US2005157913A1PendingUtilityA1
Fingerprint recognition system
Est. expiryMay 25, 2021(expired)· nominal 20-yr term from priority
G06V 40/1353G06V 40/1347G06V 10/993G06V 40/1359G06V 40/1365
38
PatentIndex Score
0
Cited by
0
References
0
Claims
Abstract
A method of analyzing and recognizing fingerprint images that utilizes vector processing of a vector field that is defined as the tangential vector of the fingerprint ridge curves is disclosed. The raw fingerprint image is divided into blocks, filtered to remove noise, and the orientation direction of each block is found. This allows the ridge curves to be enhanced and approximated by piece-wise linear approximations. The piece-wise linear approximations to the ridge curves allow the minutiae to be extracted and classified and a fingerprint minutiae template to be constructed. An enrollment process gathers multiple fingerprint images, creates fingerprint minutiae templates corresponding to the acquired fingerprint images, and stores the templates and other data associated with the respective individual or the enrolled fingerprint in a fingerprint database. In an identification process, an unknown raw fingerprint image is obtained via a fingerprint scanner and processed similarly to the enrollment process such that the fingerprint minutiae template of the unknown fingerprint is compared to one or more previously enrolled fingerprint minutiae templates. The identity of the individual associated with the unknown fingerprint is thereby ascertained. In addition, live finger detection can be accomplished in conjunction with the verification or identification process through analysis of the fingerprint image thus enhancing the security of the overall system.
Claims
exact text as granted — not AI-modified1 . A method for fingerprint recognition, the method comprising the steps of:
acquiring an enrolled fingerprint having a plurality of ridge curves and valleys; blocking the enrolled fingerprint to form a blocked enrolled fingerprint; separating enrolled foreground blocks from enrolled background blocks of the blocked enrolled fingerprint thereby forming an enhanced enrolled image, the step of separating comprising the steps of
calculating for each enrolled block in the blocked enrolled fingerprint the mean and variance of the pixel gray level within the block,
selecting as an enrolled foreground block each block having a variance that is less than a predetermined variance threshold and a mean that is greater than a predetermined mean threshold,
determining an enrolled convex hull defined by the centers of each enrolled block selected to be in the foreground,
testing each enrolled block not selected as an enrolled foreground block whether the center of the enrolled block is within the defined enrolled convex hull, and
in the event that the center of the enrolled block being tested is within the enrolled convex hull, selecting the enrolled block being tested as an enrolled foreground block;
determining an orientation field of the enrolled fingerprint; extracting the minutiae from the enrolled fingerprint; creating an enrolled fingerprint template of the enrolled fingerprint; and storing the enrolled fingerprint template in a database.
2 . The method of claim 1 further including:
acquiring an unknown fingerprint; determining an orientation field of the unknown fingerprint; extracting the minutiae from the unknown fingerprint; creating an unknown fingerprint template; comparing the unknown fingerprint template to the enrolled fingerprint template; determining the number of the extracted minutiae in the unknown fingerprint template that match the extracted minutiae of the enrolled fingerprint template; and if the number of extracted minutiae that match exceeds a predetermined threshold, providing indicia that the unknown fingerprint and the enrolled fingerprint are a match, otherwise indicate that the unknown fingerprint and the enrolled fingerprint are not a match.
3 . The method of claim 2 further including the steps of:
determining an enrolled block size such that the enrolled fingerprint ridge curves can be approximated by parallel straight lines; and blocking the enrolled fingerprint using the enrolled block size in forming the blocked enrolled fingerprint.
4 . The method of claim 3 wherein the step of determining the enrolled block size includes determining the enrolled block size according to the formula block_size=r*16/500, where r is the resolution of the enrolled fingerprint in dots-per-unit-length.
5 . The method of claim 2 further including the steps of:
determining an unknown fingerprint block size such that the unknown fingerprint ridge curves can be approximated by parallel straight lines; and blocking the unknown fingerprint using the unknown fingerprint block size forming a blocked unknown fingerprint.
6 . The method of claim 5 wherein the step of determining the unknown fingerprint block size includes determining the unknown fingerprint block size according to the formula block_size=r*16/500, where r is the resolution of the unknown fingerprint in dots-per-unit-length.
7 . The method of claim 2 further including the steps of:
blocking the unknown fingerprint to form a blocked unknown fingerprint; and separating the unknown foreground blocks from the unknown background blocks of the blocked unknown fingerprint forming an enhanced unknown fingerprint image.
8 . The method of claim 7 wherein the step of separating the unknown foreground blocks from the unknown background blocks of the blocked unknown fingerprint includes the steps of:
calculating for each block the mean and variance of the pixel gray level within the block; selecting as an unknown foreground block each block having a variance that is less than a predetermined variance threshold and a mean that is greater than a predetermined mean threshold; determining an unknown fingerprint convex hull defined by the centers of each block selected to be in the foreground; testing each block not selected as an unknown foreground block whether the center of the block is within the defined unknown fingerprint convex hull; and in the event that the center of the unknown block being tested is within the unknown fingerprint convex hull, selecting the block being tested as an unknown foreground block.
9 . The method of claim 2 further including filtering each of the enrolled foreground blocks.
10 . The method of claim 9 wherein the step of filtering each of the enrolled foreground blocks includes filtering each of the enrolled foreground blocks with a low pass filter.
11 . The method of claim 10 wherein the step of filtering each of the enrolled foreground blocks with a low pass filter includes filtering using a low pass Gaussian filter.
12 . The method of claim 7 further including filtering each of the unknown foreground blocks.
13 . The method of claim 12 wherein the step of filtering each of the unknown foreground blocks including filtering each of the unknown foreground blocks with a low pass filter.
14 . The method of claim 13 wherein the step of filtering each of the unknown foreground blocks with a low pass filter includes filtering using a low pass Gaussian filter.
15 . The method of claim 2 further including the step of determining for each of the enrolled foreground blocks in the enhanced enrolled image the corresponding orientation angle and amplitude forming an enrolled orientation image.
16 . The method of claim 15 wherein the step of determining the orientation angle and amplitude for each of the enrolled foreground blocks in the enhanced enrolled image includes finding the horizontal partial derivative and the vertical partial derivative for each of the enrolled foreground blocks.
17 . The method of claim 16 wherein the step of finding the horizontal partial derivative and the vertical partial derivative for each of the enrolled foreground blocks includes using a Sobel differential operator.
18 . The method of claim 16 , in the event that the orientation amplitude is less than a predetermined amplitude constant, further includes the steps of:
selecting a plurality of directions equally spaced about a unit circle; calculating the average gray level and standard deviation gray level curve projected along each selected direction of the respective enrolled foreground block; and selecting the orientation angle to be the one of the selected directions having the smallest standard deviation gray level curve.
19 . The method of claim 7 further including the step of determining for each of the unknown foreground blocks in the enhanced unknown fingerprint image the corresponding orientation angle and amplitude forming an unknown fingerprint orientation image.
20 . The method of claim 19 wherein the step of determining the orientation angle and amplitude for each of the unknown foreground blocks in the enhanced unknown fingerprint image includes finding the horizontal partial derivative and the vertical partial derivative for each of the unknown foreground blocks.
21 . The method of claim 20 wherein the step of finding the horizontal partial derivative and the vertical partial derivative for each of the unknown foreground blocks includes using a Sobel differential operator.
22 . The method of claim 20 , in the event that the orientation amplitude is less than a predetermined amplitude constant, further includes the steps of:
selecting a plurality of directions equally spaced about a unit circle; calculating the average gray level and standard deviation gray level curve projected along each selected direction of the respective unknown foreground block; and selecting the orientation angle to be the one of the selected directions having the smallest standard deviation gray level curve.
23 . The method of claim 15 further including the steps of:
creating a directional filter for filtering a respective one of the enrolled foreground blocks in the enrolled orientation image as a function of the orientation angle and amplitude of the respective enrolled foreground block, wherein the directional filter increases the contrast between ridges and valleys in the enrolled fingerprint orientation image along the same orientation direction as the respective enrolled foreground block; and applying the respective directional filter to each of the enrolled foreground blocks to be filtered forming a ridge-enhanced enrolled image.
24 . The method of step 23 wherein the step of creating the directional filter includes creating a filter mask having predetermined coefficients that are a function of the corresponding foreground block to be filtered.
25 . The method of step 24 wherein the step of creating the filter mask includes the steps of:
creating a square filter mask having a length equal to the period of the signal or the period of the signal plus one, whichever is an odd number; and determining the coefficients of the filter mask.
26 . The method of claim 25 wherein the step of determining the coefficients of the filter mask includes the steps of:
setting the center coefficient of the center row to a value a 0 ; setting the first and last coefficients of the center row to a value of a 0 /4; calculating the coefficients of the center row between the center coefficient and the first and last coefficient according to a cosine function and the difference between a 0 and a 0 /4; determining a number of middle rows on each side of the center row needed to adequately enhance the contrast between ridges and valleys in the fingerprint image, wherein the number of middle rows is an even number; determining the coefficients of the middle rows according to a cosine taper function between the center coefficient, c i , and c i /1.41; and determining the values, b i , of the top and bottom row of the filter mask as b i = ( - ∑ j = 1 , n a i , j ) * 1 2 , where b i is the i th coefficient of the first and last row of the mask, a i,j is the value of the i th coefficient of the j th row in a set comprised of the middle rows and the center row, where there are n rows in the set, and n is an odd number.
27 . The method of claim 19 further including the steps of:
creating a directional filter for filtering one of the unknown foreground blocks in the unknown fingerprint orientation image as a function of the orientation angle and amplitude of the respective unknown foreground block, wherein the directional filter increases the contrast between ridges and valleys in the unknown fingerprint orientation image along the same orientation direction as the respective unknown foreground block; and applying the respective directional filter to each of the unknown foreground blocks to be filtered forming a ridge-enhanced unknown fingerprint image.
28 . The method of step 27 wherein the step of creating the directional filter includes creating a filter mask having predetermined coefficients that are a function of the corresponding foreground block to be filtered.
29 . The method of step 28 wherein the step of creating the filter mask includes the steps of:
creating a square filter mask having a length equal to the period of the signal or the period of the signal plus one, whichever is an odd number; and determining the coefficients of the filter mask.
30 . The method of claim 29 wherein the step of determining the coefficients of the filter mask includes the steps of:
setting the center coefficient of the center row to a value a 0 ; setting the first and last coefficients of the center row to a value of a 0 /4; calculating the coefficients of the center row between the center coefficient and the first and last coefficient according to a cosine function and the difference between a 0 and a 0 /4; determining a number of middle rows on each side of the center row needed to adequately enhance the contrast between ridges and valleys in the fingerprint image, wherein the number of middle rows is an even number; determining the coefficients of the middle rows according to cosine taper function between the center row coefficient, c i , and c i /1.41; and determining the values, b i , of the top and bottom row of the filter mask as b i = ( - ∑ j = 1 , n a i , j ) * 1 2 , where b i is the i th coefficient of the first and last row of the mask, a i,j is the value of the i th coefficient of the j th row in a set comprised of the middle rows and the center row, where there are n rows in the set, and n is an odd number.
31 . The method of claim 25 further including the steps of:
determining a binarization threshold; and applying the binarization threshold to each pixel in the ridge-enhanced enrolled image forming a binary enrolled image, wherein if a pixel value in the ridge-enhanced enrolled image is less than the binarization threshold, the pixel value is set to zero, and if a pixel value in the ridge-enhanced enrolled image is greater than or equal to the binarization threshold, the pixel value is set to one.
32 . The method of claim 31 wherein the step of determining the binarization threshold includes setting the binarization threshold to one-half the maximum intensity value of the respective pixel.
33 . The method of claim 27 further including the steps of:
determining a binarization threshold; applying the binarization threshold to each pixel in the ridge-enhanced unknown fingerprint image forming a binary unknown fingerprint image, wherein if a pixel value in the ridge-enhanced unknown fingerprint image is less than the binarization threshold, the pixel value is set to zero, and if a pixel value in the ridge-enhanced unknown fingerprint image is greater than or equal to the binarization threshold, the pixel value is set to one.
34 . The method of claim 33 wherein the step of determining the binarization threshold includes setting the binarization threshold to one-half the maximum intensity value of the respective pixel.
35 . The method of claim 31 further including the step of reducing the width of a ridge curve contained within the binary enrolled image to a single pixel width forming a thinned binary enrolled image.
36 . The method of claim 33 further including the step of reducing the width of a ridge curve contained within the binary unknown fingerprint image to a single pixel width forming a thinned binary unknown fingerprint image.
37 . The method of claim 35 further including the step of approximating each ridge curve in the thinned binary enrolled image by a piecewise linear approximation forming a piecewise linear reduced binary enrolled image.
38 . The method of claim 37 wherein the step of approximating each ridge curve in the thinned binary enrolled image by a piecewise linear approximation includes:
finding the starting and ending points of a ridge curve in the thinned binary enrolled image; forming a line segment between the starting and ending points of the respective ridge curve; measuring the maximum distance between the line segment and the respective ridge curve; and if the maximum distance between the line segment and the respective ridge curve is greater than a predetermined error threshold, forming a first line sub-segment between the starting point of the respective ridge curve and the point of the respective ridge curve having the maximum distance from the line segment and forming a second line sub-segment between the starting point of the respective ridge curve having the maximum distance from the line segment and the ending point of the respective ridge curve.
39 . The method of claim 36 further including the step of approximating each ridge curve in the thinned binary unknown fingerprint image by a piecewise linear approximation forming a piecewise linear reduced binary unknown fingerprint image.
40 . The method of claim 39 wherein the step of approximating each ridge curve in the thinned binary unknown fingerprint image by a piecewise linear approximation includes:
finding the starting and ending points of a ridge curve in the thinned binary unknown fingerprint image; forming a line segment between the starting and ending points of the respective ridge curve; measuring the maximum distance between the line segment and the respective ridge curve; and if the maximum distance between the line segment and the respective ridge curve is greater than a predetermined error threshold, forming a first line sub-segment between the starting point of the respective ridge curve and the point of the respective ridge curve having the maximum distance from the line segment and forming a second line sub-segment between the starting point of the respective ridge curve having the maximum distance from the line segment and the ending point of the respective ridge curve.
41 . The method of claim 37 further including the step of extracting the minutiae from the piecewise linear reduced binary enrolled image to provide enrolled minutiae.
42 . The method of claim 41 wherein the step of extracting the minutiae includes:
calculating a crossing number corresponding to each ridge pixel contained within the piecewise linear reduced binary enrolled image; and determining the type of pixel as a function of the corresponding crossing number.
43 . The method of claim 42 wherein the step of calculating the crossing number includes calculating the crossing number according to
Cn
=
(
1
2
∑
i
=
1
…7
P
i
-
P
i
+
1
)
,
where P i and P i+1 are the pixels surrounding the respective ridge pixel P.
44 . The method of claim 42 wherein the step of determining the type of pixel includes the steps of:
if the crossing number equals 0 the pixel is an isolated point; if the crossing number equals 1 the pixel is an end point; if the crossing number equals 2 the pixel is a continuing point; if the crossing number equals 3 the pixel is a branching point; and if the crossing number equals 4 the pixel is a crossing point.
45 . The method of claim 39 further including the step of extracting the minutiae from the piecewise linear reduced binary unknown fingerprint image to provide unknown fingerprint minutiae.
46 . The method of claim 45 wherein the step of extracting the minutiae includes:
calculating a crossing number corresponding to each ridge pixel contained within the piecewise linear reduced binary unknown fingerprint image; and determining the type of pixel as a function of the corresponding crossing number.
47 . The method of claim 46 wherein the step of calculating the crossing number includes calculating the crossing number according to
Cn
=
(
1
2
∑
i
=
1
…7
P
i
-
P
i
+
1
)
,
where P i and P i+1 are the pixels surrounding the respective ridge pixel P.
48 . The method of claim 46 wherein the step of determining the type of pixel includes the steps of:
if the crossing number equals 0 the pixel is an isolated point; if the crossing number equals 1 the pixel is an end point; if the crossing number equals 2 the pixel is a continuing point; if the crossing number equals 3 the pixel is a branching point; and if the crossing number equals 4 the pixel is a crossing point.
49 . The method of claim 41 further including the step of removing false minutiae from the enrolled minutiae to form reduced enrolled minutiae.
50 . The method of claim 45 further including the step of removing false minutiae from the unknown fingerprint minutiae to form reduced unknown fingerprint minutiae.
51 . The method of claim 49 further including the step of creating an enrolled minutiae template using the reduced enrolled minutiae.
52 . The method of claim 51 wherein the step of creating an enrolled minutiae template includes creating a connected graph of the reduced enrolled minutiae.
53 . The method of claim 52 wherein the step of creating the connected graph includes the steps of:
for each of the reduced enrolled minutiae, forming an enrolled segment between the respective reduced enrolled minutiae and each of the other reduced enrolled minutiae that is within a predetermined distance.
54 . The method of claim 53 further including the steps of:
determining the intersection point between each enrolled segment and each ridge curve intersected by the respective enrolled segment; and determining the intersection angle between each enrolled segment and the tangential direction of the intersected ridge curve.
55 . The method of claim 50 further including the step of creating an unknown fingerprint minutiae template using the reduced unknown fingerprint minutiae.
56 . The method of claim 55 wherein the step of creating an unknown fingerprint minutiae template includes creating a connected graph of the reduced unknown fingerprint minutiae.
57 . The method of claim 56 wherein the step of creating the connected graph includes the steps of:
for each of the reduced unknown fingerprint minutiae, forming an unknown fingerprint segment between the respective reduced unknown fingerprint minutiae and each of the other reduced unknown fingerprint minutiae that is within a predetermined distance.
58 . The method of claim 57 further including the steps of:
determining the intersection point between each unknown segment and each ridge curve intersected by the respective unknown fingerprint segment; and determining the intersection angle between each unknown fingerprint segment and the tangential direction of the intersected ridge curve.
59 . The method of claim 2 wherein the step of comparing the unknown fingerprint template to the enrolled fingerprint template includes the steps of:
a) finding a matching pair of nodes in the enrolled fingerprint template and the unknown fingerprint template; b) determining a template transformation to translate and rotate the unknown fingerprint template to align the unknown and enrolled fingerprint templates; c) using the template transformation, transforming an unknown fingerprint minutiae in the neighborhood of the matching node pairs in the unknown fingerprint template to the enrolled fingerprint fingerprint template; d) computing the difference between the transformed unknown fingerprint minutiae and the enrolled minutiae; e) if the difference between the transformed unknown fingerprint minutiae and an enrolled minutiae is less than a predetermined threshold, count the transformed unknown fingerprint minutiae and the enrolled minutiae as matched; and f) in the event that there is more than one unknown fingerprint minutiae in the neighborhood of the matching node pair in the unknown fingerprint minutiae template and there is more than one enrolled minutiae in the neighborhood of the matching node pair in the enrolled minutiae template, repeat the step of computing the difference and comparing the difference to the predetermined threshold for each of the unknown fingerprint minutiae.
60 . The method of claim 59 , further including the steps of:
in the event that more than one matching node pair is found, repeating the steps a-f for each matching node pair; and selecting the matching node pair having the greatest number of matched unknown and enrolled minutiae.
61 . The method of claim 33 further including the step of detecting a living finger.
62 . The method of claim 61 wherein the step of detecting a living finger includes detecting the characteristic of a sweat pore contained within the binary unknown fingerprint image.
63 . The method of claim 62 wherein the step of detecting the characteristic of a sweat pore includes:
forming a chain code of the boundaries in the binary unknown fingerprint image; finding all clockwise closed chains; measuring the closed chains; and if the size of a closed chain exceeds a predetermined sweat pore threshold the closed chain is identified as a sweat pore in a living finger.
64 . The method of claim 2 wherein, in the event that the unknown fingerprint and the enrolled fingerprint are a match, providing access to a secured entity.
65 . The method of claim 64 wherein the secured entity is a computer.
66 . The method of claim 64 wherein the secured entity is a computer network.
67 . The method of claim 64 wherein the secured entity is data contained in a smartcard.
68 . The method of claim 64 wherein the secured entity is a cryptographic key.
69 . The method of claim 15 further including the steps of:
dividing each of the selected foreground blocks into a plurality sub-blocks; creating a core mask; convolving each of the sub-blocks of the selected foreground blocks with the core mask; normalizing the results of the convolution of each of the sub-blocks of the selected foreground blocks with the core mask; estimating the curvature in each sub-block as proportional to the convolution of the respective sub-block; determining Poincare indices of sub-blocks having a curvature that is greater than a predetermined curvature threshold; grouping the sub-blocks having a curvature that is greater than a predetermined curvature threshold according to the corresponding Poincare index; identifying the sub-blocks having a curvature that is greater than a predetermined curvature threshold as cores and deltas according to the corresponding Poincare index; if the estimate of the curvature of a sub-block exceeds a predetermined curvature threshold, surrounding the respective sub-block with a closed curve and calculate the direction integration of the closed curve; and if the calculated direction integration is substantially zero then reducing the diameter of the closed curve and recalculating the direction integration and continue to reduce the diameter of the closed curve until the value of the direction integration is non-zero.Cited by (0)
No later patents cite this yet.
References (0)
No backward citations on record.