Face Recognition Technique using ICA and LBPH

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 04 Issue: 08 | Aug -2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 5.181 | ISO 9001:2008 Certified Journal | Page 104
FACE RECOGNITION TECHNIQUE USING ICA AND LBPH
Khusbu Rani1, Sukhbir kamboj2
1RESEARCH SCHOOLAR
2ASSISTANT PROFESSOR
Dept. of Computer Science and Engineering Global Research institute of Management and Technology
Kurukshetra Haryana, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract Most of the image processing techniques such as
edge detection, segmentation, object tracking, pattern
recognition etc. do not perform well in the occurrenceof noise.
Thus, image restoration as a preprocessing step is performed
before applying the image to any of the beyond mentioned
techniques. presents a process offacerecognitionsystemusing
principle analysis with Back propagation neural network
where features of face image has been combined by face
detection and edge detection technique. In this system, the
performance has been analyzedbasedontheproposedfeature
fusion technique. At first, the fussed featurehasbeenextracted
and the dimension of the feature vector has been reduced
using Indepdent Component Analysis method and LBPH
Key Words: Face Recognition, LBP,ICA,PCA
1.INTRODUCTION
Face recognition concept of featureextractionanddetection,
is a small capacity for human beings. Humanhavedeveloped
this skill to correctly and instantaneously recognize things
around us after millions of years of evolution. The
necessitate for machine intrusion in face recognition to
create the whole process gives rise to Automated Face
Recognition (AFR) that simulates the Human Vision System
(HVS).[1] The past few decades have seen AFR receive
immense attention due to its myriad applications in fields of
security and surveillance.Implementationsincomputers are
much more complex though not impossible. Image
processing (in this specific case, leading to face recognition)
by computers usually takes place in this order:
1. Reduction of high-dimensional real-world data set
to lesser dimensions in order to facilitate faster
processing speeds on relatively low-end machines.
2. Implementing a machine learning algorithm to
train with a test data-set.
Automated Face recognition is a particularly attractive
approach. Although research in automatic face recognition
has been conducted since the 1960s, this problem [2] is still
largely unsolved.Recentyearshaveseensignificantprogress
in this area owing to advances in face modeling and analysis
techniques. It has a wide number of applications including
security, law enforcement, person verification, Internet
communication, Pattern Recognition and computer
entertainment. The first large scale application of face
recognition was carried out in Florida. Face recognition has
two main steps: feature extraction and classification. Image
processing technique has been applied to evaluate feature
from image [3] database and there are some classification
techniques that are applied to recognize the unknown face
image. The overview of current system is demonstrated in
figure 1.1.
Fig. 1.1: System’s overview.
To develop a helpful and appropriate face recognition
system numerous factors need to be taken in hand.
1. The overall speed of the system from detection to
recognition should be acceptable.
2. The accuracy should be high
II. Local Binary Pattern
There exist several methods for extracting the most useful
features from (preprocessed) face images to perform face
recognition. One of these feature extraction methods is the
Local Binary Pattern (LBP) method. This relative new
approach was introduced in 1996 by Ojala et al. [5]. With
LBP it is possible to describe the texture and shape of a
digital image. This is done by dividing an image into several
small regions from which the features are extracted (figure
1.1).

Fig :1.1 A preprocessed image divided into 64 regions
These features consist of binary patterns that describe the
surroundings of pixels in the regions. The obtained features
from the regions are concatenated into a single feature
histogram, which forms a representation of the image.
Images can then be compared by measuring the similarity
(distance) between their histograms. According to several
studies [2, 3, 4] face recognition using the LBP method
provides very good results, both in terms of speed and
discrimination performance. Because of the way the texture
and shape of images is described, the method seems to be
quite robust against face images with different facial
expressions, different lightening conditions, image rotation
and aging of persons.
We explained how the LBP-methodcanbeappliedonimages
(of faces) to extract features which can be used to get a
measure for the similarity between these images. The main
idea is that for every pixel of an image the LBP-code is
calculated. The occurrence of each possible pattern in the
image is kept up. The histogram of these patterns,alsocalled
labels, forms a feature vector, and is thus a representation
for the texture of the image. These histograms can then be
used to measure the similarity between the images, by
calculating the distance between the histograms.
III. Independent Component Analysis (ICA)
Independent component analysis (ICA) is an overview of
PCA. There are a number of algorithms for performing ICA
[9],[10] This method is also called blind source separation
(BSS), the major objective of this method is to minimizes the
second order and higher order dependencies in the input
and produces a collection of statistically source vectors. In
other word this method is mainly used in high order
dependencies. We applied ICA technique on the set of two
images architectures
Fig. 3.1: Image synthesis model for Architecture I.
To find a set of IC images, the images in X are considered to
be a linear combination of statistically independent basis
images, S, where A is an unknown mixing matrix. The basis
images were estimated as the learned ICA output U.
Architecture II, based on[11]and[12]..Each image in the
dataset was considered to be a linear combination of
underlying basis images in the matrix A. The basis images
were each associated with a set of independent “causes,”
given by a vector of coefficients in S. The basis images were
estimated by A = W , where W is the learned ICA weight
matrix. The Architecture II is below given:
Fig. 3.2: Image synthesis model for Architecture II.
Architecture I treated the images as arbitrary variables and
the pixels as outcomes, whereas Architecture II treated the
pixels as arbitrary variables and the images as outcomes.
Both ICA Architectures gives bettertorepresentations based
on PCA for recognizing faces across changes in phrase. .
Classifiers that joint both the ICA representations gave the
greatest performance.
IV Proposed Face Recognition Technique
Face recognition is the current area of research for its wide
range of practical applications.Thereare numerous numbers
of face recognition methods. They are further categorized
into two categories: appearance-based and feature-based
approaches. Feature based techniques extract face feature
indicators based ongeometrical relationships&propertiesof
each face characteristics like eyes, nose, mouth, and chin.
There recognition accuracy depends upon face feature
extraction techniques. This is not reliable in practical
applications. Appearance-based processes uses global face
features depend on intensity vector representation. These
approaches widely utilize by many researchers. Face-
recognition performance significantly decreases if there are
variations in the pose, illumination, and size of the input
image.
Many techniques are proposed to tackle this problem.
Recognition system using PCA and BPNN provides high
recognition rate and fast execution time. PCA is used for
feature extraction and space dimension reduction. BPNN is

used for image classifications. Recognition rate and
execution time are two main parameters, which are
measured during implementation of LBPH+ICA. The
working model is made up of three steps as shown in figure
Algorithm LBPH and ICA
To implement the face recognition in this research work, we
proposed the Local Binary patterns methodology and ICA .
Local Binary Pattern works on local features that uses LBPH
and ICA operator which summarizes the local special
structure of a face image. LBPH is defined as an orders set of
binary comparisons of pixels intensities between the center
pixels and its eight surrounding pixels. Local Binary Pattern
do this comparison by applying the following formula:
LBPH (Xc,Xy) = ∑7
n=0 s(in-ic) 2n
Where ic corresponds to the value of the center pixel (𝑥,𝑦 𝑐 ),
in to the value of eight surrounding pixels. It is used to
determine the local features in the face and also works by
using basic LBP operator.Featureextractedmatrixoriginally
of size 3 x 3, the values are compared by the value of the
center pixel, then binary pattern code is produced and also
LBP code is obtained by converting the binary code into
decimal one.
Input: Training Image set. Output: Feature extracted from
face image and compared with center pixel and recognition
with unknown face image.
1. Initialize temp = 0
2. FOR each image I in the training image set
3. Initialize the pattern histogram, H = 0
4. FOR each center pixel tcȯ I
5. Compute the pattern label of tc,LBP(1)
6. Increase the corresponding bin by 1.
7. END FOR
8. Find the highest LBPH and ICA feature for each face image
and combined into single vector.
9. Compare with test face image.
Algorithm Step ICA :
1. Initialize temp = 0
2. for i=1:R
3. for j=1:C
4. value=I(i,j);
5. freq(value+1)=freq(value+1)+1;
6. probf(value+1)=freq(value+1)/numofpixels;
end
end
7. for i=1:size(probf)
8. sum=sum+freq(i);
cum(i)=sum;
probc(i)=cum(i)/numofpixels;
output(i)=round(probc(i)*no_bins);
end
YES
NO
Fig4.1 :Flowchart of the LBP Process
Training and Test Face
Recognized image is obtained by calculating the Euclidean
distance between the weight vector of test face and kth
training image. The image whichhaslessdistanceisdetected
as output image. It must closely resemble the input
face.There are 400 images in ORL face database. We have to
select 20 training images & one test image. 20 training
images are selected randomly form 3 subjects out of 40
subjects. Each subject contains 10 images of same person in
different expression, pose and illumination
Capture face image
The face image is divided into
several block
Block LBPH are connected into
Histogram Calculated for each block
Facial recognition are represented
using LBPH
Face
Image

Training set
Figure 5.1: Training set of 20 images from ORL face
database
The common feature of all images is calculated by mean
image. When we subtract the mean image from all the
training images, we get normalized images. Normalize face
image represents all the unique featuresinrespectiveimage.
Figure 5.2: Recognise face
We compare our recognition method with the original LBP
method, and we use original LBP method by split face image
to 4×4 grids. Then we use our method to test face
recognition by 53 landmarks of face and we can set the grid
of a landmark point as 9×9, 7×7, 5×5 and 3×3. Thespeedand
true positive rate are compared for different
So far we have concerned ourselves by application of ICA in
terms of blind source separation. In this sectionutilization in
image processing will be explained [1, 5]. The main concept
of ICA applied to images insists on the idea that each image
(subimage) may be perceived as linear superposition of
features ai(x, y) weighted by coefficients si. In case of ICA,
features are represented by columns of mixing matrix A and
si are elements of appropriate sources. In addition ICA
features are localized and oriented and sensitivetolinesand
edges of varying thickness of images (see Figures 5.2 ).
Furthermore the sparsity of ICA coefficients should be
pointed out. It is expected that suitable soft-thresholding on
the ICA coefficients leads to efficient reducing of Gaussian
noise.
Figure 5.3: Histogram recognize image
Recognized image is obtained by calculating the weight
vectors of all training images. Weight vectors denote the
contribution of each Eigenfacetoall trainingimages.Highest
weight vector means highest contributionofEigenface.With
the help of weight vectors, we calculate Euclidean distance
between the weight vector of test face and kth training
image. The image which has lessEuclidean distance is
detected as output image. It must closely resemble the input
face.
V .Conclusion
The proposed algorithm gives better performance in
comparison with LBPH, ICAandotherexistingnoiseremoval
algorithms in terms of MSE However the time required
executing this algorithm is bit more than the existing
algorithms .The performance of the algorithm is tested
against Face Database images at low, medium and high
densities, showing the effectiveness how impulse noise is
removed through the colour images. It yields better results
than existing methods even at very high noise densities of
80% and 90%. Both visual and quantitative results are also
demonstrated
REFERENCES
[1] ZahidMahmood, Tauseef Ali, Samee U. Khan, “Effects of
pose and image resolution on automatic face recognition”,
IEEE transaction, vol. 5, pp. 111-119, 2016
[2] Ms. Madhavi R. Bichwe, Ms. RanjanaShende, “Face
Recognition in a Video by pose variations”, IEEE
International Conference on Computer, Communicationand
Control, pp: 1-5, 2015
[3] A.H. Boualleg, Ch. Bencheriet and H. Tebbikh, “Automatic
Face recognition using neural network-PCA”, IEEE
International Conference on Information and
Communication Technology, vol. 1, pp. 1920-1925, 2006
[4] Lih-Heng Chan, Sh-HussainSalleh, Chee-Ming Ting, A.K
Ari, “PCA And LDA-Based Face Verification using Back-

Propagation Neural Network”, IEEE International
Conference on Information Science, Signal Processing and
their Applications, pp. 728-732, 2010
[5] Rajath Kumar M. P., KeerthiSravan R., K. M. Aishwarya,
“Artificial Neural Networks for Face Recognition using PCA
and BPNN ”, IEEE International Conference on Information
and Communication Technology, pp. 1-6, 2015
[6] TahiaFahrinKarim, MollaShahadatHossainLipu, Md.
LushanurRahman, Faria Sultana, “Face Recognition Using
PCA-Based Method”, IEEE International Conference on
Advance Management Science, volume 3, pp. 158-162, 2010
[7] Athmajan, Rajasinghe, Senerath, Ekanayake,
Wijayakulasooriya, “Improved PCA Based Face Recognition
Using Similarity Measurement Fusion”, IEEE International
Conference on Industrial and Information Systems, pp. 360-
365, 2015.
[8] KolhandaiYesu, HimadriJyotiChakravorty,
PrantikBhuyan, RifatHussain, Kaustubh Bhattacharyya,
“Hybrid Features Based Face Recognition Method using
Artificial Neural Network”, IEEE National conference on
Signal Processing, pp. 40-45, 2012
[9] Mrs. Abhjeet Sekhon, Dr. Pankaj Agarwal, “Face
Recognition using Back Propagation Neural Network
Technique”, IEEE International Conference on Advances in
Computer Engineering and Applications, pp. 226-230, 2015
[10] Hemant Singh Mittal, Harpreet Kaur, “Face Recognition
using PCA & Neural Network”,IEEEInternational Conference
on Information Science, Signal Processing and their
Applications, volume 3, pp. 158-162, 2013. Recognition
Letters, (2003).
[11] WeichengShen and Tieniu Tan,“AutomatedBiomertics-
based personal Identification”, Arnold and Mabel Beckman
Center of the National Academics of Sciences and
Engineering in Irvine, CA, August 1998.
[12] Michal Chora, “Emerging MethodsofBiometricsHuman
Identification” Image Processing Group, Institute of
Telecommunications, 695-882, IEEE, (2007).

Face Recognition Technique using ICA and LBPH

More Related Content

What's hot (20)

Similar to Face Recognition Technique using ICA and LBPH (20)

More from IRJET Journal (20)

Recently uploaded (20)

Face Recognition Technique using ICA and LBPH