Fingerprints

recognition
using
neural
network
Politecnico di Milano Polo Regionale di Como
Methods and Technologies for Image Processing

Author: Alessandro BAFFA 682075

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 1

 Agenda

 Introduction
 Features of fingerprints

The pattern recognition

system Why using neural

network?
 Preprocessing
The goal of this method
system
 Feature extraction and selection
 Invariant recognition
 Result
 References
Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 2
 Features of fingerprints

Fingerprints are imprints

formed by friction ridges of
the skin in fingers and
thumbs.
 Their pattern are
permanent and
unchangeable on each
 finger during all the life;
They are individual (the
probability that two fingerprints
are alike is about 1 in
1.9x10^15)
 They have long been used for
identification

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 3

 The pattern recognition system

 Image acquisition converting a scene into an

array of numbers that can be manipulated by a
computer
 Edge detection and thinning are parts of the
preprocessing step which involves removing noise,
enhancing the picture and, if necessary,
segmenting the image into meaningful regions

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 4

 The pattern recognition system

 Feature extraction in which the image is represented by a

set of numerical “features” to remove redundancy from the
data and reduce its dimension
 Classification where a class label is assigned to the
image/object by examining its extracted features and
comparing them with the class that the classifier has
learned during its training stage.
 The main focus of this method is on these two last
parts

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 5

 Why using neural network?

 Neural network enable solutions to be found

to problems where algorithmic methods are
too computationally intensive or do not
 exist
The problem of feature extraction
and classification seems to be a
 suitable application for neural nets.
They offer significant speed advantages
over conventional techniques

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 6

 The goal of this method

 This proposed method is based on a data

model for fingerprints that is structural
rather than coordinate.
 This structural data model is robust with
respect to traslation, rotation and
distortion

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 7

 Preprocessing system

The first phase of the work is to capture the

fingerprints image and convert it to a digital
representation of 512x512 by 256 gray
levels.

 Histogram equalization technique is used

to increase the contrast if the illumination
condition is poor

But we are only interested in binary

information

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 8

 Preprocessing system

Binarization is usually performed by

using Laplacian edge
detection operator
 Local derivative operator such as
“Roberts”, “Prewitt” or “Sobel”
 Thresholding tecnique

The binary image is further enhanced

by a thinning algorithm which
reduces the image ridges to a
skeletal structure

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 9

 Preprocessing system

 The thinning algorithm while

deleting unwanted points should
not:
 Remove end points
 Break connectedness
 Cause excessive erosion of the
region
After obtaining the binary form of the
fingerprint image, there may be some
irregularities caused by skinfolds and
contiguous ridges or spreading of ink due to
finger pressure, and so on..

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 10

 Preprocessing system

 To remedy this problem, smoothing

is necessary and includes:
 Filling holes
 Deleting redundant points
 Removing noisy points
 Filling potential missing points

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 11

 Feature extraction and
selection
 Selection of good feature is a crucial step in the
process since the next stage sees only these features
and acts upon them.
 150 different minutiae type have been identified but in
practice only ridge ending and ridge bifurcation are
used.

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 12

 Feature extraction and
selection
 Good features are those satisfying two
requirements:
 Small intraclass invariance (i.e. slightly
different shapes with similar general
characteristics should have numerically
close values)
 Large interclass separation (i.e. features
from different classes should be quite
different numerically)

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 13

 Feature extraction and
selection
 A multilayer perceptron network of three layers
is trained to detect the minutiae in the thinned part
image of size 128x128
 The first layer has nine units associated with the
components of the input vector
 The hidden layer has five units
 The output layer has one unit corresponding to the
number of the classes

The network is trained to output ‘1’ when the input

window is centered on the feature to be located and it
outputs ‘0’ if minutiae are not present

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 14

 Feature extraction and
selection
 the network is trained by using
the backpropagation learning
technique and the weight
change is updated according to

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 15

 Feature extraction and
selection
 The trained network is then used to analyze the
complete image by raster scanning the fingerprint via
window of size 3x3
 In order to prevent the falsely reported features and
select “significant” minutiae, two more rules are
added to the system to guarantee perfect ridge forks
are detected while excluding all other features:
 At those potential minutiae feature points we examine
them by increasing the window size to 5x5
 If two or more minutiae are too close togheter, we
ignore all of them

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 16

 Feature extraction and
selection

 Distribution of minutiae of two

identical fingerprints 2(a) before and
2(b) after applying the rules

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 17

 Invariant recognition

 The location of a reference point of the fingerprints is

important for invariant recognition and has to be
determined
 Contour tracing is used to find one or more
turning points (i.e. points with maximum rate of
change of tracing movement)
 This points are then used to find the reference
point

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 18

 Invariant recognition

 The Euclidean distance d(i) from each feature point i

to the reference point are calculated
 The distance to the center confers the property
of positional invariance
 The data are then sorted in ascending order
from d(0)
to d(N)operation gives the data the property of
 this
rotational invariance
 In order to make the data becomes invariant to scale
change, it is normalized to unity by the shortest
distance d(0), i.e. dist(i) = d(0)/d(i), i = 0..N
 This will weight those feature points nearer to the center
more heavly because usually these points are more
significant in classification.

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 19

 Invariant recognition

 The centroidal data patterns should be

shift, scale and rotational
independent
 Also the invariant feature vectors are
in the range [0,1] and they can be
directly used as the training/stored
vectors in the MLP classifier

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 20

 Result

 The recognition rate of fingerprints

depends much on the quality of the
fingerprints and effectiveness of the
preprocessing system level used in edge
 Such as the thresholding
detection
 If there are too many broken lines or noisy
points in the image, the preprocessing system
contour tracing may fail.
 An intelligent connection algorithm to recover
broken lines and suppress spurious irregularities is
necessary

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 21

 Reference
s
 W.F. Leung – S.H. Leung, W.H. Lau –
Andrew Luk, Fingerprints recognition
using neural network
 M.T. Leung – W.E. Engeler – P. Frank,
Fingerprints image processing using
neural network
 Jacques de Villiers – Etienne Barnard,
Backpropagation neural nets with one
and two hidden layers
 Andrew Luk – S.H. Leung – C.K. Lee – W.H.
Lau, A two level classifier for fingerprint
recognition

Alessandro Baffa 682075 - "Fingerprints recognition using Neural Network" 22