Fingerprint fuzzy vault method based on (k, w) threshold secret sharing scheme

A fingerprint fuzzy vault method based on (k, w) threshold secret sharing schemeTechnical field

[0001] The present invention belongs to the technical field of pattern recognition and cryptography, and specifically relates to a technology, a threshold secret sharing scheme and an automatically aligned fingerprint fuzzy vault scheme.

Background technique

[0002] Secret sharing is a very important branch in the field of modern cryptography, and it is also an important research content in the direction of information security. In 1979, Shamir and Blakley independently proposed the concept of key decentralized management. The mechanism to realize this idea is called the (dagger-threshold scheme. This scheme is to divide a key (called a shared key) into w parts (Called w sub-keys or shadows, respectively handed over to W individuals for custody, so that the determined remediation VW) satisfies: (1) Among these W individuals, any F individual can use their sub-keys to recover the shared Key; (2) Anyone-in-one collaboration is not helpful in restoring shared keys. This idea of decentralized key management makes key management more secure and flexible, but each members sub-key has security risks. Members adopt fingerprint fuzzy vault method to protect their sub-keys.

[0003] In 2002, A. Juels and M. Sudan proposed the A fuzzy vault scheme. In their fuzzy vault method, the key of the user's unique set A mixed user is entered into the Reed-Solomon-based vault. The user can use the set B, which has most of the same elements as the set A, to recover the key.

[0004] Based on the idea of the fingerprint fuzzy vault scheme of global domain registration, the fuzzy vault scheme can be used to protect the sub-keys of each member. At this time, the security of this kind of key decentralized management is based on the difficulty of polynomial reconstruction and the fact that the user's biological characteristics are not leaked.

Summary of the invention

[0005] Under real and reliable experimental conditions, the present invention provides a set of practical fingerprint fuzzy vault methods based on Lu Song's threshold secret sharing scheme. This is a set of solutions that not only effectively protects the user's fingerprint data, but also ensures the security of the shared key.

[0006] A fingerprint fuzzy vault method based on the (dagger threshold secret sharing scheme) includes a shared key distribution stage and a shared key reconstruction stage: the shared key distribution stage also includes the binding of the fingerprint fuzzy vault and the user subkey The process and the shared key binding process; the shared key reconstruction phase also includes the release process of part of the user's sub-key and the shared key release process.

[0007] The shared key distribution stage is specifically as follows:

1. Step 1 of the binding process between the fingerprint fuzzy vault and the user's sub-key: W users input their personal registered user names and extract personal fingerprint characteristics. The plane coordinates and directions of fingerprint features are linearly mapped to [0, 255], which are represented by 8 bits respectively. X, y represent the plane coordinates of the fingerprint feature point, & represents the ridge direction of the fingerprint feature point, and * represents the type of the fingerprint feature point. Among them, the type of fingerprint feature points only uses end points and cross points. When its type is an end point, ί = 0; when its type is a cross point, ϊ = 1_οThe fingerprint characteristics of each user are expressed as: (and, nine'A._f,) (4 = 1.---¾), also ShiΛ*-Niuru) (/₃ =1,..-^）,…，

<img file="CN102946310A_D0001.tif" />

[0008] Step 2. W users construct different polynomials:

% + + --+i_llu^, mod j?

Shame)=+Wei+ --+ Know J mod/?

,

ΛΦ) = + good + ...+ such as # τηοάρ polynomial coefficients are all 16-bit random numbers, /= 1,2,...,w, = 0,1,---,8, j? = 65537 is a prime number. As-u is regarded as the sub-key of user 1, ...,% as, ... I is regarded as the sub-key of user w. k is the bit string length of the shared key £ = ^ + 1)/correction, and the gate is round up operation.

[0009] Step 3. Calculate the hash value of the polynomial coefficient bit string corresponding to each user. Each user is stored in the form of "registered user name: hash value".

[0010] Step 4. User y connects the plane coordinates% and Dan.* of each feature point of the fingerprint in series to form a 16-bit number = [such as %], and then calculate. The set of real fingerprint points obtained by the user's corpse is recorded as 6 = {e.g. $, paper, %£)%)} ο The set of real fingerprint points obtained by all users is recorded as G = g, E, ..., ®.

[0011] Step 5. Add a single tuple Js consisting of random numbers as a hash point, the heart is a 16-bit random number, the lip is an 8-bit random number, and · is a 16-bit random number, Raspberry can only randomly take values 0 and 1, right=12, record the set of hash points as CM (...£) call, Yan Xinyu v. F©)}. Mix the set &amp; and c to get the vault set. Jie and store, where Jie = ear = (lang Έ Xing Dan)}, / = =% + island +... + - + ", guess =. or Wowjin or Jiashi) ο

[0012] 2. Shared key binding process step 1. Use the shared key $ to construct a polynomial Q only. The binary string of ε is divided into blocks to form part of the coefficients of the coarse one-degree polynomial on the [M]. The remaining m-ί-1 coefficients are 16-bit random integers, where also = the constant term of the polynomial is A 16-bit check code.

[0013] Step 2. Slide rule also), fart=12... . Obtain the set W ΈΈ mixed false point set M = {, such as) m, ... use, where hong, h_ABoth are 16-bit random integers and dragon*Fu Le) ο

[0014] Step 3. Mix and scramble the collection gold and GC to obtain the collection GF and store it.

[0015] The shared key reconstruction phase is specifically as follows: When a shared key holder recovers the shared key 5, they will do the following:

1. Part of the user's sub-key release process Step 1. The shared key holder enters the fingerprint, and the plane coordinates and direction of each feature point of the extracted query fingerprint image are linearly mapped to [0.255], which is represented by 8 bits. . Query the set of characteristic points of the fingerprint 7; Guangli. Ma Yichao "Qiao=12...this}.

[0016] Step 2. Decompose the first element of the tuple in the vault set $ to get F = still = (right swollen; Yu, bow)}, weak %He;.

[0017] Step 3. Select a feature point of the query fingerprint from £ = Li, and use "Q as a reference point" to calculate the rotation angle and position offset between a point in Γ and the reference point.

Ru=% Yi Yi

[0018]] Phase NS (1) cold = | real one bow | Step 4. According to the transformation amount calculated by the formula (1), calibrate all the remaining feature points of the query fingerprint. Let the fingerprint feature point characteristics after calibration are as follows:

J = S + District Factory Airlines) cosΔί?-0" -Renpi SinΔ&amp; + song = film/ + (Declare + ~j^^) cosA^+ήρ

^.r=(4.r ^{+ A}^^{mDd360 (2)} \_^^T - Among them, 21,..., the plane coordinates of the calibrated feature points are 5L, the direction is winter, and the type is .

[0019] Step 5. Match the calibrated feature point feature set close = = (·, £, )} with the set. If it satisfies the formula (3), it is considered a matching point.

[0020]

<img file="CN102946310A_D0002.tif" />

(3) Among them, 5 α is the set threshold. According to the number of matching points, a matching number P is obtained with the 5th query fingerprint feature point and the jth vault point as a pair of reference points.

[0021] Step 6. After traversing the remaining points in J, sequentially calculate (1), (2), (3) to obtain the corresponding matching numbers J respectively. Select one of the largest number of matching objects g Ο

[0022] Step 7. Repeat steps 3, 4, 5, and 6 to compare the brews obtained each time and keep a larger number of matches. If the matching score is greater than the threshold, it means that the query fingerprint matches the registered fingerprint. At the same time, a set of matching points will be obtained YA C amma = {winterρ, Wuhukou w Ίγ,^) |&lt;P⁼1.2,...,5ο

[0023] Step &amp; Use Newton's interpolation method to reconstruct a polynomial method [still, at this time, the holder of the shared key is required to enter the user name. Calculate the hash value addition of the coefficient bit string of the polynomial factory 00, and compare it with the hash value added by the user name index. If they are equal, the polynomial reconstruction is correct, otherwise, the user is required to re-enter the fingerprint. If the user is asked If the fingerprint is re-entered more than 3 times, the user is regarded as an illegal user.

[0024] Step 9. After each shared key holder correctly reconstructs the corresponding polynomial, extract the corresponding subkey of the shared key holder from the corresponding polynomial.

[0025] 2. The shared key release process matches the corresponding set of two-tuples from the set GF, and uses Newton interpolation to reconstruct the polynomial IQ. At the same time, calculate the check code of the polynomial coefficient bit string except the constant term, and compare Whether the check code is equal to the constant term of the polynomial. If they are equal, the released shared key is correct.

[0026] This eclipse M) threshold secret sharing scheme makes key management more secure and flexible, but the subkey of each member has hidden security risks. The feature of the present invention is that while the fingerprint fuzzy vault method is used to protect the shared key, the release of the key is safely and conveniently shared through the fingerprint characteristics of the user. The release process of the key is quite simple, making the key sharing scheme more practical Sex.

Description of the drawings

[0027] FIG. 1 is a flow chart of the shared key binding process; FIG. 2 is a flow chart of the shared key release process; FIG. 3 is a partial fingerprint image in the fingerprint database under test; FIG. 4 is an extraction from the registered fingerprint image The feature point map; Figure 5 is the feature point map extracted from the query fingerprint image.

Detailed ways

[0028] The present invention will be further described below in conjunction with the accompanying drawings.

[0029] The shared key distribution stage is specifically as follows (as shown in Figure 1):

1. The binding process of the fingerprint fuzzy vault and the user's sub-key. Step 1. Each user enters his or her registered user name and fingerprint. Part of the fingerprint image in the fingerprint database for the test is shown in Figure 3. Perform a series of preprocessing operations such as segmentation of the fingerprint image, calculation of the direction field and gradient, equalization, convergence, smoothing, enhancement, binarization, thinning, etc. to obtain a clear binary image with fingerprint feature information. Then extract all the feature points in the image, filter and remove the pseudo feature points, and retain the true feature points of the original image, as shown in Figure 4.

[0030] Step 2. The plane coordinates and directions of each feature point of the fingerprint are linearly mapped to [0. 255], respectively₈ Bit representation. Work, y represents the plane coordinates of the fingerprint feature point, &amp; represents the ridge direction of the fingerprint feature point, and f represents the type of the fingerprint feature point. Among them, the type of fingerprint feature points only uses end points and cross points. When its type is an end point, ί=ο; when its type is a cross point,_f = i οThe fingerprint characteristics of each user are represented as =,

<img file="CN102946310A_D0003.tif" />

<img file="CN102946310A_D0004.tif" />

<img file="CN102946310A_D0005.tif" />

1,... few)

[0031] Step 3. W users construct different polynomials Jing, Jing still,...,/&gt;): ^(u) = ^₀ + But Β + ...+ station/ mod ρ = + + ...+ mod ρ

The coefficients of polynomials such as "are 16-bit random numbers, /= 1,, Ji = 0,1,...,8, ρ = 65537 is a prime number. ,b...u is regarded as the subkey of user 1,..., Xiaoyi... is regarded as the subkey of user w. The eagle is the bit string length of the shared key, the heart adds /16], £ = 0 + 1)/cho, and the gate is the round up operation.

[0032] Step 4. Calculate the hash value of the polynomial coefficient bit string corresponding to each user. User corpse computing

11-11Μ, Η(·) is a one-way hash function that generates a 32-bit number. Each user is stored in the form of "registered user name: hash value".

[0033] Step 5. The user's corpse connects the plane coordinate pairs of each feature point of the fingerprint in series to form a 16-bit number = [Xi 1%], and then calculates Guang Yu). The set of fingerprint real points obtained by user r is denoted as ©s%, XianluoΜ. Collect the fingerprint real points of all users and write it as G = (q,q,...,G_w）。

[0034] Step 6. Add f tuples composed of random numbers (%%·, such as) as a hash point, f is a 16-bit random number, lip is an 8-bit random number, and 6 is a 16-bit J can only randomly take values 0 and 1, Λ =12..., F ο denote the set of hash points as. ={© heart, "%) is called rejection, 6 slipper ordering station Yan Shi©"}. Mix the set G and c to get the vault set Γ and store it, where Jie=qiao=(skillful4Hearty)}, eight 1,...chuan, + M_s ",= C h or Ujr call, Shen'h or &amp; (bow, stone).

[0035] 2. Shared key binding process step 1. Use the shared key $ to construct a polynomial only feather. The binary string of $ is divided into blocks to form part of the coefficients of the coarse _1 degree polynomial on the edge, and the remaining ffi-ί-1 coefficients are 16-bit random integers, where m = £xk.

[0036] Ruler = other + such as +...+ such as _1 furnace, the constant term of the polynomial is a 16-bit check code, that is, ruler = Order (by 41 state-J ο where the shared key J II -II iV/[0037] Step 2. Calculate Μ.Λ), =. Get the set stuffing = (% swollen Zhang Ru))}. Miscellaneous false point set song={(Panqiu) bubble=12_, volume, among them, , dragon are 16-bit random integers and dragon*) ο

[0038] Step 3. Mix and scramble the collection GS and GC to obtain the collection GF and store it.

[0039] The shared key reconstruction stage is specifically as follows (as shown in Figure 2):

Once the shared key holders recover the shared key s, they will do the following:

1. The release process of part of the user's subkey step 1. The shared key holder enters the fingerprint, performs segmentation operation on the input query fingerprint image, calculates the direction field and gradient, equalizes, converges, smooths, enhances, and binarizes A series of pre-processing operations such as thinning, etc. obtain a clear binary image that retains the fingerprint feature information. Then extract all the feature points in the image, and filter and remove the false feature points. Finally, the actual feature points of the query fingerprint are extracted, as shown in Figure 5. The plane coordinates and directions of each feature point of the extracted query fingerprint image are linearly mapped to Congjixi, which are represented by 8 bits respectively. Query the characteristic points of the fingerprint

Set (heart% ) ΙΉ, ... use.

[0040] Step 2. Decomposing the first element of the tuple in the vault set Jie = = (bow, ";,, zero £. Stone", gr strict; subtle;,.

[0041] Step 3. Select a feature point of the query fingerprint from the bird and add = [Oh Bi Kuang Tao Ru] as a reference point, and calculate a point in F = £, £,;, and the rotation angle and position of the reference point Offset.

[0042]

<img file="CN102946310A_D0006.tif" />

Step 4. According to the transformation amount calculated by formula (1), calibrate all remaining feature points of the query fingerprint. Let the fingerprint feature point characteristics after calibration are as follows:

Er = + District factory heart P cut also Riyi factory EQ plus + &amp; * = blood + heart-Lingering) + -j^₄)cosA5+Af &amp;_jr =(^._Γ + Δφπιο(1360 (2)

<img file="CN102946310A_D0007.tif" />

Where τ = 1,..., must, the plane coordinates of the calibrated feature points are χ_λι ,Person, direction is, type is ".

[0043] Step 5. Set the calibrated feature point feature set as ogle=ear=(-, hit "No way)} and set

V = (r.=(x.,y.^.,f.)} for matching, if it satisfies the formula (3), it is regarded as a matching point.

[0044]

<img file="CN102946310A_D0008.tif" />

(3) Where &amp; b is the set threshold. After traversing the points in the vault and the vault, the number of matching points is obtained, that is, the first query fingerprint feature point and the nth vault point are used as a pair of reference points for a matching score ratio.

[0045] Step 6. After traversing the remaining points in P, sequentially calculate (1), (2), (3) to obtain the corresponding matching scores, respectively, called "-select one of the largest matching scores sc* ο

[0046] Step 7. Repeat steps 3, 4, 5, and 6, compare the samples obtained each time, and retain a larger matching score. If the matching score is greater than the threshold, it means that the query fingerprint matches the registered fingerprint. At the same time, according to "calling department=calling", when the 5th query fingerprint feature point and the /th vault point are used as a pair of reference points for matching, the number of query fingerprint pre-registered fingerprint matching points is the largest. According to formulas (1), (2), (3), match the query fingerprint and the registered fingerprint again, and you will get a set of matching points called =Dai Yin cold heart swollen jade#, Shi Feng I corpse 12...3

[0047] Step &amp; Use Newton's interpolation method to reconstruct the polynomial: Zhou=Heart+*&gt; +...+skillful#, at this time, the holder of the shared key is required to enter the user name. Calculate the polynomial factory (the hash value of the coefficient bit string 11-11^..) and compare it with the hash value indexed by the user name. If they are equal, the polynomial reconstruction is correct; otherwise, the user is required to re-enter the fingerprint. If the user is required to re-enter the fingerprint more than 3 times, the user is regarded as an illegal user.

[0048] Step 9. When all the shared key holders have correctly reconstructed the corresponding polynomials, extract the corresponding subkeys of the shared key holders from the corresponding polynomials.

[0049] 2. The shared key release process matches the corresponding set of two-tuples from the set, and the polynomial reconstructed by Newton's interpolation method is Q(easy Ο [0050] Gong Ge = Ji + Yu 2 + ... + Xin At the same time, it calculates the check code of the polynomial coefficient bit string except the constant term, and compares whether the check code is equal to the constant term of the polynomial. That is, if the plex=that is %skin||..., the released shared key is correct Otherwise, it prompts that the shared key cannot be released correctly.