CCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete Record
Signature Recognition Poster
1. FRAUD DETECTION USING SIGNATURE RECOGNITION
TEAM ID :- 44039
INTRODUCTION
IMPLEMENTATION
REFERENCES
CONCLUSION
ABSTRACT
Training Signature
Images
Feature Extraction
Parameter Extraction
Global Extraction
Local Extraction
Test Signature
Images
Preprocessing
Gray Scale
Noise Removal
Border Elimination
Edge Detection
Image Normalization
Recognition and Verification
Genuine or Forgery
Feature Database
(SQLite)
Preprocessing
Gray Scale
Noise Removal
Border Elimination
Edge Detection
Image Normalization
Feature Extraction
Parameter Extraction
Global Extraction
Local Extraction
SYSTEM FLOW - CHART
Canny Edge Detection
Gray Scale conversionGray Scale
Capture Image
ORB Feature Extract
RESULT ANALYSIS
Biometric Recognition is an automatic Identification of an
Individual based on physiological and behavioral characteristics.
Biometrics traits (Signature, Voice, Iris, Fingerprint, Face etc.) are
preferred to traditional methods like Password, PIN number,
smartcards etc. The Biometric Characteristics of Individuals are not
easily transferable, they are Unique and not been stolen. Within
Biometric methods, Automatic Signature Recognition is an
important research area because of the social, legal and wider
acceptance of Handwritten Signature as Identification.
A Handwritten Signature is an important attribute of human
being which can be used to authenticate human identity. Signature
verification system involves concept that are forgeries, features and
performance evolution parameters.
Types of Forgery:
1. Random
2. Skilled
3. Simple
Performance Evolution Parameter:
1. False Acceptance Rate (FAR)
2. False Rejection Rate (FRR)
The signature of person is important biometric of human being
which can be used to authenticate human identity and provide
authorization of legal document. Signature verification used in
Banking , Passport verification , Public examination , Bank -
Cheque.
When any unauthorized person wants to imitate signature and
steal their identity at that time fraud things happened. So therefore it
has long been target of fraudulence and signature verification is very
important.
The image of signature is collected by camera of mobile which
extract some spatial and dynamic features based on image
processing techniques like Gray scale conversion, Noise removal ,
Normalization and Feature Extraction techniques. The signature
matching is performed by matching algorithms.
FEATURES
It does not required any external hardware.
It provide research area for android application
developer to work in field of artificial intelligence.
It allow user to identify fraud things.
It Provide Mobility.
It allow users to add new set of training signature
images to train network which recognize signature fraud or
genuine.
[1] Ashish Dhawan, Aditi R. Ganesan, “Handwritten
Signature Verification”, The University of Wisconsin..
[2] Brooks, F. (1995) The Mythical Man Month, Addison-
Wesley.
[3] K.A. Vala, N. P. Joshi, “A Survey on Offline Signature
Recognition and Verification Schemes”, International
Journal of Advanced Research in Electrical, Electronics and
Instrumentation Engineering, Gujarat, India, March-2014.
[4] Madhuri Yadav, Alok Kumar, Tushar Patnaik,
Bhupendra Kumar, “A Survey on Offline Signature
Verification”, International Journal of Engineering and
Innovative Technology (IJEIT), January-2013.Result
Akshay R. Panchal (120840131027)
Santosh M. Ladani (120840131053)
Tejraj G. Thakor (130843131018)
Dhruvin L. Bhalodiya (120840131021)
Prof. Bhagyasri G. Patel
Department of Computer Engineering, FETR, Bardoli.
Guided By:
In our proposed Android Application which is based on
Off-line Signature Verification Approach. In Our
proposed application first apply some image processing
techniques so we can get normalized form of image after
that with use of some feature extraction techniques.
Our Application extract the maximum features from
the signature images, which is then match with the
training sets of images using the Support Vector Machine
(SVM) approach which is quite easy to implement and
provide better accuracy than other signature matching
techniques. Our proposed system is mainly developed for
to reduce the fraud level in daily business transaction and
in other government official paper work.