There are several stages in all biometric recognition systems
[1]. Accordingly, fingerprint recognition systems comprise basic
stages listed as fingerprint image acquisition, image segmentation
and normalization, image enhancement, orientation field (OF) estimation,
feature extraction and fingerprint matching. Each of them
might comprise sub-stages or pre and post-processes. A typical
fingerprint identification system is illustrated in Fig. 1.
All fingerprint recognition systems include a key stage called
matcher. The function of this stage is of paramount importance
in finding the same fingerprint impressions. Fingerprint matching
has been approached by various strategies, such as image based
[2], ridge pattern based, point (minutiae) based [3], texture structure
based as a statistical approach [4] and graph based schemes
[5]. Among these methods, minutiae based approach is more reliable
and robust and therefore it is more conventional. However, the
accuracy of this method strictly depends on the quality of fingerprint
images, they are unable to tolerate large amounts of nonlinear
distortion in the fingerprint ridge structures. On the other hand,
all previous matching techniques rely on fingerprint ridge orientation
extraction. Fingerprint orientation estimation gives important
information about ridge and texture pattern of fingerprint images.