This repository hosts two projects that dive into the realm of biometric identification using signature and fingerprint analysis. By leveraging image processing techniques and pattern recognition, these projects aim to establish secure and accurate identification systems.
In response to the COVID-19 pandemic, the widespread adoption of face masks presented unique challenges to facial recognition systems. Our research focuses on advancing the efficiency of these systems in distinguishing between genuine individuals and imposters, even when masks obscure facial features. We introduce a novel dataset and present our findings, demonstrating an improvement in the accuracy of imposter classification.
Our dataset is a meticulously curated combination of the CASIA Web Face and CASIA Web Masked Face datasets. It was designed to train models to recognize individuals with a higher degree of accuracy, specifically tailored to the nuances introduced by mask-wearing.
Access our comprehensive dataset here.
-
Initial Findings Presentation: An overview of our initial hypotheses and research approach. View Presentation
-
Research Paper Presentation: Detailed insights and references that underpin our research findings. Explore the Presentation
-
Research Paper: The full research paper documenting our methodology, data analysis, and conclusions. Download PDF
This project is designed to detect signature forgeries using structural similarity indexes and image processing methods. It demonstrates the ability to differentiate between genuine and forged signatures effectively.
- Import Dataset
- Perform Canny Edge Detection
- Apply Morphological Operations for image enhancement
- Compute Structural Similarity Index of genuine and Forged Signatures
The goal of this notebook is to develop a fingerprint identification system. The system processes the fingerprint images, enhancing and extracting unique features such as minutiae to match fingerprints with high precision.
- Fingerprint segmentation
- Estimation of local ridge orientation
- Estimation of local ridge frequency
- Fingerprint enhancement
- Detection of minutiae positions
- Estimation of minutiae directions
- Creation of local structures
- Fingerprint comparison
- Apply the entire process on a new dataset
- Create a pipeline
To utilize these notebooks:
- Clone the repository.
- Install any required dependencies listed in the notebooks.
- Run the Jupyter Notebooks in an environment that supports Python and necessary libraries like OpenCV, NumPy, and SciPy.
- Familiarity with Python and Jupyter Notebooks
- Understanding of image processing and machine learning basics
We welcome contributions and suggestions. Feel free to fork the repository, make changes, and submit a pull request.