On-the-Fly Metacarpus Vein based Biological Recognition based on Image Processing

Abstract

The security of Palm vein technology is paramount as it ensures that no one can replicate the technology. The vein pattern of the human palm is employed in this palm system for identity verification. This technique is more secure than other biometric systems as it is a contactless system and each person has a different vein pattern. The main objective is to build a unique, cheap and reliable system to replace the contact-based systems. Images from a near-infrared camera are used as they help in creating the clear veins required for the ideal functioning of the system. To determine the appropriate profile, we must process the image and perform grayscale thresholding before inverting it. Image segmentation is performed on the image to obtain meaningful edges. Morphological operations are employed to describe the pattern of veins after removing noise from images. The resulting grayscale image is divided into 20x20 sub-regions before feature matching.Feature vectors for these sub-region are used to represent mean absolute deviation (MAD) These feature sets form the input to the backpropagation neural network.Therefore, the accuracy rate of classification using the feature matching method can goup to75%.