This project aims at developing a system for acquiring images of finger veins and processing them using MATLAB for the purpose of authentication. It includes designing of hardware for image acquisition, coding the matching algorithm for processing the finger vein pattern and training and testing of algorithm module. Typical Finger vein recognition system consists of image acquisition module, image preprocessing, feature extraction, and matching.
Finger Vein patterns can be viewed through an image sensor sensitive to infrared light. Infrared light passing through the tissues of the human body is blocked by hemoglobin. As hemoglobin exists densely in blood vessels, infrared light passing through veins appears as dark shadow lines in the captured image.
Preprocessing step includes image segmentation in which captured image is divided into multiple parts. Each of the pixels in a segment will be similar with respect to some properties, such as color, texture or intensity. The aim of segmentation is to change the representation of an image into something that is easier to analyze. Image segmentation is used to locate objects and boundaries in an image. Segmentation is the process by which we are assigning a label to every pixel in an image. Pixels sharing the same label will have certain similar visual characteristics.
The finger vein image features are extracted using wavelet transform and line detection. Wavelet transform is a mathematical function which divides a function into its different frequency components .Wavelet transform analyzes each individual component with a resolution that matches its scale.HAAR wavelet transform multiplies a function against the HAAR wavelet with various shifts and stretches. HAAR transform is easy to implement and is able to analyze the local features. These characteristics make HAAR wavelets applicable for Finger vein recognition algorithm.
At last, matching with database is a final decision making step to get a result from the finger vein recognition algorithm. In the matching stage two types of errors are considered FAR (False Acceptance Rate), FRR (False Rejection Rate). FRR is the rate of occurrence of a scenario in which two fingerprints from same finger fails to match (the matching score is below the threshold) while FAR is the rate of occurrence of a scenario in which two fingerprints from different fingers will match (matching score is greater than the threshold). EER is the error rate at which the FAR equals the FRR and is therefore, suitable for measuring the overall performance of biometric recognition system.
Sample image and its feature extracted image are shown below.
In the verification stage, newly captured finger vein image is applied to preprocessing stages, and at last vein image is replaced with the feature extracted image. Finally that extracted image is sent to an authentication stage. This stage will match the newly feature extracted image with the database image, after matching it will create a match score of each finger vein images in the database. Depending on the match score authentication is carried out. This project implements a highly secured authentication system based on using finger vein recognition.