many minutiae-based fingerprint recognition system. General, fingerprint recognition system consists of fingerprint image acquisition, preprocessing, thinning, postprocessing, fingerprint feature extraction, andfingerprint matching. However, the identification performance of fingerprint recognition system on the different quality of extraction of minutiae is very sensitive. Spurious minutiae being produced, valid minutiae being lost, and the feature of minutiabeing wrongly labeled are resulted in caused the undesirable result. Hence, poor quality of the actual minutiae feature of a fingerprint image constitutes the single most cause of poor accuracy performance of a fingerprint recognition system. Therefore, it is important to quantify the quality of minutiae for either seeking a better representation of the minutia or for subjecting the poor minutia to alternative methods of processing (e.g., matching). In this thesis, we explore a definition of quality of extracted minutiae and present detailed algorithms to measure minutiae quality. The proposed quality measure has been developed with the use of fingerprint image processing properties and the alignment of orientation field. The goal of the thesis is to obtain the value of the minutia quality quantified by above method. Based on value ofthe minutia quality, we can express clear its stability. For security, we combine information of the minutiae quality with the minutiae-based fingerprint matching techniques existed to make the matching more powerful. In experiments, it is shown that the matching error rate is reduced compared with uncombined the minutiae quality method, and their performance is evaluated using a FVC2002 dataset.