VeinRecognitionSystem

Overview

This project aims to implement an automated personal identification system using finger vein biometrics. Utilizing Convolutional Neural Networks (CNN) and traditional machine learning methods, the system provides robust, efficient identification while ensuring user privacy.

Table of Contents

1. Introduction
2. Objectives
3. Existing vs Proposed System
4. Modules
5. Process Model
6. Software Requirements Specification
7. Data Flow Diagram
8. Running the Project
9. Conclusion

Introduction

Automated personal identification has become critical in various sectors such as e-governance and security. Finger vein identification leverages unique anatomical patterns, making it a reliable alternative to more commonly altered biometric traits like fingerprints and facial features. This project focuses on using CNNs to improve the accuracy and efficiency of finger vein authentication systems.

Objectives

• To develop an automated personal identification system using finger vein images.
• To compare the performance of CNN with traditional machine learning algorithms such as Support Vector Machines (SVM).
• To reduce template size while enhancing prediction accuracy.

Existing vs Proposed System

Existing System

• Hand-Crafted Features: Current systems rely heavily on manually designed features.
• Disadvantages:
  • Longer processing time
  • Lower accuracy

Proposed System

• Machine Learning & Deep Learning: Utilizes SVM and CNN algorithms.
• Advantages:
  • Reduced processing time
  • Increased accuracy and prediction capabilities

Modules

1. Upload Finger Vein Dataset: Load the dataset and visualize finger images.
2. Preprocess Dataset: Resize, shuffle, and normalize images, splitting into training (80%) and testing (20%).
3. Run SVM Algorithm: Train an SVM model and evaluate prediction accuracy.
4. Run CNN Algorithm: Train a CNN model and evaluate prediction accuracy.
5. Comparison Graph: Visualize accuracy comparison between SVM and CNN.
6. Identify Finger Vein from Test Image: Upload a test image for identification using the trained CNN model.

Process Model

SDLC (Software Development Life Cycle)

• Stages:
  • Requirement Gathering
  • Analysis
  • Designing
  • Coding
  • Testing
  • Maintenance

The SDLC framework ensures a systematic approach to developing, testing, and maintaining the software.

Software Requirements Specification

Overall Description

A comprehensive SRS that includes:

• Business Requirements: What must be delivered for value.
• Product Requirements: Characteristics of the system.
• Process Requirements: Activities undertaken by the development organization.

Feasibility Study

1. Economic Feasibility: Cost vs. benefits analysis.
2. Operational Feasibility: Meeting organizational operating requirements.
3. Technical Feasibility: Ensuring a robust and secure implementation.

Data Flow Diagram

A DFD illustrates how data flows through the system:

1. Upload Dataset
2. Dataset Upload Success
3. Preprocess Dataset
4. Dataset Preprocess Success
5. Run SVM
6. SVM Run Success
7. Run CNN
8. CNN Run Success
9. Predict Disease
10. Prediction Success

Running the Project

1. Initialize: Double-click run.bat to start the application.
2. Upload Dataset: Click the "Upload Finger Vein Dataset" button to load the dataset.
3. Preprocess Data: Click the "Preprocess Dataset" button to prepare images.
4. Run Algorithms: Train both SVM and CNN models via their respective buttons.
5. Comparison: Generate a comparison graph to visualize the performance of both algorithms.
6. Identify Test Image: Upload a test image for identification.

Conclusion

The proposed system demonstrates significant improvements in both accuracy and efficiency for finger vein identification compared to existing methods. By leveraging advanced machine learning techniques, this project enhances security while maintaining user privacy.

---