Proposed Scheme for Finger Vein Identification Based on Maximum Curvature and DirectionalFeature Extraction Using Discretization
| Autor: | Wong Yee Leng, Siti Mariyam Shamsuddin, Yuhanim Hani Yahaya |
|---|---|
| Rok vydání: | 2018 |
| Předmět: |
Scheme (programming language)
Discretization Computer science business.industry Feature extraction Process (computing) 020207 software engineering Pattern recognition 02 engineering and technology Identification (information) Feature (computer vision) 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Artificial intelligence business Representation (mathematics) Focus (optics) computer computer.programming_language |
| Zdroj: | Lecture Notes in Electrical Engineering ISBN: 9789811082757 |
| DOI: | 10.1007/978-981-10-8276-4_18 |
| Popis: | Finger vein identification has becoming increasingly noticeable biometric trait. The finger vein pattern provides high distinguishing features that are difficult to counterfeit because it resides underneath the finger skin. The performance of finger vein identification is highly depending on the meaningful extracted features from feature extraction process. Previous works have developed new methods for better feature extraction. However, most of the works focus on how to extract the individual features and not presenting the individual characteristic of finger vein patterns with systematic representation. Therefore, in this paper we propose an improved scheme of finger vein feature extraction method by adopting Discretization method. The finger vein feature extraction is based on combination of Maximum Curvature and Directional Feature (MCDF) feature extraction. After the extraction, the MCDF features value are then fed into Discretization module. The extracted features will be represented systematically by discriminatory feature values. The features values are informative enough to reflect the identity of an individual. The experimental result shows that the proposed scheme using Discretization produce identification accuracy performance above 95.0%. This shows that the proposed scheme produce good performance accuracy compared to non-discretized features. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|