Method for fingerprint classification

摘要

The method for fingerprint classification uses a local gradient directional binary pattern (LGDBP) descriptor. The method acquires digital images of fingerprints from a scanner or the like, and the LGDBP descriptors corresponding to the directional ridge patterns are calculated. Using the LGDBP descriptors as a fingerprint representation, an extreme learning machine neural network with a radial basis function kernel may is used to reduce substantially the search space to a predefined number of classes of known fingerprints to be searched to identify the fingerprint.

主权项

1. A method for fingerprint classification, comprising the steps of:
scanning a fingerprint to generate a fingerprint image, FP, having a horizontal resolution of M pixels and a vertical resolution of N pixels; generating a set of responses for each (i,j)-th pixel of the fingerprint image FP, Gk(i,j), where 1≦i≦M and 1≦j≦N, by convolving the fingerprint image FP with a set of filters, Gθk, for k=1, . . . , 8, such that Gk(i,j)=FP*Gθk over k=1, . . . , 8, wherein Gθk=Gx cos θk+Gy sin θk, whereθk=(k-1)⁢π8for k=1, . . . , 8, and Gx and Gy are, respectively, horizontal and vertical Gaussian derivative filters respectively given byGx=-x2⁢πσ4⁢ⅇ-x2+y22⁢σ2⁢⁢and⁢⁢Gy=-y2⁢πσ4⁢ⅇ-x2+y22⁢σ2,where σ is a parameter representative of width of a Gaussian function;
thresholding each response Gk(i,j) such that, for each (i,j)-th pixel, if a value of the respective response Gk(i,j) is greater than zero, then a corresponding thresholded value, sk, is equal to one, and if the value of the respective response Gk(i,j) is less than or equal to zero, then the corresponding thresholded value, sk, is equal to zero; generating a local gradient directional binary pattern code by forming a binary number from the values of sk over k=1, . . . , 8 as s1 s2 s3 s4 s5 s6 s7 s8; generating a local gradient directional binary pattern code for each (i,j)-th pixel; computing a distribution of the local gradient directional binary pattern codes for all said pixels in the fingerprint image FP as a histogram and forms descriptor; and using the histogram and forms descriptor and a neural network to reduce substantially a search space to a predefined number of classes of fingerprints to be searched in order to identify the scanned fingerprint.