Towards Building a Better Biometric System Based on Vein Patterns in Human Beings

Until a couple of years ago, lifting fingerprints from photos was considered to be an impossible task. But with all the technological advancements coupled with the ingenuity of hackers, spoofing fingerprint information is not so difficult now. Add to this, the fact that human fingerprints are not always unique, and we see that we need to stop relying heavily on fingerprint-based security systems. So a project that extracts the vein distribution patterns in human palms was devised as a solution to this problem. Biometric identification using IR images of palm vein patterns offers more distinctiveness, robustness and difficulty to plagiarize when compared to similar mechanisms. De-oxidized blood in human veins has a property of absorbing the light rays emitted by an infrared source. This property of IR absorption makes the veins distinctively visible when exposed to IR rays. An image of a human palm is prone to noise when captured under infrared light. Canny's edge detection algorithm was employed on an image dataset where all the images are preprocessed using several steps which will be discussed in detail in the following sections. The detected edges act as authentication information for security systems. A feature extraction algorithm is implemented on these edge-detected images and the result is used to classify the subjects.