Ubiquitious Computing system : Integrating RFID with Face Recognition systems
1. Ubiquitious Computing system : IntegratingUbiquitious Computing system : Integrating
RFID with Face Recognition systemsRFID with Face Recognition systems
Shahryar Ali
Bahria University – Final year Project
BS- Telecom
2. Project OverviewProject Overview
Student Attendance System using RFID and Face
Recognition
RFID Tag (Card) for each Student
Student passes his tag through the RFID Reader for
Attendance
Camera is used to capture the image of the student.
Face Recognition (PCA) is used for student verification.
3. Benefits and FeaturesBenefits and Features
Automatic Attendance
Records Entry Time of Students
No Class Attendance Sheets
No Need to Submit Attendance Sheets
Efficient and Error free
4. Introduction to Radio FrequencyIntroduction to Radio Frequency
Identification (RFID)Identification (RFID)
Automatic identification technology
Radio frequency (RF) waves
Reads Encoded Digital Data
Barcode v.s RFID
Applications in Military, Government agencies, Library
and Time and access control.
5. Components of RFID System:
RFID Tag
RFID Reader/Interrogator
RFID Middleware/Host PC
Types of RFID System:
1. Active vs. Passive
2. Read Only vs. Read/Write
6. Working of RFID SystemWorking of RFID System
Methods of Power Transfer
Magnetic Induction
Electromagnetic wave capture
Near field v.s Far field Communication
Passive RFID Basics
Database Management
Application
7. Frequency RangesFrequency Ranges
FREQUENCY
RANGE
BAND READ RANGE ADVANTAGES DISADVANATAGE
S
APPLICATIONS
LF 125-135 kHz Below .5
meter
Accepted
Worldwide
Short read range,
slow read rates
Access Control
, Animal
Tracking
HF 13.56 MHz Below 1 meter Quick read rates Require high
power
Item tracking,
libraries
UHF 860-969
MHz
3 meters High read
range, very
quick read rates
Doesn’t operate
well near water
or metals
Supply chain,
parking lot
access
Microwave 2.45 GHz 1 meter Fastest read
range
Clear path
required
Supply chain
8. RFID StandardsRFID Standards
ISO STANDARD CATEGORY DEFINED TOPICS
ISO 11784 Animal Tracking Frequency, Baud-Rate, Code
Structure
ISO 18000 Air Interface Standard Frequency, Anti-collision Protocol
ISO/IEC 14443 Proximity Cards Frequency, RF Power, Physical
layer
ISO 15693 Vicinity Cards Frequency, RF Power, Physical
layer
ISO 15691 Supply Chain and Item
Management
Data Processing , Application
Commands
ISO 18047-4 Testing Purposes System Functionality
9. Selection of RFID system:
RFID Reader Selection Criteria
RFID Tag Selection Criteria
RFID System for the Implementation:
PUA-310-0 Proximity Reader
Standard Light Proximity Card
10. Reader SpecificationsReader Specifications
Brand Pegasus
Model PUA-310-0
Country of Origin Taiwan
Reading Range 5cm
Suitable Temperature -10 to 55 °C
Power supply 12 V DC
Operating Frequency 125 KHz
Modulation Type ASK
Interface RS-232
Dimension 116 L x 76W x 13H mm
Weight 250grams
Cost Rs.7000
12. Installation and TestingInstallation and Testing
Installation:
• Environmental Analysis
• Cabling
• Transmission Rate
• Power
Testing :
• HyperTerminal Settings
• Reader Output
13. 12-digit ASCII code
“Start of Text” O2H and “End of Text” O3H
HEX Equivalent: 02H , 30H, 36H , 30H, 30H, 33H, 33 H, 32H,
35H, 31H, 41H, 03H
Total Card Length is 12 Bytes(96-bit)
14. Implementation of StudentImplementation of Student
Attendance SystemAttendance System
Application Overview
Tool for Implementation
Serial Communication in MATLAB
Database of RFID Tags in MATLAB
Programming the Application
Demonstration in MATLAB GUI
Recording Information in Excel Sheet
15.
16. Programming the ApplicationProgramming the Application
Key-less System
Start of Attendance
Start Time of Class
Student Attendance
Security
Student Entry Time
End of Attendance
End Time of Class
20. Problem and the Need of FaceProblem and the Need of Face
RecognitionRecognition
Illegal use of RFID Card
Not considered in Time and Access Control systems.
Use of Biometrics
Real-time Face Recognition System
21. Introduction to Face RecognitionIntroduction to Face Recognition
Unique facial features
Identification
Security Systems
Real-time
`
Image Processing
22. How Face Recognition SystemHow Face Recognition System
Works?Works?
Image Acquisition Face Detection Feature Extraction
Feature Matching
Face
Database
Face Classification
23. Methods of Face Recognition:
Principal Component Analysis (PCA)
Independent Component Analysis (ICA)
Linear Discriminant Analysis (LDA)
Problems in Face Recognition:
Variation in Scale
Variation in Orientation
Variation in illumination
Variation in expression
25. Recognition of FacesRecognition of Faces
Representation of Images
Face Databases
• The Color FERET Database
• The Yale Face Database
• The ORL Database of Faces
26. Flow Diagram (PCA)Flow Diagram (PCA)
Image Acquisition
(Face Database)
Normalization of
facial Images
Calculation of Mean
Image
Calculation of
EigenFaces
Calculation of Weights of
EigenFaces
Unknown Input
Image Normalization
Calculate the
Euclidean Distance
Calculation of weight
of input Face Image
Classification of Input
Face Image
28. Transformation of Images:
Represent an N x N image as N^2-dimensional vector.
Calculate the Mean Facial Image:
Mean image is the average of all images.
29. Subtract the Mean Image from Sample
images:
Each Face images differs from the average face image by a
vector.
Calculation of Eigenfaces:
Let’s assume a face image is of size 128 x 128.
After transforming the image into N^2 dimension, it becomes a
vector of dimension 16384.
Task is to describe these images in low dimensional subspace i.e.
Data Reduction.
Calculate Covariance Matrix which consists of all image vectors.
30. Size of the covariance matrix (C=A AT ) is N^2 x N^2.
Computationally not feasible to calculate.
Construct a matrix L = ATA of size M x M where M = Number of
Training Images.
Eigenvectors of “L” are calculated.
From these M eigenvectors, only P are chosen according the
criterion.
These P Eigernvectors are called Eigenfaces.
A= [Ф1 Ф2 …… ФM ]
31. Calculation ofWeightVectors:
After transformation of the images into eigenfaces,the weight
vectors are formed of both Training and Test Images.
Calculate the Euclidean Distance:
The Euclidean distance between two weight vectors provides a
difference between the two images i and j.
d( i, j) = || i - jΩ Ω Ω Ω ||2
where ω = weight, μ = eigenvector, Γ = new input image, Ψ = mean face
32. Implementation of Face RecognitionImplementation of Face Recognition
(PCA) in MATLAB and Results(PCA) in MATLAB and Results
Face Databases:
• Yale Face Database
• Real-Time Face Database
Implementation in MATLAB:
• Acquisition Module
• PCA module
• Recognition Module
• Classification Module
40. Recognition Rate using Different NumberRecognition Rate using Different Number
of Training and Test Imagesof Training and Test Images
Number of Training
Images
Per Individual
Number of Testing
Images
Per Individual
Recognition
Rate
(%)
1 10 15.00
2 9 25.00
3 8 40.00
4 7 62.00
5 6 71.00
6 5 70.00
7 4 75.00
8 3 77.00
9 2 86.60
10 1 86.60
42. Camera for Image AcquisitionCamera for Image Acquisition
Logitech 2 MP HD Webcam C600
Brand/Model Logitech C600
Technology RightLight Technology
Sensor True 2-Mega pixel
Resolution 1600 x 1200
Frames per Second Upto 30 fps
Others USB 2.0 , Fixed focus
43. Face Detection LibraryFace Detection Library
Face Detection library created by W. Kienzle, G. Bakir, M. Franz
and B. Scholkopf is used.
It is non-commercial Dynamic-link library.
The MATLAB version is used which implements a single function.
The method of reduced state vectors machines to increase the
accuracy of Face Detection is used.
It detects the faces by adjusting the threshold.
The Image must be uint8 grayscale.
The Red Rectangle shows the face detected.
Library has the capability to detect all the faces in the image.
44. The program is created to crop the detected face.
All the Training Images of six students taken at Bahria University
are captured with their faces detected using this library.
Each image is of size 128 x 128.
These 60 images form a Real-time face Database.
45. The complete Real-time Face Database of 6 students (10 images
per Individual).
46.
47. Recognition Rate using Different NumberRecognition Rate using Different Number
of Training and Test Imagesof Training and Test Images
Number of Training
Images
Per Individual
Number of Testing
Images
Per Individual
Recognition
Rate
(%)
1 9 64.0
2 8 68.3
3 7 74.0
4 6 77.0
5 5 79.7
6 4 87.5
7 3 87.5
8 2 91.2
9 1 95.0
48. Integration of RFID and FaceIntegration of RFID and Face
RecognitionRecognition
Key-less System
Start of Attendance
Start Time of Class
Student Attendance
Security
Student Entry Time
End of Attendance
End Time of Class
Student Face Verification
Student Face Verification Record
49.
50. Conclusion and Future workConclusion and Future work
Attendance system using RFID and Face Recognition is a very
unique application.
Attendance and access control systems are common in most of
the modern offices.
But they don’t consider the fact that RFID can be very insecure if
some illegal person gets access to the RFID card.
Integration with face recognition system solves this problem.
A student is identified by RFID card and is verified by the Face
Recognition system.
51. The application can be enhanced to take the attendance of all
university students .
With the use of multiple RFID and Face Recognition systems for
different classrooms.
To handle a large database, a Database management system
would be required.
It can easily be integrated in MATLAB using the Database
Toolbox.