Gradient Based Image Segmentation for Vein Pattern

The biometrics identification technology based on vein pattern has being developed rapidly in recent years. However, the vein image acquired by near-infrared (NIR) imaging device has low contrast and usually has a mount of noise which make an effective image segmentation be a great challenge. Thresholding is a popular vein image segmentation method, which is easy to calculate and can shorten the recognition time. But the thresholding method is limited for the vein image with low contrast and lots of noise. In this paper, we propose a new vein image segmentation method based on gradient enhancement to resolve the conflict between low contrast vein image and high quality image segmentation. Using local gradient in Sauvola equation, we got better effect in our own hand dorsal vein image database.