Enhancement of an Automatic Fingerprint Identification System Using a Genetic Algorithm and Genetic - PowerPoint PPT Presentation

1 / 40
About This Presentation
Title:

Enhancement of an Automatic Fingerprint Identification System Using a Genetic Algorithm and Genetic

Description:

Enhancement of an Automatic Fingerprint Identification System Using a Genetic ... (Farina et al., 1999; Fan et al., 2000; Tan and Bhanu, 2002a) ... – PowerPoint PPT presentation

Number of Views:133
Avg rating:3.0/5.0
Slides: 41
Provided by: pok67
Category:

less

Transcript and Presenter's Notes

Title: Enhancement of an Automatic Fingerprint Identification System Using a Genetic Algorithm and Genetic


1
Enhancement of an Automatic Fingerprint
Identification System Using a Genetic Algorithm
and Genetic Programming
Wannasak WetcharapornNachol ChaiyaratanaSanpacha
i Huvanandana
2
Aim
3
Overview
  • Introduction
  • Fingercode System
  • Feature Pruning Using a GA
  • Modification of the Matching Mechanism Using GP
  • Matching by a Combined 1-Norm and GP-Generated
    Operator
  • Conclusions

4
Introduction
5
Block Diagram for a Conventional Automatic
Fingerprint Identification System
6
Automatic Fingerprint Identification System
  • Classification (Henry, 1905)
  • Syntactic Approach (Moayer and Fu, 1986 Blue et
    al., 1994)
  • Structural Approach (Hrechak and Mchugh, 1990
    Karu and Jain, 1996 Hong and Jain, 1999 Cho et
    al., 2000)
  • Neural Network Approach (Blue et al., 1994 Mitra
    et al., 1994 Halici and Ongun, 1996 Jain et
    al., 1999)
  • Statistical Approach (Coetzee and Botha, 1999)
  • Recognition
  • Conventional Minutiae-Based Approach (Farina et
    al., 1999 Fan et al., 2000 Tan and Bhanu,
    2002a)
  • Evolutionary Minutiae-Based Approach (Tan and
    Bhanu, 2002b Tan and Bhanu, 2006)
  • Texture-Based Approach (Jain et al., 2000)

7
Fingercode System
8
Texture Feature Extraction
  • Find the reference point
  • Divide the image into bands and sectors
  • Apply 8 directional Gaborfilters
  • Calculate the standard deviation of filtered
    pixels in each sector
  • Use the standard deviation as a feature in the
    fingercode

9
Feature Pruning Using a GA
10
Individual (1024 bits Binary)
01011011011100100110111011011000010100001000110110
10110100101010001000101100000010000101011100110101
1010000001001110110010110011 11000111001100100111
01001001100001011000011111011110011011001000000110
00100000101100101011011000001100100110010100010100
10100010 1101101101100000101001001101101011110011
00001101110010111111000010010000101001010110101101
00101010000010000001010010110010110100 1011000010
01101101100111111010101010000101110001110011101000
01100101000101100001000010100011000000100101100011
100111000010100101 110001100100010011001001010010
01000011010010010110000010000100100101001000000100
000000000111101010101100100000000001001100101101
01000011001010001100110101100011000000111000101000
00100001001101001000000000010001011111001000010000
0000010110001010010010010110 01011110111010011001
00111110100000000111010001010010110010001100100000
11110100100010000001000000010110111010101110100100
00010000 1000101001110001011011010110010001000011
00100101000100011101011110011110101000000010000100
00010001010110000110011000110001000010
11
Fitness Evaluation
12
GA Parameter
13
AND OR and Majority Vote
14
How to Create Validation Sets
User Database
Intruder Database
15
GA Result
Percent ()
Generation
16
GA Result
Original (1024 Feat.)
Evolutionary (419 Feat.)
AND Func. (384 Feat.)
OR Func. (512 Feat.)
Majority Vote (492 Feat.)
Efficacy ()
17
GA Result
Original (1024 Feat.)
Evolutionary (419 Feat.)
AND Func. (384 Feat.)
OR Func. (512 Feat.)
Majority Vote (492 Feat.)
18
Feature Vector
Eight Filters
19
Modification of Matching Mechanism Using GP
20
Individual (Grow Method)
  • Maximum Tree Depth 10
  • Function Set , -
  • Terminal Set 0.25, 0.50, 0.75, 1.25, 1.50,
    1.75, 2.00, B1, B2 B492

Depth 0
Depth 1
Depth 2
Depth 3
21
GP Parameter
22
GP Result
Percent ()
Generation
23
GP Result
Original
GA (Majority Vote)
GP
Efficacy ()
24
GP Result
Original
GA (Majority Vote)
GP
25
Matching by a Combined 1-Norm and GP-Generated
Operator
26
GP Tree for Each Code-Group
27
GP Parameter
28
GP 1-Norm Result
Percent ()
Generation
29
GP 1-Norm Result
GA (Majority Vote)
GP
GP 1 Norm
Efficacy ()
30
GP 1-Norm Result
GA (Majority Vote)
GP
GP 1-Norm
31
Conclusions
N/A
N/A
32
Thank You
33
Appendix
34
Evolutionary Computation Algorithm
35
Stochastic Universal Sampling Selection
X1,
New population
X1,
X1,
X2,
X4,
X5,
X6,
X7
36
Uniform Crossover
?
?
?
?
?
?
?
?
?
?
0
1
0
1
1
1
1
0
0
1
Parent 1
Offspring 1
Offspring 2
0
1
0
1
1
1
0
1
1
1
Parent 2
37
Bit-Flip Mutation
0
1
0
1
1
1
1
0
0
1
0
1
0
1
38
Individual (Tree Structure)
Tree Structure
( (- Bit2 Bit23) ( 1.50 Bit78))
LISP S-expression
39
Crossover
40
Mutation

-

-
1.75
B43
B24
B3
B108
Write a Comment
User Comments (0)
About PowerShow.com