Page 27 - 2018

used the most widely adapted Gradient based orientation estimation and also enhanced
it. The performance is done using comparison of existing gradient based method with
enhanced gradient based method and prove that proposed enhanced method give better
result.

Third major challenge in the fingerprint is makes the ridge ends and ridge bifurcation
more visible and distinguishable from the image. The binarization and thinning process is
used for that purpose. The binarization is the way which convert the gray-scale image
into binary image means black and white image which distinguish the fingerprint image
in ridge part and non-ridge part. In proposed research work apply global threshloding,
famous otsu thresholding and local adaptive thresholding using different matrix size like
5x5, 9x9, 15x15. The comparison above method prove that local adaptive thresholding
apply with 9x9 matrix size give better performance.

Fourth major challenge is computational time, the thinning process is used to reduce the
size of pattern width up to 1-pixel and thin the image which is become appropriate for
minutiae extraction. As well as it reduce the execution time for feature extraction process.
In proposed research work the zhnag-suen’s algorithm is implemented and also
enhanced it. The performance of enhanced zhang-suen’s algorithm is compared with
existing zhang-suen’s as well as Hilditch’s algorithm. The comparison prove that the
enhanced zhang-suen’s algorithm give best result.
The thinned skeleton image is become input image for extracting the minutiae points. In
proposed work extract ridge ending and ridge bifurcation as minutiae. The minutiae
points are extracted from the skeleton image using Rutovitz Crossing Number (CN)
method.

Fifth major challenge is remove the spurious minutiae which are extracted at the time of
feature extraction stage. In proposed research work used some condition to remove that
spurious minutiae.

A two final major challenges in fingerprint recognition are most important that is
reducing false positive during minutia detection and matching of unequal number of
minutia features. This problems are arrived because of rotation of fingerprint. To,
overcome these problem used core point detection using famous Poincare Index method
as well as enhanced it to found the exact core point. Then after in proposed work apply
minutiae based approach for matching fingerprint images. In that combine two
approaches: approach based on alignment and approach based on local minutiae: radius:
50px from central minutiae, which reduce false positive minutiae matching. As well as it
provide better results while comparing minutia sets of different sizes as well as in slightly
different orientation during the matching process. The experimental results on the

08