Fingerprint Enhancement Base On Non-separable Wavelet Transform

The use of fingerprints as a biometric is both the oldest mode of computer-aided, personal identification and the best prevalent in use today. As a fingerprint verification system (FVS) grows more and more mature, it becomes the most popular and reliable biometric identification technique. Fingerprint image enhancement is a quite important step in FVS.Fingerprint enhancement is necessary for low quality images. The purpose of fingerprint enhancement is to improve the clarity of ridge and valley structures, to avoid generating pseudo-features, and to guarantee the accuracy and reliability of feature extraction. Therefore, it draws the substantial amount of attention recently. The performance of a fingerprint feature extraction and matching algorithm depends critically upon the quality of the input fingerprint image, while the'quality'of a fingerprint image cannot be objectively measured; it roughly corresponds to the clarity of the ridge structure in the fingerprint image. Where as a good quality fingerprint image has high contrast and well defined ridges and valleys, a poor quality fingerprint is marked by low contrast and ill-defined boundaries between the ridges.In this thesis, we develop a new method for fingerprint enhancement based on non-separable wavelet transform. Our method produces high contrast between ridges and valleys. We first decompose the fingerprint image using the non-separable wavelet transform, which can decompose the fingerprint image efficiently and obtain wavelet coefficients, then modifies the coefficients by apply the adaptive approach to reduce the noises and increase the contrast between ridges and valleys according to the geometry feature of images. Then we apply the inverse wavelet transform to map the result. We present our results using a two-dimensional median filter to reduce noise and preserve edges. The result generally exhibits more fidelity to the original in comparison to the alternative approaches.Compared with the traditional wavelet, our research demonstrates that the three high frequency sub-images generated by non-separable wavelet transform can extract more creases and no longer extensively focus on the three special directions. The non-separable wavelet is one of the filter banks, it features capturing the singularities in all directions. It is also holding the ability of multiresolution analysis and low computational complexity as traditional wavelet. Thus, the high-frequency sub-images of non-separable wavelet transform can reveal more desirable features in comparison with the separable wavelet transform.The proposed approach is applied to contrast enhancement, which is one of the fundamental topics in image processing, pattern recognition and computer vision. The contrast, especially in homogeneous areas, can be limited to avoid amplifying any noise that might be present in the image. Contrast enhancement methods can be categorized as either indirect or direct approaches [1], the indirect method is to modify the histogram, which is not efficient and effective, since it only stretches the global distribution of the intensity. The direct method is to define a measurement of the contrast and use it to enhance the contrast; direct methods have better performance than that of indirect methods.Our results comparison with other enhancement methods; Histogram equalization (HE), Contourlet enhancement, Gabor filters, short time Fourier transform (STFT) analysis and separable wavelet transform. Histogram equalization cannot work well, when there are high intensity distributions in the image, it misses pixels with intensity values that occur less frequently in the image. In comparison between wavelet and contourlet based contrast enhancements, two parameters must be considered: smoothness and detail enhancement. Wavelet is better, when a smooth image is desired after enhancement. When use fingerprint enhancements based on short time Fourier transform (STFT) analysis and Gabor filters, there are advantages and disadvantages of analysis merely in spatial domain or frequency domain. It is known that enhancement of fingerprint images can be performed on either binary ridge images or direct gray images. The binarization will generate more spurious minutiae structures and lose some valuable original fingerprint information. Therefore, our enhancement algorithm is performed on gray images directly. And in the results we compare between separable wavelet transform and non-separable wavelet transform, when we use 2-D separable wavelet filter banks are simply the tenor product of 1-D wavelet filter banks. Though 1-D wavelet filter banks are proved to be compact supported, its tenor product can, however, reveal the singularities in the three directions (i.e. horizontal, vertical and diagonal) only.The experimental results show that our proposed algorithm can increase the contrast between ridges and valleys, and reduce noises in the gray-scale fingerprint images. We use the fingerprint database FVC2002 to experiment on the results. The results demonstrate that the proposed method is very efficient in contrast enhancement without under-enhancement and over-enhancement, and it is superior to some other existing methods.