IJRE | Vol. 03 No.

05 | May 2016 15

Face Recognition in Non-Uniform Motion Using Raspberry Pi

Janhavi Katle1, Sayali Pujari2,Samrudhi Joshi3 , Tejal Ranpise4,, Prof. Sanjeev Kumar Angadi5
1234
BE IT, 5Assistant Professor,Department of Information Technology,
RMD Sinhgad School of Engineering, Pune, Maharashtra, India.

Abstract: A face recognition system based on the Raspberry II. BACKGROUND

Pi is proposed in this paper. Face recognition system consists
of detection and recognition of using Haar like feature based 2.1 Existing System
cascade classifier. Features of face are extracted using Eigen Face recognition technology has evolved as a solution to
faces algorithm. Proposed architecture generates one full face address the needs in order to perform identification and
review. The various areas in which face recognition system verification of identity claims. A number of face recognition
can be used are home automation for security purpose, systems has been developed by improving the feature
employee login system, automatic student attendance etc. extraction methods and dimensionality reduction techniques in
Digital image features of Haar used in object detection. pattern recognition with high rate of success. MOBILAP is
Adjacent rectangular regions at a specific location in a implemented in existing system. MOBILAP results on the
detection window are considered, sums of the pixel intensities Labeled Faces in the dataset using the unsupervised protocol.
in each region and difference between these sums calculated. System also evaluates the performance of MOBILAP on our
The subsections of an image are categorized using the own real dataset captured using a camera that contains
difference between the sums. The proposed system has an significant blur, illumination and pose variations.
image database with human faces. An observation is made
that among all faces the the region of the cheeks is brighter 2.2 Proposed System
than region of eyes. So a common Haar like feature for face The proposed system combines all the preceding steps into a
detection is a set of two adjacent rectangle. Speed of self contained real time GUI program to allow immediate use
calculation is main advantage of a Haar like feature over most of the face recognition system rather than providing a
other features. Haar like feature of any size can be calculated command line tool that processes image files in an offline
in constant time due to the use of integral images. manner ,one should be able to modify the behavior of the
Keywords:-Raspberry Pi, Haar feature, cascade classifier, system for your own purposes, such as to allow an automatic
Eigen faces algorithm. login of your computer or if you are interested in improving
I. INTRODUCTION the recognition reliability then you can read conference papers
Face recognition is the process of putting a label to a known about recent advances in face recognition to potentially
face. In over two decades face recognition has been one of the improve each step of the program until it is reliable enough for
most interesting and important research fields. It is due to the your specific needs. For example, you could improve the face
need of automatic recognitions and surveillance systems, the preprocessing stages, or use a more advanced machine
interest in human visual system on face recognition, and the learning algorithm, or an even better face verification
design of human computer interface, etc. These researches algorithm.
involve knowledge and researchers from disciplines such as
neuroscience. A bunch of papers have been published to
overcome difference factors (such as illumination, expression,
scale, pose, etc) and achieve better recognition percentage. In
this paper, general ideas and structures of recognition,
important issues and factors of human faces, critical
techniques and algorithms are discussed and finally a
comparison and conclusion is given. Face recognition systems
comes under security systems that are able to automatically
detect and identify human faces. Face recognition systems
depend on a recognition algorithm. But the most of the
algorithm considers somewhat global data patterns while
recognition process. This will not yield accurate detection and Fig.1. Proposed system
recognition system. The proposed system will be developed so III. TECHNOLOGY & TECHNIQUES USED:
that it can be able to give accurate face detection and
recognition. 1) Raspberry Pi:-The Raspberry Pi is a low cost, credit card
Security and authentication of a person is a key part of any sized computer that can be attached into a computer monitor or
organization. Face recognition is one of the many techniques TV, and uses a standard keyboard and mouse. It is a small
used for this purpose. It is an effective means of a human device that enables people of all ages to learn computing, and
authentication the advantage of this approach is that, it makes to learn how to programming languages like Python. It is
easy to detect changes in the face pattern of an individual to a capable of doing everything youd expect a desktop computer
great extent. The recognition system can tolerate local to do, from browsing the internet and playing HD video, to
variations in the face expression of an individual. So face making spreadsheets, word processing, and playing games.
recognition can be used as a key factor in crime detection The Raspberry Pi can interact with the outside world, and has
mainly to identify criminals. been used widely in many applications. Raspberry Pi was
build for kids all over the world to learn to program and
understand how computers work.

ISSN 2348-7852 (Print) | ISSN 2348-7860 (Online) ijre.org

IJRE | Vol. 03 No. 05 | May 2016 16

Fig. 2.Raspberry Pi Model B

Table.1.Cascade Classifier reliability and speed
2) OpenCV: - OpenCV is a library of programming functions
mainly aimed at real time computer vision. It was originally
developed by Intel research center in Russia, later supported IV. DESIGN
by Willow Garage and now maintained by It seez. The library
is cross platform and free for use under the open source BSD
license. OpenCV is a machine learning software library. It
provides an infrastructure for computer vision applications and
accelerates the use of machine perception in the commercial
products. It is a BSD licensed product therefore making it easy
for businesses to utilize and modify the code. The library has
more than 2500 optimized algorithms, which includes a both
classic and state of the art computer vision and machine
learning algorithms. The algorithms can be used to detect and
recognize faces, identify objects, classify human actions in
videos, track camera movements, track moving objects, extract
3D models of objects, bind images together to produce a high
resolution image of an entire scene, find similar images from Fig.3. System Architecture
an image database, remove red eyes from images taken using
flash, follow eye movements,etc. According to a survey over A face recognition system is a computer application capable of
47 thousand people use OpenCV and estimated number of identifying or verifying a person from a digital image.
downloads exceeds 7 million. The library is used largely in Description of process is given as follows:
companies, research groups and by governmental bodies.
1. Image:-
3) Cascade Classifier:-The word cascade in the cascade It is the image which is distorted and need processing or we
classifier means that the resulting classifier consists of several can say that these are the images present in test set and training
simpler classifiers. They are applied subsequently to a region set to test the Face Recognition System.
of interest until at some phase the candidate is rejected.
Cascade classifier supports boosting. At present stage Discrete
Adaboost, Real Adaboost, Gentle Adaboost and Logitboost are
supported. Decision tree classifiers with at least 2 leaves are
basic classifiers. The inputs to the basic classifiers are HAAR Fig.3. Image preprocessing steps
like features, and are calculated as described below. The
current algorithm uses the following Haar like features: 2. Face Detection:-
The key function of face detection step is to determine whether
human faces appear in a given image and location of the faces.
Patches containing each face in the input image is the expected
output of this step. This makes face recognition system easy to
design. Face alignment are performed to justify orientation of
patches. Face detection could also be used for video and image
classification, etc.

3. Feature Extraction:-
In feature extraction step, patches of human face are extracted
from images after face detection. These patches of face are
directly used for face recognition which gives unfavorable
results. This is because each patch usually contains over 1000
pixels, which are too large to build well made recognition
system. Face patches may be taken from different camera
alignments, illuminations, and may suffer from occlusion and
clutter. Feature extractions are performs information packing,
dimension reduction, silence extraction, and noise cleaning.

ISSN 2348-7852 (Print) | ISSN 2348-7860 (Online) ijre.org

IJRE | Vol. 03 No. 05 | May 2016 17
After this step, a face patch is usually transformed into a [5] Eigenfaces for recognition, M. Turk and A. Pentland,
vector with fixed dimension and their corresponding locations. Journal of Cognitive Neuroscience 3, pp. 71-86
Feature extraction can either included in face detection step or [6] Eigenfaces vs. Fisherfaces: Recognition using class
face recognition step. specific linear projection, P.N. Belhumeur, J. Hespanha and D.
Kriegman, Proceedings of the IEEE Transactions on PAMI
4. Face Recognition:- 1997, Vol. 19, Issue 7, pp. 711720
After formulating the representing each face, the last step is to [7] Face Recognition with Local Binary Patterns, T. Ahonen,
recognize the identities of these faces. A face database is A. Hadid and M. Pietikinen, Computer Vision - ECCV 2004,
required to build to achieve automatic recognition. Several pp. 4694
images are taken for each person and their features are
extracted and stored in the database. Then comparison of
features to each face class stored in the database is done. VII. BIOGRAPHIES:-

5. Verification & Identification:- Ms. Janhavi Katle

There have been many researches and algorithms proposed to RMD Sinhgad School of Engg.
deal with this classification problem. Verification and Warje Pune
identification are two general applications of face recognition. Dept. Of Information Technology
If given a face image, we want the system to tell who the
person is or the most probable identification is this is called to
face identification and if given a face image and a guess of the
identification and we want the system to tell true or false about
the guess this is called verification. Ms. Sayali Pujari
RMD Sinhgad School of Engg.
V. CONCLUSION Warje Pune
This paper has shown all the steps required to create a real Dept. Of Information Technology
time face recognition system just using basic algorithms. We
used face detection to find the location of a face within the
camera image, followed by several forms of face
preprocessing to reduce the effects of different lighting
conditions, camera and face orientations. We then trained an
Eigen faces machine learning system with the preprocessed Ms. Samrudhi Joshi
faces we collected, and finally we performed face recognition. RMD Sinhgad School of Engg.
Warje Pune
VI. REFERNCES Dept. Of Information Technology
[1] Rapid Object Detection using a Boosted Cascade of
Simple Features, P. Viola and M.J. Jones, Proceedings of the
IEEE Transactions on CVPR 2001, Vol. 1, pp. 511-518 Ms. Tejal Ranpise
[2] An Extended Set of Haar-like Features for Rapid Object RMD Sinhgad School of Engg.
Detection, R. Lienhart and J. Maydt, Proceedings of the IEEE Warje Pune
Transactions on ICIP 2002, Vol. 1, pp. 900-903 Dept. Of Information Technology
[3] Face Description with Local Binary Patterns: Application
to Face Recognition, T. Ahonen, A. Hadid and M. Pietikinen,
Proceedings of the IEEE Transactions on PAMI 2006, Vol. 28,
Issue 12, pp. 2037-2041
[4] Learning OpenCV: Computer Vision with the OpenCV
Library, G. Bradski and A. Kaehler, pp. 186-190, O'Reilly
Media.

ISSN 2348-7852 (Print) | ISSN 2348-7860 (Online) ijre.org