Research On Feature Extraction For Low Quality Fingerprint Image

Fingerprint characteristics was the most reliable and stable characteristics in the biological characteristics, occupy the important position of biological recognition. But because of outside noise, acquisition device technology, there is low quality fingerprint image, thus affecting the fingerprint identification effect. To reduce the noise influence on fingerprint identification pretreatment, the process of fingerprint image is need, such as fingerprints image enhancement.This paper look up a large number of literature, and analyze the fingerprint identification technology at home and abroad, according to the current research in fingerprint image preprocessing, fingerprints, feature extraction and matching. This paper main research fingerprint image enhancement, feature extraction, features, etc.The paper analyzes the classic Gabor filter's method, and find that the filter exist that enhancement effect is different in all directions, block uneven enhancement effect. By understanding of the transformation of Gabor filter, made new improvements in filter parameter, the width of the filter. Make filtering effect more optimized for the feature extraction and matched.According to distortion fingerprint problems and its characteristics, this paper make a method based on fingerprint image center with curvilinear coordinates, to extract fingerprint feature and match fingerprint image. In the same Gabor filtering processing, compare to the rectangular coordinate matching method, the EER value dropped four percentage points. Experiments show that, based on fingerprint image center with curvilinear coordinates, the method of the fingerprint image matching distortion of obvious effect, can overcome the impact of fingerprint distortion, and achieved good effect.