W. Tao and M. Xie (PRC)
Binarization, fingerprint, discrete wavelet transform, identification,
In this paper, a new binarization method for fingerprint is presented. It is mainly based on the fact that the fingerprint ridges and valleys are regions where their second derivatives are positive and negative, respectively. First, by implementing the discrete wavelet transform (DWT) for the original fingerprint image, we can get the approximation sub-image. Second, histogram equalization is applied on the approximation sub-image to obtain the equalized sub-image. Our binarization algorithm is performed on the equalized sub-image. Next, we calculate the block orientations of the equalized sub-image using its gradient information. Last, in order to get the binary image of the equalized sub-image, the second derivative at each pixel is estimated based on the intensity values of pixels along the orthogonal direction of the block orientations. The method is tested on the fingerprints of FVC2000 [16] for fingerprint identification algorithms. Experimental results show that our algorithm is efficient and robust.
Important Links:
Go Back
