An improved method of identification based on thermal palm vein image

  • Authors:
  • Ran Wang;Guoyou Wang;Zhong Chen;Jianguo Liu;Yu Shi

  • Affiliations:
  • National Key Laboratory of Science & Technology on Multi-spectral Information Processing, Institute for Pattern Recognition and Artificial Intelligence, Huazhong University of Science and Tech ...;National Key Laboratory of Science & Technology on Multi-spectral Information Processing, Institute for Pattern Recognition and Artificial Intelligence, Huazhong University of Science and Tech ...;National Key Laboratory of Science & Technology on Multi-spectral Information Processing, Institute for Pattern Recognition and Artificial Intelligence, Huazhong University of Science and Tech ...;National Key Laboratory of Science & Technology on Multi-spectral Information Processing, Institute for Pattern Recognition and Artificial Intelligence, Huazhong University of Science and Tech ...;National Key Laboratory of Science & Technology on Multi-spectral Information Processing, Institute for Pattern Recognition and Artificial Intelligence, Huazhong University of Science and Tech ...

  • Venue:
  • ICONIP'12 Proceedings of the 19th international conference on Neural Information Processing - Volume Part II
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Biological characteristics based on face, fingerprint and iris images have been extensively studied and used for the identification in the past few decades. As a new-born method, thermal palm vein pattern is gathering more and more attention all over the world. An improved method of identification based on thermal palm vein image is presented in the paper. Five steps are needed to have a person verified: 1) acquisition of infrared palm vein image; 2) detection of ROI (region of interest); 3) enhancement of the palm vein image; 4) features extraction of the palm vein patterns; 5) matching the features between the real palm vein and sample data. Experiments have been carried out on 178 different images and 176 images of them are correctly recognized with two in failure. Experiments show the detection rate is satisfied.