Finger-vein recognition based on improved Zernike moment

Abstract

In modern time, finger-vein recognition technology has become increasingly popular. Basically, the finger-vein recognition process involves finger-vein image acquisition, feature extraction and recognition. The recognition algorithm is the key research issue. Because of the differences between acquisition devices and individuals, the performance of the algorithm is affected by image rotation, translation and noise in the process of finger vein recognition. To solve the problems, a recognition algorithm based on an improved Zernike moment is proposed in this paper. Firstly, Sobel operator is used to perform ROI extraction from the original finger-vein image, effectively. Secondly, a series of image processing techniques, such as Curvature detection, threshold segmentation, image thinning and expansion, are adopted to extract the finger-vein lines and generate a registration template. Lastly, the Zernike moment features of the image and a combination matching algorithm of template matching and Hu moment matching, are utilized to realize the accurate identification of the finger vein. The system was verified by experiments and the results indicate that our algorithm can greatly reduce the influence of image rotation, translation, and noise in the recognition process.