A Simple and Efficient Method for Finger Vein Recognition.

Area of Science:

• Biometrics
• Pattern Recognition
• Computer Vision

Background:

• Finger vein recognition is a promising biometric technology due to its high distinctiveness and security.
• Traditional methods often suffer from redundant feature data and reliance on image segmentation.
• Existing approaches face challenges with data redundancy and segmentation sensitivity.

Purpose of the Study:

• To propose a new finger vein recognition algorithm using code generation.
• To overcome limitations of traditional methods, such as data redundancy and segmentation dependency.
• To develop a computationally simple and data-efficient biometric identification technique.

Main Methods:

• Finger vein images are divided into blocks to compute mean values.
• Centrosymmetric coding is applied to the generated matrix to create feature codewords.
• Image segmentation is not required, simplifying the process.

Main Results:

• The proposed method generates concise feature codewords, reducing data requirements.
• Similarity is effectively measured using the ratio of minimum Hamming distance to codeword length.
• Experiments on public databases demonstrate superior performance compared to state-of-the-art methods.

Conclusions:

• The novel code-generating algorithm offers a simple, efficient, and accurate approach to finger vein recognition.
• The method shows significant potential for practical applications in biometric matching.
• This technique addresses key challenges in current finger vein recognition systems.