50120140504002

International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-6367(Print),

ISSN 0976 - 6375(Online), Volume 5, Issue 4, April (2014), pp. 11-23 © IAEME

11

A NEW FACE RECOGNITION SCHEME FOR FACES WITH EXPRESSIONS,

GLASSES AND ROTATION

Walaa M Abdel-Hafiez1, Mohamed Heshmat

2, Moheb Girgis

3, Seham Elaw

4

1, 2, 4

Faculty of Science, Mathematical and Computer Science Department,

Sohag University, 82524, Sohag, Egypt

3Faculty of Science, Department of Computer Science,

Minia University, El-Minia, Egypt

ABSTRACT

Face recognition is considered as one of the hottest research areas in computer vision field.

The purpose of the proposed research work is to develop an algorithm that can recognize a person by

comparing the characteristics of his/her face, which may have expressions, glasses and/or rotation, to

those of known faces in a database. This work provides a simple and efficient technique to recognize

human faces. The new method is based on variance estimation of the three components of the color

faces images and facial features extraction of the most facial features. The features under

consideration are eyes, nose and mouth. The technique used to extract facial features was developed

based on feature location with respect to face dimensions. The proposed algorithm has been tested on

various face images and its performance was found to be good in most cases. Experimental results

show that our method of human face recognition achieves very encouraging results with good

accuracy, great speed and simple computations.

Keywords: Face Recognition, Facial Features Extraction, Color Spaces, Variance Estimation.

I. INTRODUCTION

Face recognition has been used in various applications where personal identification is

required like, visual attendance systems where student identification and recognition are achieved

through face recognition. Face recognition [1-4] has been used also in gaming applications, security

systems, credit-card verification, criminal identifications, and teleconference, and short face

recognition applications are widely used in many corporate and educational institutions.

INTERNATIONAL JOURNAL OF COMPUTER ENGINEERING &

TECHNOLOGY (IJCET)

ISSN 0976 – 6367(Print)

ISSN 0976 – 6375(Online)

Volume 5, Issue 4, April (2014), pp. 11-23

© IAEME: www.iaeme.com/ijcet.asp

Journal Impact Factor (2014): 8.5328 (Calculated by GISI)

www.jifactor.com

IJCET

© I A E M E



12

Human faces are complex objects with features that can vary over time. However, humans

have a natural ability to recognize faces and identify persons with just a glance. Our natural

recognition ability also extends beyond face recognition, where we are equally able to quickly

recognizing patterns, sounds or smells. Unfortunately, this ability does not exist in machines, thus the

need to simulate recognition artificially in our attempts to create intelligent autonomous machines.

Facial feature recognition is an example of popular applications for artificial intelligence systems.

Face recognition by machines has various important applications in real life, such as, electronic and

physical access control, national defense and international security. Simulating our face recognition

natural ability in machines is difficult but not impossible. Throughout our lifetime, many faces are

seen and stored naturally in our memories which forming a kind of database. Machine recognition of

faces requires also a database, which consists of facial images that may include different face images

of the same person. The development of intelligent face recognition systems requires providing

sufficient information and meaningful data during machine learning of a face.

Face Recognition can be defined as the visual perception of familiar faces or the biometric

identification by scanning a person's face and matching it against a database of known faces. In both

definitions, the faces can be identified as familiar or known faces.

One of the main challenging problems in building an automated system that perform face

recognition and verification tasks is face detection and facial feature extraction. Though people are

good at face identification, recognizing the human faces automatically by computer is a very difficult

task. Face recognition is influenced by many complications, such as the differences of facial

expression, the light directions of imaging, and the variety of posture, size and angle. Even for the

same person, the images taken in a different surrounding condition may be unlike. The problem is so

complicated that the achievement in the field of automatic face recognition by computer is not as

satisfied as the finger prints [5].

The objective of facial feature localization is to detect the presence and location of features

after the locations of faces are extracted by using any face detection method. The challenges

associated with face and facial feature detection methods can be attributed to the following

factors [6]:

• Intensity: There are three types of intensity: color, gray, and binary.

• Pose: Face images vary due to the relative camera-face pose (frontal, 45º, profile), and some

facial features such as an eye may become partially or wholly occluded.

• Structural components: Facial features such as beards, mustaches, and glasses may or may not

be present.

• Image rotation: Face images directly vary for different rotations.

• Poor quality: Image intensity in poor-quality images, for instance, blurry images, distorted

images, and images with noise, becomes unusual.

• Facial expression: The appearance of faces depends on a personal facial expression.

• Unnatural intensity: Cartoon faces and rendered faces from 3D model have unnatural intensity.

• Occlusion: Faces may be partially occluded by other objects such as hand, scarf, etc.

• Illumination: Face images vary due to the position of light source [6].

Phimoltares et al. [6] presented algorithms for all types of face images in the presence of

several image conditions. There are two main stages in their method. In the first stage, the faces are

detected from an original image by using Canny edge detection and their proposed average face

templates. Second, a proposed neural visual model (NVM) is used to recognize all possibilities of

facial feature positions. Input parameters are obtained from the positions of facial features and the

face characteristics that are low sensitive to intensity change. Finally, to improve the results, image

dilation is applied for removing some irrelevant regions.



13

Nikolaidis and Pitas [7] proposed a combined approach for facial feature extraction and

determination of gaze at direction that employs some improved variations of the adaptive hough

transforms for curve detection, minima analysis of feature candidates, template matching for inner

facial feature localization, active contour models for inner face contour detection and projective

geometry properties for accurate pose determination.

Koo and Song [8] suggested defining 20 facial features. Their method detects the facial

candidate regions by haar classifier, and detects eye candidate region and extracts eye features by

dilate operation, then detect lip candidate region using the features. The relative color difference of

a* in the L*a*b* color space was used to extract lip feature and to detect nose candidate region and

detected 20 features from 2D image by analyzing end of nose.

Yen and Nithianandan [9] presented an automatic facial feature extraction method based on

the edge density distribution of the image. In the preprocessing stage, a face is approximated to an

ellipse, and a genetic algorithm is applied to search for the best ellipse region match. In the feature

extraction stage, a genetic algorithm is applied to extract the facial features, such as the eyes, nose

and mouth, in the predefined sub regions.

Gu et al. [5] proposed a method to extract the feature points from faces automatically. It

provided a feasible way to locate the positions of the two eyeballs, near and far corners of eyes,

midpoint of nostrils and mouth corners from face image.

Srivastava [10] proposed an efficient algorithm for facial expression recognition system,

which performs facial expression analysis in a near real time from a live web cam feed. The system is

composed of two different entities: trainer and evaluator. Each frame of video feed is passed through

a series of steps, including Haar classifiers, skin detection, feature extraction, feature point tracking,

creating a learned support vector machine model to classify emotions to achieve a tradeoff between

accuracy and result rate.

Radha and Nallammal [11] described a comparative analysis of face recognition methods:

principle component analysis (PCA), linear discriminant analysis (LDA) and independent component

analysis (ICA) based on curvelet transform. The algorithms are tested on ORL Database.

Kumar et al. [12] presented an automated system for human face recognition in a real time

background world for a large homemade dataset of persons' faces. To detect real time human face

AdaBoost with Haar cascade is used and a simple fast PCA and LDA are used to recognize the faces

detected. The matched face is then used to mark attendance in the laboratory, in their case.

El-Bashir [13] introduced a method for face recognition. After a preprocessing and

normalization stage to the image, PCA is applied to recognize a specified face. If the face is not

recognized correctly, then more features are extracted: face color and moment invariant. The face is

recognized again using decision tree.

Javed [14] proposed a computer system that can recognize a person by comparing the

characteristics of face to those of known individuals. He focused on frontal two-dimensional images

that have been taken in a controlled environment, i.e. the illumination and the background were

constant, and used the PCA technique. The system gives good results especially with angled face

views.

Pattanasethanon and Savithi [15] presented a novel technique for facial recognition through

the implementation of successes mean quantization transform and spare network of winnow with the

assistance of eigenface computation. After having limited the frame of the input image or images

from web-cam, the image has cropped into an oval or ellipse shape. Then the image is transformed

into grey scale color and is normalized in order to reduce color complexities. They also focused on

the special characteristics of human facial aspects such as nostril areas and oral areas, compared the

images obtained by web-cam with images in database, and have good accuracy with low time.



14

Wu et al. [16] presented a system that can automatically remove eyeglasses from an input

face image. The system consists of three modules: eyeglasses recognition, localization and removal.

Given a face image, first, an eyeglasses classifier is used to determine if a pair of eyeglasses is

present. Then, a Morkov chain Monte Carlo method is applied to locate the glasses by searching for

the global optimum of the posteriori. Finally, a novel example based approach has been developed to

synthesize an image with eyeglasses removed from the detected and localized face image. The

experiments demonstrated that their approach produces good quality of face images with eyeglasses

removed.

Chen and Gao [17] presented a local attributed string matching (LAStrM) approach to

recognize face profiles in the presence of interferences. The conventional profile recognition

algorithms heavily depend on the accuracy of the facial area cropping. However, in realistic scenarios

the facial area may be difficult to localize due to interferences (e.g., glasses, hairstyles). The proposed

approach is able to efficiently find the most discriminative local parts between face profiles

addressing the recognition problem with interferences. Experimental results have shown that the

proposed matching scheme is robust to interferences compared against several primary approaches

using two profile image databases (Bern and FERET).

This paper presents a new face recognition method for faces with expressions, glasses and/or

rotation. The proposed method uses variance estimation of RGB components to compare the

extracted faces and the faces in the database used in comparison. In addition, Euclidean distance of

facial features of the extracted faces from test image and faces extracted from the database after a

variance test is used.

The rest of this paper is organized as follows. Section II describes the methodology of the

proposed method with its stages: variance estimation, feature extraction, method representation and

the proposed algorithm. Section III presents the results and method analysis. Section IV draws the

conclusion of this work and possible points for future work.

II. METHODOLOGY

The face and facial feature detection algorithms are applied to detect generic faces from

several face images. Most automatic face recognition approaches are based on frontal images. Facial

profiles, on the other hand, provide complementary information of the face that is not present in

frontal faces. Fusion of frontal and profile views makes the overall personal identification technique

foolproof and efficient.

The proposed face recognition method is based on the average variance estimation of the

three components of RGB faces images, and the extraction of the most facial features. The features

under consideration are eyes, nose and mouth. The technique used to extract facial features is based

on feature location with respect to the dimensions of the face image.

Given a face image, which obtained from a camera or preprocessed previously, our goal is to

identify this face image using a database of known humans' faces. Therefore, our algorithm is divided

into three main steps. First: variance estimation of faces images. Second: facial feature extraction,

an effective method to extract facial features like eyes, nose and mouth depending on their locations

with respect to the face region is used, which we have developed before in [18]. Third: similar face

identification or image searching; the goal of this step is to scan the database of known faces to find

the most similar faces to the test face.



15

1. Variance Estimation

Variance calculation is a very light calculation and considered as an important constraint to

prove similarity between two images. Let x be a vector of dimension n, the variance of x can be

calculated as follows:

n

xi

xn

i

∑ −

= =1

)(

var

2

, (1)

where x is the mean value of x .

However, it is not necessary that the two images which have the same variance to have the

same contents. Different images may have the same value of variance because variance estimation is

totally depending on the values of image pixels and their mean value. So the variance is used at first

to filter the database of faces and extract faces that have the same or close value to variance of the

input face image, then another test is required to detect the most similar faces to this test face [18].

When working with RGB color images, there are three values for each pixel in the image,

representing the red, green, and blue components. To compute the variance of RGB image, the

variance for each color is calculated separately. So there are three values for variance, one for the red

values, another for the green values and third for the blue values [18], which are calculated as

follows:

2 2 2

1 1 1

( ) ( ) ( )

, , ,

n n n

r r g g b b

i i ired green blue

x x x x x x

v v vn n n

= = =

− − −

= = =

∑ ∑ ∑ (2)

To simplify the comparison, the average of the three values is computed as follows:

( ),

3

red green bluev v vv

+ += (3)

2. Facial Features Extraction

In this part of work, the aim is to compare two color faces to decide whether they both belong

to the same person or not and detect the similarity between them using Euclidean distance. RGB

(Red, Green and Blue) color space, fig. 1, which is used here, is an additive color system based on

tri-chromatic theory. It is often found in systems that use a CRT to display images. The RGB color

system is very common, and is being used in virtually every computer system as well as television,

video etc [19], [20].

Red Green Blue

Figure 1: RGB color model



16

In RGB color model, any source color (F) can be matched by a linear combination of three

color primaries, i.e. Red, Green and Blue, provided that none of those three can be matched by a

combination of the other two, see fig. 1.

Here, F can be represented as:

,F r R gG b B= + + (4)

where r, g and b are scalars indicating how much of each of the three primaries (R, G and B) are

contained in F. The normalized form of F can be as follows:

' ' ' ,F R R G G B B= + + (5)

where

' / ( ) ,

' / ( ) , (6)

' / ( ) ,

R r r g b

G g r g b

B b r g b

= + +

= + +

= + +

To extract facial features, we used our method proposed in [18], which is based on feature

location with respect to the whole face region. By detecting the candidate regions of left eye, right

eye, nose and mouth, by training, then applying the obtained dimensions of each region on several

other faces with the same size, the results were very good, as shown in fig. 2.

Given a face image of 200 pixels height and 200 pixel width, after training with a lot of

images, we found that the candidate region of eyes is located between rows 60 and 95, columns 25

and 80 for right eye and columns 115 and 170 for left eye. The candidate region for the nose is

located between rows 110 and 145 and columns 75 and 125 and the candidate region for the mouth is

located between rows 145 and 185 and columns 60 and 135. When applying the dimensions obtained

by training on many face images, we found that they there were suitable for any face image with the

same width and height even it has expression, as shown in fig. 2.

Figure 2: Examples of Feature extraction

(6)



17

This feature extraction technique can be generalized and the candidate region for each feature,

which is based on height and width of the face image to match any face image size can be as follows:

• Right eye: Rows from (height/3.3) to (height /2.1)

Columns from (width/8) to (width/2.5)

• Left eye: Rows from (height/3.3) to (height /2.1)

Columns from (width/1.7) to (width/1.17)

• Nose: Rows from (height/1.8) to (height /1.38)


• Mouth: Rows from (height/1.38) to (height /1.08)


3. Method Representation

The proposed algorithm consists of three parts. Firstly, variance estimation is applied to

extract database images, which have a close variance value to the test image. Secondly, the features

extraction method is used to extract facial features from the face images. Finally, Euclidean distance

of facial features is computed by the following equation:

[ ] [ ]( )

[ ] [ ]( )

[ ] [ ]( )

d abs test feature R matched feature R

abs test feature G matched feature G

abs test feature B matched feature B

= − +

− +

−

(7)

By applying eq. (7) to find the distance between the right eye region of the test image and the

right eye region of each image, which has variance value close to the variance value of the test image

(returned from variance test), then, by applying eq. (7) to left eye, nose and mouth regions and find

summation of these four distance values, it can be decided which of the images that have close

variance value is the most similar to test image. The steps of the proposed algorithm are shown in

fig. 3.

Step 1: Read input image.

Step 2: Read the database of images and calculate the variance of each image by using eq. (2) , (3), and

put variance values in an array,

Step 3: Calculate variance of the test image using eq. (2), (3).

Step 4: Compare variance value of test image and each image in database and keep locations of the most

similar images to test image, which satisfy the condition ( )600 variance difference 600− ≤ ≤ , in an array.

Step 5: For i=1 to number of similar images which extracted from step 4.

a) Extract facial features from each image according to location (right eye – left eye – nose –

mouth).

b) Calculate the Euclidean distance between the 3-arrays containing the RGB color values of each

feature using eq. (7) plus the Euclidean distance between the 3-arrays containing the RGB color

values of the whole test image and each similar image from step 4.

Step 6: Detect minimum distance (d ) and location of the image that has the minimum distance from step

5.

Step 7: Display the best-matched image from the database.

Figure 3: The proposed algorithm's steps



18

III. RESULTS AND DISCUSSION

The experiments were performed on a computer with 2.20 GHz speed and 4 Gbyte RAM

using several color images containing faces and a database of 10 different images with different

sizes, as shown in fig. 4. The test images include different images for the same person with different

conditions. Some images have expressions, glasses and some rotation, as shown in fig. 5. The

images in the database have been chosen carefully such that they are standard and have no

expressions if possible.

The proposed algorithm gives good results in recognizing all the test images, which belong to

the same person in the database, with different expressions, glasses and rotation. Even if the gaze

direction is different, the proposed algorithm succeeds in returning the correct location of the right

image in the used database.

Figure 4: The used database

Figure 5: The used database and some of test images

Table 1 shows some of the results obtained using 150 test RGB images of 10 different

persons and a database of 10 standard RGB images of those persons, which is shown in fig. 4. The

first column of the table shows the test face. The next columns show the results that were obtained by

applying the classical method, variance estimation formula, feature extraction method, the proposed

method and the time in each method. The classical method is the general method in comparing two

images, by comparing pixel by pixel and computing the summation of the difference of all pixels.

The classical method performs on the whole image without partitioning. The variance estimation is

5 4 3 2 1

10 9 8 7 6



19

applied by using eq. (2) and (3). We have displayed the results of the variance separately to show

how the variance computation is efficient and important in comparing similarity between images.

When we used the variance estimation as a first test in the proposed method, we noticed that it gives

the correct image location from the database if the test image and the matched image in the database

have similar conditions of illumination and background. Also, we have applied the feature extraction

method separately to study how it is efficient in face recognition. Facial features are extracted and

Euclidean distance is computed for each feature then the summation of the difference is obtained. By

comparing the difference between the test image and all the images in the database, the matched

image is detected as the image has minimum difference. It is noticed from the table that the classical

method and variance estimation method have less time than the two other methods. The execution of

the proposed method proceeds as follows: the first test (variance test) with variance difference range

equals [-600,600] is applied first to detect the images that have close variance values to the test

image. (It should be noted that the variance difference range is arbitrary and can be changed). The

algorithm returns the locations of faces whose variance value close to the variance of the test face. In

order to know which one of them is the same or the closest to the test face, the facial features of the

test face and the facial features of the obtained face images are extracted then the Euclidean distance

of their RGB components is calculated by eq. (7). The face image with the minimum distance (d) is

considered as the best-matched image and its location is returned.

The search efficiency is evaluated by how many times the distance (d) computations are

performed on average compared to the size of the database. In the proposed method the total number

of distance calculations is small, because it uses the variance test to find out the face images that have

a close variance value to the input face image, then the distance computation is performed only on

those images where their number is always small compared to the database size. But execution time

of the proposed method is high compared to the other methods because the proposed method works

in two stages or two tests variance estimation and facial feature extraction where each of these stages

take some time. The execution time depends on the database size. The execution time of the classical

method is 0.43 seconds on average, the execution time of the variance estimation method is 0.1

seconds on average, the execution time of facial feature extraction method is 0.22 seconds on average

and the execution time of the proposed method is 1.06 seconds on average, as shown in Fig. 6.

In most cases, the proposed algorithm gives good results. However, in some cases, the results

are not good, because the proposed algorithm is affected by illumination conditions in some images,

zooming and big rotation in some others, see Table 2.

Figure 6: The time chart of the proposed algorithm and comparison methods

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Tim

e (

in

se

co

nd

s)

Images



20

Table 1: Some results of the proposed algorithm and the comparison methods



21

Table 1: Some results of the proposed algorithm and the comparison methods (Continued)



22

Table 2: Some false positive results of the proposed algorithm and the comparison methods

IV. CONCLUSION

In this paper, a new method of face recognition, for faces with expressions, glasses and/or

rotation, based on variance estimation and facial feature extraction is proposed. It can be used in face

recognition systems such as video surveillance, human computer interfaces, image database

management and smart home applications. The proposed algorithm has been tested using a database

of faces and the results showed that it is able to recognize a variety of different faces in spite of

different expressions, rotation and illumination conditions. Zoomed images and their effect on the

recognition of humans need further investigation.

REFERENCES

[1] P. Phillips, P. Flynn, T. Scruggs, K. Bowyer, J. Chang, K. Hoffman, J. Marques, J. Min and

W. Worek, Overview of the Face Recognition Grand Challenge, proceedings of the 2005

IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR),

(2005).

[2] S. Lin, An introduction to Face Recognition technology, informing science issue on

multimedia informing technologies–part 2, 2000, 3(1).

[3] Z. Pan, G. Healey, M. Prasad and B. Tromberg, Face Recognition in Hyperspectral Images,

IEEE Transactions On Pattern Analysis And Machine Intelligence, 25(12), 2003, 1552-1560.

[4] V. Blanz and T. Vetter, Face Recognition Based on Fitting a 3D Morphable Model, IEEE

Transactions On Pattern Analysis And Machine Intelligence, 25(9), 2003.

[5] H. Gu, G. Su and C. Du, Feature Points Extraction from Faces, Image and Vision Computing

NZ, Palmerston North, 2003, 154-158.

[6] S. Phimoltares, C. Lursinsap and K. Chamnongthai, Face detection and facial feature

localization without considering the appearance of image context, Image and Vision

Computing, 25, 2007, 741-753.

[7] Α. Nikolaidis and I. Pitas, Facial Feature Extraction and Pose Determination, Pattern

Recognition, Elsevier, 33(11), 2000, 1783-1791.

[8] H. Koo and H. Song, Facial Feature Extraction for Face Modeling Program, International

Journal Of Circuits, Systems And Signal Processing, 4(4), 2010, 169-176.

[9] G. Yen and N. Nithianandan, Facial Feature Extraction Using Genetic Algorithm,

Proceedings of the Evolutionary Computation, 2, IEEE 2002, 1895-1900.



23

[10] S. Srivastava, Real Time Facial Expression Recognition Using A Novel Method, The

International Journal of Multimedia & Its Applications (IJMA), 4(2), 2012, 49-57.

[11] V. Radha and N. Nallammal, Comparative Analysis of Curvelets Based Face Recognition

Methods, Proceedings of the World Congress on Engineering and Computer Science, 1,

2011.

[12] K. Kumar, S. Prasad, V. Semwal and R. Tripathi, Real Time Face Recognition Using

Adaboost Improved Fast PCA Algorithm, International Journal of Artificial Intelligence &

Applications (IJAIA), 2(3), 2011.

[13] M. El-Bashir, Face Recognition Using Multi-Classifier, Applied Mathematical Sciences,

6(45), 2012, 2235 - 2244.

[14] A. Javed, Face Recognition Based on Principal Component Analysis, International Journal

of Image, Graphics and S ignal Processing, 2, 2013, 38-44.

[15] P. Pattanasethanon and C. Savithi, Human Face Detection and Recognition using Web-Cam,

Journal of Computer Science, 8(9), 2012, 1585-1593.

[16] C. Wu, C. Liu, H. Shum, Y. Xu and Z. Zhang, Automatic Eyeglasses Removal from Face

Images, ACCV2002: The 5th Asian Conference on Computer Vision, 2002.

[17] W. Chen and Y. Gao, Recognizing Face profiles in the Presence of Hairs/glasses

Interferences, 11th. International Conference Control, Automation, Robotics and Vision

(ICARCV), 2010, 1854-1859.

[18] W. Mohamed, M. Heshmat, M. Girgis and S. Elaw, A new Method for Face Recognition

Using Variance Estimation and Feature Extraction, International Journal of Emerging Trends

and Technology in Computer Science (IJETTCS), 2(2), 2013, 134-141.

[19] A. Ford and A. Roberts, Colour Space Conversions, Technical Report,

http://www.poynton.com/PDFs/coloureq.pdf, 1998.

[20] C. Yang and S. Kwok, Efficient Gamut Clipping for Colour Image, Processing using LHS

and YIQ", Optical Engineering Journal, 42(3), 2003, 701–711.

[21] U.K.Jaliya and J.M.Rathod, “A Survey on Human Face Recognition Invariant to

Illumination”, International journal of Computer Engineering & Technology (IJCET),

Volume 4, Issue 2, 2013, pp. 517 - 525, ISSN Print: 0976 – 6367, ISSN Online:

0976 – 6375.

[22] Abhishek Choubey and Girish D. Bonde, “Face Recognition Across Pose With Estimation of

Pose Parameters”, International Journal of Electronics and Communication Engineering

&Technology (IJECET), Volume 3, Issue 1, 2012, pp. 311 - 316, ISSN Print: 0976- 6464,

ISSN Online: 0976 –6472.

[23] J. V. Gorabal and Manjaiah D. H., “Texture Analysis for Face Recognition”, International

Journal of Graphics and Multimedia (IJGM), Volume 4, Issue 2, 2013, pp. 20 - 30,

ISSN Print: 0976 – 6448, ISSN Online: 0976 –6456.

[24] S. K. Hese and M. R. Banwaskar, “Appearance Based Face Recognition by PCA and LDA”,

International journal of Computer Engineering & Technology (IJCET), Volume 4, Issue 2,

2013, pp. 48 - 57, ISSN Print: 0976 – 6367, ISSN Online: 0976 – 6375.

50120140504002

Technology

introduction face recognition

face identification

face dimensions

new face recognition

machine recognition

facial feature recognition

various face images

different face images