This document outlines a research report on using image morphing and genetic algorithms to improve credit card security. It discusses using genetic algorithm (GA) and differential evolution (DE) to morph between two face images with 25 selected feature points. An experiment is proposed to compare interactive GA (iGA) and interactive DE (iDE), where a human evaluates the algorithm outputs, by running each algorithm 5 times on two face images with 20 individuals over 10 generations. The goal is to generate a total of 200 morphed images with each run to improve credit card security through image morphing.
3. Background & Purpose
Many people are using the credit card.
However, the security of the credit card is not
perfected.
The criminal humans steal the information of
the credit card and use them illegally.
We try to improve the security of the credit
card by using the image morphing
technology.
3
4. Image Morphing
2 images combining, and making a these
middle images.
To select 25 feature points of each image.
To hit the points to the feature place.
Feature points are eyes, eyebrow, nose,
mouth and etc.
To do morphing based on them.
4
5. Genetic Algorithm(GA), Differential
Evolution (DE)
GA
Genetic Algorithm is one of the Evolutionary Algorithms.
GA is the algorithm based on a mechanism
of biological evolution such as reproduction, recombinati
on of genes , natural selection ,and the survival of the fitt
est.
DE
Differential Evolution (DE) is a method of Evolution
Strategy.
Normal Evolution Strategy use gauss mutation, however
DE use sum of base vector and weighted difference
vector in substitution for gauss mutation.
5
6. Flow Chart
GA DE
initialization initialization
Yes Yes
evolution end evolution end
No
No
selection selection
crossover crossover
mutation
6
7. interactive GA (iGA), interactive DE
(iDE)
iGA and iDE are different from GA and DE in
an evaluation method.
These evaluation method are human’s
operation.
The human subjective evaluation is not
possible to a machine.
7
8. Experiment
To compare iDE and iGA.
We prepare two face images that has 25
feature points.
We set the number of individuals are 20 and
the number of generation is 10.
Each total morphed images are 200.
We run this 5 sets.
8