Signature Forgery Verification
ABSTRACT:
This paper presents an innovative approach for signature verification and forgery detection
based on fuzzy modelling. The signature images are binarized and resized to a fixed size
window and are then thinned. The thinned image is then partitioned into a fixed number of
eight sub-images called boxes. This partition is done using the horizontal density
approximation approach. Each sub-image is then further resized and again partitioned into
twelve further sub-images using the uniform partitioning approach. The features of
consideration are normalized vector angle (a) and distance (γ) from each box. Each feature
extracted from sample signatures gives rise to fuzzy sets. Since the choice of a proper
fuzzification function is crucial for verification, we have devised a new fuzzification function
with structural parameters, which is able to adapt to the variations in fuzzy sets. This function
is employed to develop a complete forgery detection and verification system.
EXISTING SYSTEM:
Signature verification and forgeIy detection relate to the process of verifying signatures
automatically and instantly to determine whether the signature is genuine or forged. There
are two main types of signature verification: static and dynamic. Static, or off-line
verification is the process of verifying an electronic or paper signature after it has been
made, while dynamic or on-line verification takes place as a subject creates his signature on
a digital tablet or a similar device. The signature in question is then compared to
previous samples
of the signer's signattire, which constitute the database or knowledge base. In the case of
an ink signature on paper, the computer requires the sample to be scanned for analysis,
whereas a digital signature is already stored in a data format that signature verification can
use.
Disadvantages
1. It takes more time
2. Less accuracy
PROPOSED SYSTEM:
In this project we per your instructions we have employed CNN and HOG algorithm to
extract features from signature images and then employ Decision tree to select relevant
features from CNN extracted features and then employ SVM, KNN and LSTM algorithm on
selected features to detect images as Original or Forge. Each algorithm performance is
evaluated in terms of accuracy, precision, recall and FSCORE. Among all algorithm LSTM is
giving best accuracy.
Advantages
[Link] takes less time
[Link] prediction result
Modules Description:
To implement this project we have designed following modules
1) Upload Signature Forgery Dataset: upload dataset to application and then read all
images and extract HOG features from each image
2) Process Dataset: this module will shuffle, normalize and split dataset into train and
test where application using 80% dataset for training and 20% for testing
3) CNN & HOG Based Features Extraction: employ CNN algorithm to extract features
from HOG generated images data
4) Decision Tree Based Features Selection: apply decision tree to selected features from
CNN extracted features
5) Run SVM Algorithm: 80% training with selected features will be input to SVM
algorithm to train a model and this model will be applied on 20% test data to calculate
prediction accuracy
6) Run KNN Algorithm: 80% training with selected features will be input to KNN
algorithm to train a model and this model will be applied on 20% test data to calculate
prediction accuracy
7) Run LSTM Algorithm: 80% training with selected features will be input to LSTM
algorithm to train a model and this model will be applied on 20% test data to calculate
prediction accuracy
8) Comparison Graph: will plot comparison graph between all algorithms Forge
Detection from Test Image: will upload test image and then application will apply
trained model to classify that image as original or Forge signature
SYSTEM REQUIREMENT:
Hardware Requirements:
Processor - I3(min)
Speed - 1.1 GHz
RAM - 4GB(min)
Hard Disk - 500GB
Software Requirements:
Operating System - Windows 10/above
Programming Language - Python 3.7.0