A micro-control capture images technology for the finger vein recognition based on adaptive image segmentation

The advantages of finger-vein recognition, compared to other biometric recognition technology commonly enlisted in identification needs such as ATM and door security, are uniqueness and living recognition, and so it has recently become one of major topics in biometric investigation. Through the irradiation of near-infrared rays, the finger-vein images are generated by blood vessels, and the texture of the image is created by the level of transparency between skeleton joints of the finger. Feature extraction of most finger-vein images employs global characteristics instead of local characteristics. This paper presents a local features method for dealing with the four segmentation of finger-vein images according to the physiological characteristics of finger. The four segmentation will be separated by a dynamic boundary line determined by the global vertical statistical quantity of the finger-vein image. Afterwards, each segmentation will be added with weighted values to enhance the fidelity of recognition. This study employs a total of 3816 finger-vein images generated by 106 adults in age of 20–60 years. The forefinger, middle finger, and little finger of each adult will be sampled for six images. Finally, the average recognition proposed by this study reaches 97%. The significant accuracy and contribution of this study is illustrated when compared to methods of global feature extraction (74.55%) and fixed-four regional segmentation feature extraction (86.48%).