Elsevier

Pattern Recognition

Volume 36, Issue 2, February 2003, Pages 303-312
Pattern Recognition

An effective algorithm for fingerprint image enhancement based on wavelet transform

https://doi.org/10.1016/S0031-3203(02)00032-8Get rights and content

Abstract

Among all the fingerprint identification/verification systems, such as minutiae-based or filterbank-based fingerprint matching, the performance relies heavily on the quality of the input fingerprint images. In this paper, we propose an effective algorithm of fingerprint image enhancement, which can much improve the clarity and continuity of ridge structures based on the multiresolution analysis of global texture and local orientation by the wavelet transform. Experimental results show that the enhanced image quality by using the wavelet-based enhancement algorithm is much better than the other existing methods for improving the minutiae detection.

Introduction

There are many human biometric features that can be used to confirm the identity, such as voice, hand geometry, face, fingerprint and retinal pattern of eyes. Among all these features, fingerprint matching is the most popular and reliable biometric technique for automatic personal identification. A fingerprint is the pattern of ridges and valleys on the surface of a fingertip. A total of 18 different types of local ridge/valley descriptions have been identified [1], and most of them are not used in automatic fingerprint identification system (AFIS). Instead, in accordance with the representation of fingerprints in the U.S. Federal Bureau of Investigation (FBI), ridge endings and bifurcations are taken as the distinctive features of the fingerprints, and the coordinates and the angle of the features are used to represent the fingerprint in the matching process [2]. A ridge ending is defined as the point where a ridge ends abruptly. A ridge bifurcation is defined as the point where a ridge forks or diverges into branch ridges. Examples of ridge ending and bifurcation are shown in Fig. 1.

The performance of an automatic fingerprint identification system relies heavily on the quality of input fingerprint images. Poor quality image badly affects the extraction of ridge minutiae and directions, so that it causes decline in the performance. Accordingly, for constructing an effective fingerprint identification system, a robust enhancement algorithm is necessary. Directional behavior is the most obvious characteristic in fingerprint image, hence there are extensive literatures on this subject [3], [4], [5], [6], [7], [8], [9], [10], [11], [12], [13]. Abutaleb et al. [3] use the fact that a fingerprint is made of white followed by black lines of bounded number of pixels in time domain. They use the genetic algorithm to generate black and white lines of different widths, and then translate the enhancement technique into the optimization problem while getting the best match with the original fingerprint. From viewing in the frequency domain, ridges and valleys in a local neighborhood form a sinusoidal-shaped wave, which has a well-defined frequency and orientation. Thus some techniques take advantage of this information to enhance gray-level fingerprint images [4], [5]. Hong et al. [6], [7] use Gabor filters as bandpass filters to remove the noise and prevent true ridge/valley structures because the filter has both frequency-selective and orientation-selective properties and has optimal joint resolution in both spatial and frequency domains [14]. Kasaei et al. [8] use the concept of statistics to calculate the Dominant Ridge Direction (DRD) for each 16×16 block. After deriving the DRD, they rotate the block image according to the dominant orientation for image projection to enhance the local ridge veins. Wahab et al. [9] define eight directional masks of spatial domain to find the local orientation of the ridge in 8×8 block. Because of the similarity of neighborhood region in fingerprint image, averaging the pixel values along the direction of ridge with their neighborhood is calculated for preserving the continuity of ridges. In addition to these enhancement techniques, design of fingerprint filters is discussed in Ref. [10] and adaptive window sliding on the grayscale image for image enhancement is presented in Ref. [11]. Wavelet theory covers quite a large area and is widely used in numerous applications. Lee et al. [12] use wavelet transform to decompose a fingerprint image into several directional sub-images. Orientation features and coherence features within these regions are calculated to recognize the fingerprint. A novel approach is also proposed by Maio et al. [13]. They extract the minutiae directly from the grayscale image without binarization and thinning. The basic idea of their method is to follow the ridge lines on the grayscale image, by “sailing” according to the fingerprint directional image.

As mentioned above, we can see that all the researches are focused on the characteristic of local ridge orientation for improving the quality of fingerprint image, but relatively the global information is ignored. Therefore, we propose an effective enhancement method, which combines the texture unit of global information with the ridge orientation of local information. Based on the characteristic of hierarchical framework of multiresolution representation, the wavelet transform is used to decompose the fingerprint image into different spatial/frequency sub-images by checking the approximation region. In considering the global information, all the texture units within an approximation sub-image are filtered by textural filters. Based on the hierarchical relationship of each resolution, all other sub-images are processed according to the related texture units within the approximation sub-image. Similarly, the directional compensation is implemented by voting technique in considering the orientation of local neighboring pixels. After these two enhancement processes, the improved fingerprint image is obtained by the reconstruction process of wavelet transform.

This paper is organized as follows. In Section 2, we briefly review the theory of wavelet transform. The application of wavelet-based textural filtering and directional recovery on ridge configuration for improving the fingerprint image is described in Section 3. Experimental results are presented in Section 4. Concluding remarks are given in Section 5.

Section snippets

Review of wavelet transform

Wavelets are families of functions ψj,k(t) generated from a single base wavelet, called the “mother wavelet”, by dilations and translations, i.e.,ψj,k(t)=2j/2ψ(2jt−k),j,k∈Z,where Z is the set of all integers, j is the dilation (scale) parameter and k is the translation parameter. Our goal is to generate a set of expansion functions such that any signal in L2(R) (the space of square integrable functions) can be represented by the seriesf(t)=j,kdj,kψj,k(t),where the two-dimensional set of

Wavelet-based fingerprint enhancement

The purpose of a fingerprint image enhancement algorithm is to improve the quality of an input image for facilitating the classification or recognition tasks. Thus, a segmentation process for capturing the fingerprint region within the input image is needed. In order to segment the fingerprint area, the input image is divided into non-overlapping blocks and a simple variance analysis on gray-level values is used for all blocks to distinguish between background areas and foreground areas (ridges

Experimental results

The wavelet-based enhancement algorithm is evaluated on 100 images of the commercially available standard database NIST 9 (Vol. 1, CD. no. 1) [20]. The CD contains 1800 fingerprint images (image size =832×768),900 images of card type 1, and 900 images of card type 2 obtained from 900 different fingers. Fingerprints on the card type 1 are scanned by using a rolled method, and fingerprints on card type 2 are scanned by using a live-scan method. A large number of NIST 9 images has much poorer

Summary

We have proposed an effective wavelet-based method for enhancement of fingerprint image, which uses both the global texture and local orientation characteristic. Based on the hierarchical relationship of 2D wavelet transform, all the detail sub-images are reconditioned by reference to the related location of the approximation sub-image. This mechanism not only saves the computational time but also effectively improves the quality of fingerprint image. One hundred image samples selected from the

About the Author—CHING-TANG HSIEH is an associate professor of Electrical Engineering at Tamkang University, Taiwan, Republic of China. He received the B.S. degree in Electronics Engineering in 1976 from Tamkang University and the M.S. and Ph.D. degree in 1985 and 1988, respectively, from Tokyo Institute of Technology, Japan. From 1990 to 1996, he held the post of Chairman of the Department of Electrical Engineering. His current research interests include speech analysis and synthesis, speech

References (22)

  • T. Kamei, M. Mizoguchi, Image filter design for fingerprint enhancement, Proceedings of ISCV’95, Coral Gables, FL,...
  • Cited by (113)

    • On estimating uncertainty of fingerprint enhancement models

      2022, Digital Image Enhancement and Reconstruction
    • A new method of micro-calcifications detection in digitized mammograms based on improved simplified PCNN

      2016, Neurocomputing
      Citation Excerpt :

      As mammography images often have a low contrast between the region of interest and the background, a proper tactic should be adopted to enhance contrast. Various algorithms for image enhancement have been widely studied, such as mathematical morphology [24], contourlet transform [8,25], histogram modification [6], Laplacian operator [26], wavelet transform [27–29] etc. Each method has its own specific limitations.

    View all citing articles on Scopus

    About the Author—CHING-TANG HSIEH is an associate professor of Electrical Engineering at Tamkang University, Taiwan, Republic of China. He received the B.S. degree in Electronics Engineering in 1976 from Tamkang University and the M.S. and Ph.D. degree in 1985 and 1988, respectively, from Tokyo Institute of Technology, Japan. From 1990 to 1996, he held the post of Chairman of the Department of Electrical Engineering. His current research interests include speech analysis and synthesis, speech recognition, natural language processing, image processing, neural networks, and fuzzy system.

    About the Author—EUGENE LAI received his B.S. degree at the Department of Electrical Engineering, National Taiwan University, Republic of China, in 1963. He received his M.S. and Ph.D. at the Department of Electrical Engineering, Iowa State University, USA, in 1969 and 1971, respectively. He is currently a professor at the Department of Electrical Engineering, Tamkang University. His major interests are in electromagnetics and semiconductor physics.

    About the Author—YOU-CHUANG WANG received the B.E. and M.E. degrees in Electrical Engineering from Tamkang University, Taiwan, Republic of China, in 1995 and 1997, respectively. He is currently pursuing the Ph.D. degree at Tamkang University. His research interests include biometrics, multimedia, and digital signal processing.

    View full text