The fingerprint classification is conducted on PolyU's (Hong Kong Polytechnic University) research database with 336 individuals. Our approach introduces computer vision pre-processing methods to capture regions of interest in fingerprint images to allow effective feature extraction.
Downloads - Project Report / Slides
- Design a Convolutional Neural Network (CNN) for feature extraction.
- Prepare dataset for pre-processing & model training.
- Tune model (fit) for multi-class classification prediction
- Evaluate model with the test samples (unknown samples)
- This was our final year ECE BTech project, which aims to investigate the performance of the state-of-the-art CNN-based Deep learning techniques as an alternative to the conventional minutiae-based fingerprint identification.
- Our proposed method achieved a classification accuracy of 94.26%.
- Chenhao Lin, Ajay Kumar, "Matching Contactless and Contact-based Conventional Fingerprint Images for Biometrics Identification," IEEE Transactions on Image Processing, vol. 27, pp. 2008-2021, April 2018.