Technical Algorithm
- Principle: Using Infrared ray to capture the specified patterns of veins in fingers.
- Finger vein is the inside feature of finger. The technique of finger vein recognition utilizes the reaction of hemoglobin of human blood and infrared ray with specified wave length. Bones, muscles, fat and skin in the finger will also affect veins, but the patterns and structure of veins cannot be changed. Therefore, CMOS module with high response curve of near-infrared ray and high transfer speed is usually used as finger vein scanner.
Capturing Finger Vein Data
Finger vein image is captured in distribution pattern. Data mainly present in the position and distribution of vein. Terminal points, bifurcation points and turning points provide the largest amount of data and thus used as featured data.
Finger Vein Features Extraction and Matching
Fingerprint and finger vein features matching is the process of matching the present finger vein with the pre-saved finger vein templates, and to determine whether 2 finger vein images are captured from the same finger. There are two major types of matching finger vein: Finger Vein Verification and Finger Vein Identification.
Finger Vein Identification (1:1) means storing a person’s personal information and finger vein features in database in a certain effective format, and re-collecting finger vein image information for 1:1 matching with the stored features and determining whether two features are from the same finger. In the meantime, the matching of finger veins may be performed in various methods including fingerprint, finger vein and fingerprint & vein.
To process finger vein recognition, no other information of the user is needed. Finger vein image data captured on site will be compared with several saved data one by one, and try to match within those data. This is a “1:N” comparison.
Finger Vein Features Extraction and Matching
Vein line characteristic:
Capturing the pattern from finger vein grayscale image. These features present a better topographical structure of veins.
Vein texture characteristic:
During the process of finger vein recognition, the texture characteristics of image are mainly presented in partial binary codes. The binary codes are transformed by the comparison of the greyscale of existing pixels and greyscale of area pixels.
Minutiae point characteristic:
The minutiae points of finger vein recognition refers to the terminal points and bifurcation points of blood vessels.
Learnt features:
Through machine learning methods, features of finger vein can be extracted. For example, through dimensional reduction by Principle Component Analysis to the effective area of finger vein image can be captured the feature of main component amount of finger vein image.
