A new concept on the continuously distributed directional image/field(CDDF) and the method to compute it in the fingerprint images are proposed

which exhibits not only good continuity

well gradualness

and excellent robustness to the noises

but very high precision

as well.Then

the classical formula to compute the Poincaré Index and the algorithm for the singularity detection are improved

so that the modified version of Poincaré Index can present not only the rotation degrees

but also the rotation direction of the vector in the vector field

exactly.Therefore

it is able to locate the singularities (core points

and delta points)at pixel level with an accuracy of only one pixel.Based on these

a novel fingerprint classification algorithm based on both the continuously distributed directional image and the modified version of Poincaré Index is developed finally

which classifies input fingerprints into 5 categories:arch

tented arch

left loop

right loop

and whorl.The experimental results obtained on the fingerprint database of Jiangsu University of Science and Technology demonstrate that this algorithm is invariant to image rotation of any degrees

and successfully solves the problem of image rotation

translation

and transformation in fingerprint classification.For the 4

000 images in this database

a classification accuracy of 97.05% for the five-class problem has been achieved.So it has better classification performance than previously reported in the literature.