A Novel Approach for Fingerprint Singularities Detection

Conference Paper (PDF Available) · October 2017with 130 Reads 
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
DOI: 10.1109/WVC.2017.00025 ·
Conference: 2017 Workshop of Computer Vision (WVC)
Cite this publication
Abstract
Fingerprints are an important biometric trait, being the most widely deployed biometric characteristic. Singularities detection represent an important step in fingerprint recognition due to plenty of comparisons required to identify a person, especially in large datasets. In this work, we propose a novel technique for fingerprint singularities detection. The input image is preprocessed and our singularity detection technique takes advantages of new ridge patterns found in the orientation image. We achieved a singularities detection rate of 93.42%
Advertisement
A Novel Approach for Fingerprint
Singularities Detection
Paulo R. P. Silva, Arnaldo G. A. Silva, João J. B. Primo, Leonardo V. Batistaand Herman M. Gomes
Informatics Center
Federal University of Paraíba
João Pessoa, Paraíba
Email: {pauloricardo, arnaldo.gualberto, leonardo}@ci.ufpb.br
Informatics Center
Federal University of Campina Grande
Campina Grande, Paraíba
Email: {joaojanduy, hmg}@computacao.ufcg.edu.br
Abstract—Fingerprints are an important biometric trait, being
the most widely deployed biometric characteristic. Singularities
detection represent an important step in fingerprint recognition
due to plenty of comparisons required to identify a person,
especially in large datasets. In this work, we propose a novel
technique for fingerprint singularities detection. The input image
is preprocessed and our singularity detection technique takes
advantages of new ridge patterns found in the orientation image.
We achieved a singularities detection rate of 93.42%.
Keywordsbiometrics, fingerprint, singularity, detection, classi-
fication
I. INTRODUCTION
Biometrics (biometric recognition) refers to the use of
distinctive anatomical and behavioral characteristics such as
fingerprints, iris, face and voice for automatically recognizing
a person. Biometrics provides better security, higher efficiency,
and, in many instances, increased user convenience. Because
biometric identifiers cannot be easily misplaced, forged, or
shared, they are considered more reliable for person recogni-
tion than traditional token or knowledge based methods. Thus,
biometric recognition systems are increasingly being deployed
in a large number of government and civilian applications [1].
Because fingerprints are unique to individuals, invariant
to age, and convenient in practice [2], they are the most
widely deployed biometric characteristics. Fingerprint are used
in every forensics and law enforcement agency worldwide rou-
tinely [1]. However, the recognition of an individual requires
the comparison between his fingerprint with all fingerprint
previously stored in the database. In large datasets, such
process becomes infeasible as it may demand a high response
time. The identification process can be improved by decreasing
search space, so fewer comparisons and iterations might be
performed.
A common approach for search space reduction is to
analyze the fingerprint with respect to its global features (i.e.,
ridges and singularities) [3] [4]. Such global features are used
to classify fingerprint in five different classes, namely: arch,
tented arch,left loop,right loop, and whorl [5]. Generally,
the identification process is then performed with fingerprints
belonging only to the same class.
Fig. 1. Singular regions (white boxes) and core points (small circles) in
fingerprint images. Source: [1]
The most important global features are the core and delta
points (also referred to as singular points or singularities). The
core point is defined as the topmost point on the innermost
ridge and a delta point is defined as the point where three
flows meet [6] (see Fig. 1). These points are highly stable and
also rotation and scale invariant. Thus, the number and the
location of core and/or delta points are widely used by most
classification methods [7].
On the other hand, the location of core and delta points
are still an open problem due to a list of factors: the quality
of fingerprint images may be poor; the presence of noise
introduced by sensors; fingerprints with only a partial image;
the presence of scars, breaks, too oily or too dry. All of these
factors make it extremely hard to classify fingerprint images
in terms of singular points [7].
In this paper, a novel singularities detection technique is
presented. The main contribution of this work refers to a
new way to analyze the fingerprint orientation image. We
also compare the proposed approach with the algorithms of
SPD2010 competition and other published works. Beyond
singularities detection, the proposed method may be mainly
used to: create new features for fingerprint matching, and
improve indexing of fingerprint images.
II. RELATED WORKS
Many works related to singular point detection or classi-
fication have been proposed in the literature. Most of them
work on the directional image to compute singularities’ loca-
tion and type. Furthermore, singularities extraction algorithms
102
2017 Workshop of Computer Vision
0-7695-6357-0/17/$31.00 ©2017 IEEE
DOI 10.1109/WVC.2017.00025
are usually based upon Poincare index (Plindex), orientation
partitioning, template convolution, the direction of curvature
and mathematical models.
A. Plindex-based methods
Poincare index-based method is the most popular algo-
rithms for singular point detection. Poincare index (a vectorial
field wrapped by a curve) was first used to extract singular
points by Kawagoe e Tojo [8], in 1984. For each pixel’s
neighborhood, the sum of differences between the angles is
computed using a sliding window over the directional image
to estimate singularities’ type and location.
The authors in [9] proposed a modified Poincare index
technique. In the first stage, the input fingerprint image is pre-
processed and, in the next step, fine orientation field estimation
is performed. Finally, the singular points are located using the
modified Poincare index technique described by the authors.
The Plindex is also used by other works in the singularities
detection stage, as can be seen in [10], [11].
Even tough such methods are simple, they may lead to the
detection of false singularities in noisy or low-quality finger-
prints. Additional preprocessing (smoothing the orientation) or
post-processing (combining other information such as quality
check and segmentation) is required to reduce false detections
[12].
B. Directional image partitioning methods
Some algorithms create clusters by grouping similar orien-
tations from the directional image. The intersection between
clusters establishes the singular point location, as shown by
[13], [14], [15].
Punnet and Phalguni [15] proposed a method to compute
a cluster through partitioning of a directional image. A post-
processing step is applied for fine singularities location and to
detect undiscovered deltas. The cluster is validated by Poincare
index analysis.
The method defined in [16] has three main steps. Firstly, the
fingerprint image is preprocessed for background segmentation
and orientation field calculation. Secondly, core and delta
points are clustered using an improvement to Poincare index
method proposed by the authors. Finally, the Gaussian-Hermite
moment distribution of candidate singularity’s surrounding
neighborhood is computed to remove false positives. Using
a sample of 100 randomly selected images from NIST-4
database, the authors achieved accuracies of 95.35% and
85.90% for core and delta points, respectively.
The pattern in fingerprint local ridge orientation map is
analyzed by [17]. The authors segment the orientation values
into 4 areas, where singular points are defined as points
connecting all the different orientation segments. A further step
is performed to recognize core and delta points and remove
false detections. Although the results have shown an average
correct detection rate of 94.05%, their method may not work
properly if a fingerprint is rotated too much.
Such methods also require high-quality images. Noisy
images may cause false intersection points, thus false singular
points.
C. Template-based methods
In template-based methods, a filter (template) is convolved
over each pixel in a fingerprint image to extract singularities
[18], [19].
In [12], convolutional masks are applied to a fingerprint
image in order to compute gradients, derivatives, and angles
for each pixel. Such information is then used for singular point
extraction as described in the paper. The method is applied
to 75 fingerprint images randomly selected from FVC 2004
(DB4) database. The authors report missing rates of 7.14% and
7.69% for core and deltas, respectively. In addition, 5.71% and
7.69% of false core and deltas are generated by the method.
Complex filters are used by [20] for automatic extraction
of singularities in fingerprint images. The filtering is applied
to the orientation field in multiple resolution scales estimated
from the global structure of the fingerprint, i.e. the overall
pattern of the ridges and valleys. The DB2 set of FVC2000
database is used to validate the experiments (100 persons with
8 images/person). The correct detection rate is nearly 95%.
However, only core points are detected.
A method for singular point detection using a bank of
discrete Fourier filters is presented in [21]. The advantages
of Poincare index are firstly applied for selection of candi-
date blocks in orientation map. A set of 90 discrete Fourier
filters are then convolved with the orientation image and
the responses are inspected for singular point detection. By
using Fourier transform, the time-consuming problem of Gabor
filters is avoided. The experimental results showed that the
bank of filters took about 0.02 seconds for evaluation, instead
of 12 seconds by Gabor transform.
D. Curvature-based methods
The algorithms based on the curvature of the singular
points’ orientation report considerable results to detect singular
points, as acute gradient changes are found around singularities
location [22]. However, such methods are not robust to noise
and false singularities are detected.
The work of [23] proposes a novel approach for fingerprint
singular points detection based on orthogonal theory. In the
first step, the input image is normalized to remove the effects
of sensor noise and deformation due to the finger pressure
differences. Then, the orientation field and double orientation
field are estimated to detection of core and delta points. Only a
reduction in computational costs and complexity are reported.
The authors in [24] described a hybrid technique for sin-
gular point detection based on the Direction of Curvature and
Poincare Index. The modified DC (called Enhanced Detection
of Curvature, EDC) technique is used to estimate the core and
delta point in a quicker manner and with low computational
cost. On the other hand, the Enhanced Poincare Index (EPCI)
provides a more accurate location of core and delta points. The
paper reports 98.5% of accuracy.
E. Mathematical models
Mathematical models to detect singularities can also be
found in the literature. Usually, such models attempt to de-
scribe core and delta points and define its location by a
103
Fig. 2. An overview of the proposed method
mathematical formulation. Some papers related to this kind
of approach are described below.
A method combining singular candidate analysis with an
extended relational graph is proposed in [25]. Both local
and global features of the ridge distribution are extracted for
singularity detection. Such features are used to estimate the
singular points probabilities. According to the authors, the
algorithm is robust to local noise. The method is evaluated
in FVC2000 and FVC2002 databases. The accuracy reported
ranges from 80.8% up to 97.8%.
The work of Jin Qi e Suxing Liu [26] presents a method
based on a complex polynomial model (called Zero-pole
model) and a sliding window to detect singular points. Zero-
pole model of the directional image is essentially a complex
rational polynomial which zero and pole are considered either
singular points loop or delta, respectively. The method achieves
38% of correct detection rate in the SPD2010 test dataset (290
images with 297 cores and 144 deltas in the test set). Other
approaches using zero-pole model has also been proposed [27],
[28], [29].
III. METHODOLOGY
The proposed method is based upon 2 main steps: pre-
processing and singularities detection. The Fig. 2 shows an
overview of our method.
A. Preprocessing
A critical step in singularities detection is digital image
preprocessing due to the poor quality and the presence of noise
(e.g. spurious minutiae) in some fingerprint images. The main
goal of preprocessing is to improve image quality in order to
assist the progress of later steps.
Firstly, our preprocessing technique applies an histogram
equalization followed by expansion to improve image quality.
Then, the region of interest (ROI) is extracted as in [30] and
enhanced by iterative Gabor filters [31]. Finally, the orientation
image is computed using bi-dimensional gradient vectors [1].
B. Singularities Detection
Our singularities detection method receives 3 input images:
ROI, enhancement, and orientation. The main contribution of
our technique is related to novel ridge patterns found in the
orientation image. Instead of analyzing the orientation image
by its angle values, we treat the orientation image purely as
Fig. 3. The new pattern of ridges. Left column: two samples of original
images from SPD2010 test database. Right column: region of interest with
normalized orientation.
an image. The angles computed by the bi-dimensional gradient
vectors are used as image pixels, but the range is normalized
from [0-180] to [0-255]. We observed that singularities can be
found at the end of lines as can be seen in Fig. 3. According to
our knowledge and previous research in literature, none other
paper has used this approach before, proving the originality of
our method.
The visible lines in Fig. 3 befall because of the abrupt
orientation change between angles close to zero and angles
close to π. Moreover, at the end of these lines, the angles are
close to π
2. Two important advantages of this method can be
cited: the ability to detect singularities in latent fingerprints
(see Fig 4 (c)); and the robustness to rotated fingerprint (see
Fig 4 (a) and (b)). We observed this pattern remains even in
fingerprints without singularities. In these cases, there is no
line ending away from ROI border, and thus, there are no
singularities.
To detect singularities, the Canny filter is applied to high-
light edges of the lines found in the orientation image normal-
ized as described above. The resulting image is then dilated
104
Algorithm 1 Endings Extraction
1: I[][] image
2: endings_set ←{}
3: DX[] ←{0,+1,+1,+1,0,1,1,1}
4: DY [] ←{1,1,0,+1,+1,+1,0,1}
5:
6: for i8, I.width 8do
7: for j8,I.height8do
8: if I[i][j]==Black then
9: neighbours 0
10: for k0,8do
11: if I[i + DX[k]][j + DY[k]] == Black then
12: neighbours neighbours +1
13: if neighbours == 1 then
14: endings_set ←{(i, j)}
TABLE I. SPD2010 DATASET DESCRIPTION
Images Cores Deltas
Train 210 240 92
Test 290 297 144
Total 500 537 236
to prevent noncontinuous lines and the thinning operation is
performed to reduce all lines to single pixel thickness. Finally,
we run the Algorithm 1 to detect singularities. The algorithm
searches for a black pixel surrounded with only one other black
pixel in its 8-neighbourhood, representing a line ending. All
line endings are obtained, but only the points located away
from ROI border are considered.
IV. EXPERIMENTAL RESULTS
We evaluate our experiment using the publicly available
SPD2010-Train and SPD2010-Test datasets [32], which were
previously used for a singularity detection competition in 2010.
A summary of this dataset can be seen in Table I. Cores
and Deltas were manually labeled and follow Henry’s singular
point definition [5] to determine its locations. This competition
has its own evaluation metric which we followed to compare
with the algorithms published. In addition, our approach is also
compared with the work OFθ [15].
Table II presents the results achieved by our approach in
both train and test dataset given by the SPD2010 competition.
In total, we detected 724 out of 773 all singularities presented
in the dataset (93.66%). We can also observe the low difference
between the metrics between train and test dataset, which
proves the robustness of our method.
We conducted an errors analysis to understand both miss
and false alarm rates in the proposed approach. The most of
the errors were related to singularities definitions. First, our
method detects loops in the curve just below or above the in-
nermost ridge where the ground truth is located (see Fig. 6(a)).
TABLE II. RESULTS OF THE PROPOSED APPROACH ON SPD2010
DATA SET
Dataset Precision Recall F-score Accuracy
SPD2010-Test 0.90 0.92 0.93 93.42%
SPD2010-Train 0.94 0.94 0.94 93.98%
Thus, 212 loops were wrongly detected since the distance was
greater than the competition’s acceptance threshold (10 pixels)
(see Fig. 6(b)). On the other hand, the proposed method always
detect 2 loops for singularities of type whorl, but only one
is labeled. Although it represents correct detections, 25 extra
loops were classified as false alarm according to ground truth
dataset. Secondly, we faced problems with singularities too
close or even away to ROI border (10 deltas and 1 loop),
also reported in the literature [15] (Fig. 5). Finally, the poor
quality images presented in SPD2010 dataset affected our
preprocessing steps: either the segmentation removed the areas
where singularities were located or the noise disturbed the
emergence of line endings (Fig. 7). Nevertheless, the problem
cited are related mainly to the preprocessing step. Therefore,
if we improve such step only, the proposed approach may be
improved by a large margin and accomplish equal or even
better results than compared papers.
The comparison between our approach and both methods
of Punnet and Phalguni can be seen in Table III. The results
showed our method achieved similar results to the compared
algorithms. In comparison to OFθ, we achieved better results
for almost all metrics, except for detection recall of deltas.
Otherwise, only our F-score for detection of deltas was better
than OF ¯
θ, although the other metrics had comparable results.
However, Punnet and Phalguni made some changes to the
original dataset: 24 deltas were removed and replaced by
40 whorls (8% of all singular points). They explained such
modification arguing that some deltas near to ROI border were
excluded by their segmentation method, while some real loops
were not labeled in the ground-truth dataset.
V. C ONCLUSIONS
This work presented a novel method for singularities de-
tection based on a new analysis of ridge patterns. First, our
preprocessing step is composed of 3 main steps in order to
improve image quality and compute the orientation image in
a new manner. Then, the singularities are detected and post-
processing techniques are performed to improve singularities
location and avoid miss-detections. We achieved a singularities
detection rate of 93.42%in SPD2010 dataset.
We expect to improve our singularities detection method as
future research. First, we may apply noise filtering methods to
our preprocessed image. Thus, our method may become robust
to noise and make the singular point location more accurate.
Finally, we can improve our method by adding singularities
classification using Machine Learning algorithms.
TABLE III. COMPARISON BETWEEN OUR APPROACH AND THE
ALGORITHM PROPOSED BY [15]
Algorithm SP type Precision Recall F-score Accuracy
core 0.84 0.93 0.79 93%
OFθ
delta 0.75 0.93 0.69
core 0.99 0.98 0.97 100%
OF¯
θ
delta 0.96 0.91 0.89
core 0.90 0.97 0.93 94.18%
Proposed method
delta 0.91 0.88 0.90 88.14%
105
(a) (b) (c) (d) (e) (f) (g)
(h) (i) (j) (k) (l) (m) (n)
Fig. 4. Examples of our method robustness to rotated images. In each column, we can see the original images rotated (first line) and orientation images with
singularities detected (red circle) by our method. The red point in original images represents the reference point for the orientation image.
(a) (b)
Fig. 5. Mislabeled singular points in SPD2010 dataset: (a) Delta (red triangle)
labeled out of image. (b) No visible labeled core (red circle)
(a) (b)
Fig. 6. Samples of the errors reported by the proposed approach. The ground
truth detection and ours are represented by red and green, respectively. (a)
The divergence in singularities definition. (b) Correct (core) detection out of
threshold (10 pixels).
REFERENCES
[1] D. Maltoni, D. Maio, A. Jain, and S. Prabhakar, Handbook of fingerprint
recognition. Springer Science & Business Media, 2009.
[2] E. Liu, H. Zhao, F. Guo, J. Liang, and J. Tian, “Fingerprint segmentation
based on an adaboost classifier,Frontiers of Computer Science in
China, vol. 5, no. 2, pp. 148–157, 2011.
[3] S. Javed and A. Usman, “Computerized system for fingerprint classifi-
cation using singular points,” in Multitopic Conference (INMIC), 2011
IEEE 14th International. IEEE, 2011, pp. 96–101.
[4] H.-W. Jung and J.-H. Lee, “Fingerprint classification using the stochas-
tic approach of ridge direction information,” in Fuzzy Systems, 2009.
FUZZ-IEEE 2009. IEEE International Conference on. IEEE, 2009,
pp. 169–174.
[5] E. R. Henry, Classification and uses of finger prints. HM Stationery
Office, 1905.
[6] L. C. Jain, U. Halici, I. Hayashi, S. Lee, and S. Tsutsui, Intelligent
biometric techniques in fingerprint and face recognition. CRC press,
1999, vol. 10.
[7] Q. Zhang and H. Yan, “Fingerprint classification based on extraction
(a) (b)
(c) (d)
Fig. 7. Examples of lost singular points introduced by preprocessing step.
(a)-(b) line ending distortion caused by Canny edge detector. (c)-(d) joining
lines.
and analysis of singularities and pseudo ridges,” Pattern Recognition,
vol. 37, no. 11, pp. 2233–2243, 2004.
[8] M. Kawagoe and A. Tojo, “Fingerprint pattern classification,” Pattern
recognition, vol. 17, no. 3, pp. 295–303, 1984.
[9] A. Tariq, M. U. Akram, and S. A. Khan, “An automated system for
fingerprint classification using singular points for biometric security,
in Internet Technology and Secured Transactions (ICITST), 2011 Inter-
national Conference for. IEEE, 2011, pp. 170–175.
[10] C. Jin and H. Kim, “Pixel-level singular point detection from multi-
scale gaussian filtered orientation field,” Pattern Recognition, vol. 43,
no. 11, pp. 3879–3890, 2010.
[11] J. Zhou, J. Gu, and D. Zhang, “Singular points analysis in fingerprints
based on topological structure and orientation field,” Advances in
Biometrics, pp. 261–270, 2007.
[12] M. M. U. Khan and M. S. Sadi, “An efficient approach to extract
singular points for fingerprint recognition,” in Electrical & Computer
Engineering (ICECE), 2012 7th International Conference on. IEEE,
2012, pp. 13–16.
[13] P. Ramo, M. Tico, V. Onnia, and J. Saarinen, “Optimized singular point
detection algorithm for fingerprint images,” in Image Processing, 2001.
Proceedings. 2001 International Conference on, vol. 3. IEEE, 2001,
pp. 242–245.
[14] C.-Y. Huang, L.-m. Liu, and D. D. Hung, “Fingerprint analysis and
singular point detection,” Pattern Recognition Letters, vol. 28, no. 15,
pp. 1937–1945, 2007.
[15] P. Gupta and P. Gupta, “A robust singular point detection algorithm,”
106
Applied Soft Computing, vol. 29, pp. 411–423, 2015.
[16] Y. Yin, D. Weng, H. Li, and W. Ma, “A new robust method of
singular point detection from fingerprint,” in Information Science and
Engineering, 2008. ISISE’08. International Symposium on, vol. 2.
IEEE, 2008, pp. 330–334.
[17] L. Wang, N. Bhattacharjee, and B. Srinivasan, “Fingerprint reference
point detection based on local ridge orientation patterns of fingerprints,”
in Neural Networks (IJCNN), The 2012 International Joint Conference
on. IEEE, 2012, pp. 1–8.
[18] K. Nilsson and J. Bigun, “Localization of corresponding points in
fingerprints by complex filtering,” Pattern Recognition Letters, vol. 24,
no. 13, pp. 2135–2144, 2003.
[19] A. K. Jain, S. Prabhakar, L. Hong, and S. Pankanti, “Filterbank-based
fingerprint matching,” IEEE transactions on Image Processing, vol. 9,
no. 5, pp. 846–859, 2000.
[20] K. Nilsson and J. Bigun, “Complex filters applied to fingerprint images
detecting prominent symmetry points used for alignment,” in Interna-
tional Workshop on Biometric Authentication. Springer, 2002, pp.
39–47.
[21] C.-J. Lee, I.-H. Jeng, T.-N. Yang, C.-J. Chen, and P. Su, “Singular point
detection in fingerprint images by a bank of discrete fourier filters,” in
Intelligent Information Hiding and Multimedia Signal Processing, 2009.
IIH-MSP’09. Fifth International Conference on. IEEE, 2009, pp. 250–
253.
[22] H. Chen, L. Pang, J. Liang, E. Liu, and J. Tian, “Fingerprint singular
point detection based on multiple-scale orientation entropy,IEEE
Signal Processing Letters, vol. 18, no. 11, pp. 679–682, 2011.
[23] H. Yin, X. Jing, and S. Sun, “A novel algorithm for fingerprint
singular points detection based on vecrot orthogonal theory,” in Cloud
Computing and Intelligence Systems (CCIS), 2011 IEEE International
Conference on. IEEE, 2011, pp. 486–489.
[24] Z. Saquib, N. Salam, R. Nair, and N. Pandey, “An enhanced method
for singular points detection in fingerprint images,” in Computer and
Communication Technology (ICCCT), 2010 International Conference
on. IEEE, 2010, pp. 206–209.
[25] T. Ohtsuka, D. Watanabe, D. Tomizawa, Y. Hasegawa, and H. Aoki,
“Reliable detection of core and delta in fingerprints by using singu-
lar candidate method,” in Computer Vision and Pattern Recognition
Workshops, 2008. CVPRW’08. IEEE Computer Society Conference on.
IEEE, 2008, pp. 1–6.
[26] J. Qi and S. Liu, “A robust approach for singular point extraction based
on complex polynomial model,” in Proceedings of the IEEE Conference
on Computer Vision and Pattern Recognition Workshops, 2014, pp. 78–
83.
[27] B. G. Sherlock and D. M. Monro, “A model for interpreting fingerprint
topology,Pattern recognition, vol. 26, no. 7, pp. 1047–1055, 1993.
[28] R. Cappelli, A. Erol, D. Maio, and D. Maltoni, “Synthetic fingerprint-
image generation,” in Pattern Recognition, 2000. Proceedings. 15th
International Conference on, vol. 3. IEEE, 2000, pp. 471–474.
[29] L. Fan, S. Wang, H. Wang, and T. Guo, “Singular points detection
based on zero-pole model in fingerprint images,” IEEE Transactions on
Pattern Analysis and Machine Intelligence, vol. 30, no. 6, pp. 929–940,
2008.
[30] I. L. Andrezza, V. d. L. Erick, A. d. S. Marinho, A. E. de Oliveira,
J. R. Marques, A. Pedro Jr, and V. Leonardo, “Uma maneira simples de
obter regiões de interesse em imagens de impressões digitais,” Simpósio
Brasileiro em Segurança de Informação e de Sistemas Computacionais
- SBSeg, pp. 359–368, 2011.
[31] F. Turroni, R. Cappelli, and D. Maltoni, “Fingerprint enhancement
using contextual iterative filtering,” in Biometrics (ICB), 2012 5th IAPR
International Conference on. IEEE, 2012, pp. 152–157.
[32] M. F., H. P. Oliveira, and A. Campilho. SPD2010 - Fin-
gerprint Singular Points Detection Competition. Available at:
<http://paginas.fe.up.pt/ spd2010/>.
107
  • Conference Paper
    This paper focuses on a general framework for singular point extraction from vector field. We design a new index of singular point based on complex polynomial model. We test our method in the publicly available benchmark dataset of the singular point detection competition (SPD2010). Our algorithm gets the best results and produces large margins compared to the top five algorithms which took part in the public competition. We also compare our algorithm with the state-of-the-art singular point detection algorithm (called 'ZPM' method) with the benchmark. The performance of our method is much better than that of the state-of-the-art method.
  • Article
    This paper proposes an efficient algorithm to extract the singular points which can be used to classify the given fingerprint. It makes use of a novel algorithm which is a hybrid of orientation field, directional filtering and Poincare Index based algorithms to detect singular points, even when the fingerprint is of low quality or singular point is occluded. Locations of detected singular points are not much accurate and thus they are further refined. Also, some delta points which lie near to the border, may be missed out at the time of detection. Efforts are made to retrieve these missed points. The proposed algorithm also determines the direction of a singular point along with its type (either core or delta). It uses these detected singular points to classify accurately arch, tented arch, left loop, right loop, double loop and whorl type fingerprint patterns. It can handle efficiently the cases of missing delta points during fingerprint classification. The proposed algorithm has been tested on three publicly available databases. It reveals that the proposed algorithm exhibits better singular points detection and fingerprint classification performance in comparison to other well known algorithms.
  • Conference Paper
    Fingerprint analysis is typically based on the location and pattern of detected singular points in the images. These singular points (cores and deltas) not only represent the characteristics of local ridge patterns but also determine the topological structure (i.e., fingerprint type) and largely influence the orientation field. A core-delta relation is used as a global constraint for the final selection of singular points. This paper presents and discusses rules for implementing a structural fingerprint classifier. The proposed approach is divided in two main phases: Singular points extraction and recognition algorithm. The fingerprint classification is heavily depended on the singularities, i.e., core point and delta points. For the recognition of fingerprint a new technique is proposed based on the number of singular points and their types.
  • Conference Paper
    This paper introduces an approach to detect reference points for fingerprint images from their local ridge orientation maps. This approach classifies detected singular points into several core and delta point pairs. The validation procedure contains two stages in which the first stage is to remove false pairs of core and delta points and the second stage is to do some post processing of these unpaired singular points. Furthermore, this approach could be extended for reference point detection of arch and partial fingerprint images. The experiments of singular point detection part are conducted on FVC 2002 database 1 and 2. The result shows the average correct rate of singular point detection is 94.05% on the above databases.
  • Conference Paper
    The performance of Automatic Fingerprint Identification Systems (AFIS) relies on the quality of the input fingerprints, so the enhancement of noisy images is a critical step. We propose a new fingerprint enhancement algorithm that selectively applies contextual filtering starting from automatically-detected high-quality regions and then iteratively expands toward low-quality ones. The proposed algorithm does not require any prior information like local orientations or frequencies. Experimental results over both real (FVC2004 and FVC2006) and synthetic (generated by the SFinGe software) fingerprints demonstrate the effectiveness of the proposed method.
  • Article
    AFIS (automated fingerprint identification system) is very popular now days for biometric security. Fingerprint classification plays a key role in identifying fingerprints and also helps in fingerprint matching. This paper presents a new classification technique based on the detection of singular points (core and delta points). This fingerprint classification technique consists of four steps. In the first step, preprocessing (segmentation and normalization) of input fingerprint image is done. In the second step, fine orientation field of image is estimated. In the third step, singular points are located using modified Poincare index technique. In the fourth step, classification is done on the basis of singular points. The proposed technique was tested on FVC2004 database and the results show a significant improvement in reducing the misclassification errors.
  • Article
    Automated fingerprint identification system (AFIS) is very popular now days for biometric security because of the uniqueness of individual's fingerprint. The need for fingerprint classification arises due to very large fingerprint databases resulting in long response time which is unsuitable for real time applications. Hence in order to reduce number of comparisons fingerprint classification is necessary. It also plays a key role in identifying fingerprints. In this paper we have proposed a new classification technique based on the detection of singular points (core and delta points) consisting of four stages. In the first stage, preprocessing of input fingerprint image is done followed by fine orientation field estimation in second stage. In the third stage, singular points are located using modified Poincare index technique and hence in the fourth stage, classification is done on the basis of these singular points. The proposed technique was tested on NIST 4 database and the results show a significant improvement in classification of different types of fingerprints.