WO2015101197A1 - 一种基于模糊特征点信息的指纹模板及指纹识别方法 - Google Patents

一种基于模糊特征点信息的指纹模板及指纹识别方法 Download PDF

Info

Publication number
WO2015101197A1
WO2015101197A1 PCT/CN2014/094781 CN2014094781W WO2015101197A1 WO 2015101197 A1 WO2015101197 A1 WO 2015101197A1 CN 2014094781 W CN2014094781 W CN 2014094781W WO 2015101197 A1 WO2015101197 A1 WO 2015101197A1
Authority
WO
WIPO (PCT)
Prior art keywords
fingerprint
feature point
point information
fuzzy feature
fuzzy
Prior art date
Application number
PCT/CN2014/094781
Other languages
English (en)
French (fr)
Inventor
石丰
刘中秋
李健
Original Assignee
石丰
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 石丰 filed Critical 石丰
Priority to US14/902,288 priority Critical patent/US9886619B2/en
Publication of WO2015101197A1 publication Critical patent/WO2015101197A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/12—Fingerprints or palmprints
    • G06V40/1365—Matching; Classification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/12—Fingerprints or palmprints
    • G06V40/1365—Matching; Classification
    • G06V40/1371—Matching features related to minutiae or pores
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/12—Fingerprints or palmprints
    • G06V40/1347—Preprocessing; Feature extraction

Definitions

  • the invention relates to the field of fingerprint recognition, in particular to a fingerprint template and a fingerprint identification method based on fuzzy feature point information.
  • Fingerprint recognition is to identify a person's real-life identity by comparing a person's on-site fingerprint with his pre-saved fingerprint template.
  • the so-called fingerprint template records the complete information of the fingerprint feature points, including the coordinates, attributes, angles, and even the ridge width of the feature points.
  • a typical fingerprint template is established as follows: a human fingerprint image is collected by a fingerprint reading device, and the collected original image is processed to make it clearer. Then extract the fingerprint feature points (the feature points are generally the endpoints of the fingerprint, the fork point, the center point, etc.), and obtain the data description (coordinates, attributes, angles, etc., feature point information) in the geometric sense of the feature points, and finally save the fingerprint. template. After the scene fingerprint image is collected, it will also process the feature points and acquire the feature point information. Furthermore, the fingerprint identification process is completed by comparing the on-site fingerprint feature point information and the fingerprint template feature point information. Therefore, the fingerprint template is a core data structure in the fingerprint identification method.
  • each fingerprint identification method is based on the fingerprint template of the complete feature point information, that is, the fingerprint template records the complete feature value information of the feature point, including the coordinates and attributes of the feature point (ie, the endpoint, the fork point, the center point described above) Other data such as type), angle, and even ridge width.
  • the fingerprint template of the complete feature point information records the complete feature value information of the feature point, including the coordinates and attributes of the feature point (ie, the endpoint, the fork point, the center point described above) Other data such as type), angle, and even ridge width.
  • FIG. 1 A typical schematic diagram of the existing fingerprint feature point information, as shown in FIG. 1 , wherein the length of each data can be set according to actual conditions, for example, the data length of the angle is 8 bits in FIG. 1 , in actual situations Can be set to 9bits or 7bits, here is just an example.
  • the existing fingerprint template has a defect that cannot be overcome by itself: since the fingerprint template records the value of the complete feature point information of the feature point, once the fingerprint template is leaked, it may be used by a malicious person, according to The information of the fingerprint template restores the fingerprint image or creates a fingerprint image containing the same feature points. This poses a danger to the safety and privacy of users. There are such problems in various fingerprint identification methods currently applied.
  • the fingerprint template cannot be updated because the template generated by this method is true.
  • the point is constant, the hash point is randomly generated, when the attacker gets more than two templates, very It is easy to get the real point;
  • a large number of hash points (generally more than 10 times of the real fingerprint feature points) need to be added in the fingerprint template, resulting in an increase in the amount of data, and the algorithm difficulty of the comparison process is correspondingly increased, and the fingerprint template is There is still complete feature point information, which cannot completely overcome the defect that feature point information is reduced to fingerprint.
  • the present invention provides a fingerprint template based on fuzzy feature point information for the above-mentioned deficiencies of the prior art, and the present invention is implemented by the following technical solutions:
  • a fingerprint template based on fuzzy feature point information comprises: a plurality of fuzzy feature point information established according to partial feature point information of a template fingerprint, and the plurality of fuzzy feature point information is insufficient to restore the template fingerprint;
  • the values, C 0 to C n are coefficients having a predetermined rule.
  • the predetermined rule includes: C 0 to C n include a first part and a second part, the value of the first part is randomly generated, and the value of the second part is a value obtained by calculating the value of the first part according to a predetermined rule.
  • the fingerprint template is incomplete feature point information, and only the coordinate information and attribute information of the feature point are recorded, and the value of a polynomial is recorded. This information is not enough to restore the fingerprint image, nor is it enough to be used for fingerprint comparison alone, thus protecting the privacy and security of the user.
  • the invention further provides a fingerprint identification method based on fuzzy feature point information, which is realized by the following technical solutions:
  • a fingerprint recognition method based on fuzzy feature point information comprises the steps of: S1: acquiring a fingerprint template based on fuzzy feature point information and a fingerprint on the spot; S4, comparing the on-site fingerprint with a fingerprint template based on the fuzzy feature point information;
  • the fingerprint template based on the fuzzy feature point information includes: a plurality of fuzzy feature point information established according to part of the feature point information of a template fingerprint, and the plurality of fuzzy feature point information is insufficient to restore the template fingerprint;
  • Step S4 includes: substituting the corresponding feature point information of the on-the-spot fingerprint into a polynomial, calculating the values of C 0 to C n and determining whether the calculated C 0 to C n satisfy the predetermined rule, and if the judgment result is yes, the on-site fingerprint and the blur based The fingerprint template of the feature point information is matched. If the judgment result is no, the live fingerprint does not match the fingerprint template based on the fuzzy feature point information.
  • the step S1 and the step S4 further comprise: S2, performing fingerprint registration on the fingerprint of the site and the fingerprint template based on the fuzzy feature point information;
  • Step S2 includes the steps of:
  • the geometric data includes: a maximum side length, a second largest side length, a minimum side length, a maximum angle, a sub-large angle, a minimum angle, a maximum angular vertex abscissa, a maximum angular vertex ordinate, a maximum angular vertex to a sub-large angle vertex.
  • the vector the vector from the largest angular vertex to the smallest angular vertex, the attribute of the largest angular vertex, the attribute of the next largest vertex, and the attribute of the smallest angular vertex.
  • the calculation of the on-site fingerprint of step e is compared to the rotation angle and the translation distance of the fingerprint template based on the fuzzy feature point information, and the rotation and/or translation of the on-site fingerprint according to the calculation result includes: the maximum angular vertex of the live fingerprint For the reference point, the rotation angle and the translation distance of the matched non-isosceles triangle are calculated respectively, and the calculated highest frequency value is used as the rotation coefficient and the translation coefficient of the field fingerprint for rotation and translation.
  • the matching succeed in step c includes the following conditions:
  • the difference between the length of the three pairs of edges corresponding to the fingerprint template of the fingerprint based on the fuzzy feature point information is within the second threshold
  • the angle difference between the three-diagonal angle corresponding to the fingerprint of the scene and the fingerprint template based on the fuzzy feature point information is within the third threshold.
  • the step further includes a step between step S2 and step S4:
  • step S3 Calculating the number of feature points of the abscissa, the ordinate and the attribute of the on-site fingerprint after the rotation and translation, and the registration information of the fingerprint template based on the fuzzy feature point information, if the calculation result is not greater than the fourth threshold, the scene The fingerprint does not match the fingerprint template based on the fuzzy feature point information. If the calculation result is greater than the fourth threshold, step S4 is performed.
  • C 0 to C n include a first portion and a second portion.
  • the value of the first portion is randomly generated, and the value of the second portion is a value obtained by calculating the value of the first portion according to a predetermined rule.
  • the fingerprint template is incomplete feature point information, only the coordinate information and attribute information of the feature point, and the value of a polynomial are recorded. This information is not enough to restore the fingerprint image, nor is it enough to use the fingerprint comparison alone.
  • the feature point information of the on-site fingerprint into the polynomial solution coefficient and judging whether the coefficient satisfies the predetermined rule, it is judged whether the on-site fingerprint matches the template fingerprint. . Therefore, the privacy and security of the user are guaranteed.
  • FIG. 1 is a schematic structural diagram of an existing fingerprint feature point information
  • FIG. 3 is a schematic diagram showing a fingerprint template structure based on fuzzy feature point information and a data length thereof according to an embodiment of the present invention
  • FIG. 4 is a schematic diagram of a non-isosceles triangle established by the present invention according to three feature point information in a fingerprint template based on fuzzy feature point information;
  • FIG. 5 is a schematic diagram of a non-isosceles triangle established by the present invention according to three feature point information of a scene fingerprint
  • Figure 6 is a schematic diagram showing geometric data of a schematic diagram of a non-isosceles triangle in the present invention.
  • FIG. 7 shows a fingerprint recognition method based on fuzzy feature point information in an embodiment of the present invention.
  • the fingerprint template based on the fuzzy feature point information provided by the present invention comprises a plurality of fuzzy feature point information components, and the structure thereof is as shown in FIG. 2, and includes registration information and comparison information.
  • the registration information is the value of the partial feature information of the feature points of the fingerprint, which is insufficient to restore the fingerprint image, and is not sufficient for the comparison, but can be used for registration.
  • the comparison information is the value of the other part of the feature point information of the feature point information except the registration information, and the value of one of the feature point information may be the value of the plurality of feature point information, which is blurred when the template is formed. Processing, the original features cannot be restored, and are used for comparison in the present invention.
  • Registration information including the abscissa x, ordinate y, and attributes of the feature points. Attributes refer to the type of feature points, that is, endpoints, fork points, center points, and so on.
  • the alignment information contains a polynomial value P(X i ).
  • n is the order of the polynomial.
  • n generally takes an integer of 8 or greater, which means that there are at least 9 or more quantized feature points between the fingerprint template and the live fingerprint, which is not limited by the present invention.
  • C 0 ⁇ C n there are n+1 coefficients in total, then in the subsequent matching process, at least n+1 matching points need to be found in the live fingerprint, and substituted into polynomial to solve C 0 ⁇ C n .
  • X i is a value quantized by one or more feature point information other than the registration information in the feature point information.
  • the values of C 0 to C n can be obtained, and the obtained rules of C 0 to C n can be determined by judging Whether it is consistent with the rules of C 0 to C n when calculating the polynomial in the previous fingerprint template. If they match, the on-site fingerprint matches the fingerprint template, and vice versa.
  • Step 1 A preferred fingerprint template based on fuzzy feature point information is established.
  • the fingerprint reading device collects the template fingerprint, and processes the collected template fingerprint, including image refinement, enhancement, denoising, etc., and then extracts part of the fingerprint feature points (endpoint, cross point, center point, etc.) to obtain
  • the feature point information includes coordinate, attribute and angle information of the feature point, and is established into a plurality of fuzzy feature point information to form a fingerprint template. Since the fingerprint feature points of the extracted part are created into a plurality of fuzzy feature point information, and the fuzzy feature point information is not complete feature point information, it is not enough to be reduced to the template fingerprint.
  • the fingerprint template includes at least 27 fuzzy feature point information, and the data structure of each fuzzy feature point information includes an abscissa of 8 bits, an ordinate y of 8 bits, a feature point attribute of 1 bits, and a polynomial of 16 bits.
  • the value P(X i ) C 0 +C 1 X i +C 2 X i 2 +. «+C 8 X i 8 .
  • the purpose of quantification is to tolerate certain measurement errors.
  • C 0 to C 3 are 16-bit random numbers, and C 0 to C 3 are hashed to generate a 128-bit sequence, and the first 80 bits of the sequence are divided into 5 groups, each group of 16 bits, which can sequentially obtain C 4 ⁇ C 8 .
  • the coefficients C 0 to C 8 do not enter the fingerprint template, but the values of the polynomials are stored in the fingerprint template, and C 0 to C 8 have established certain rules with each other.
  • the fingerprint template created by the above steps only records the fuzzy feature point information, and due to the randomly generated relationship of C 0 to C 3 , even if the same fingerprint is acquired twice, the generated fingerprint template is different. Therefore, it is more effective to protect the privacy and security of users.
  • Step 2 Before the feature point information of the on-site fingerprint is substituted into the polynomial for comparison, the registration process of the on-site fingerprint and the fingerprint template is also set in this embodiment.
  • a first non-isosenic triangle as shown in FIG. 4 (ie, other triangles not including an isosceles triangle and an equilateral triangle) may be constructed.
  • a second non-isosceles triangle as shown in FIG. 5 can be constructed, and the geometric data of the two non-isosceles triangles are respectively recorded, as shown in FIG. 6.
  • the geometric data of the first non-isosceles triangle includes: maximum side length l1, second largest side length l2, minimum side length l3, maximum angle 1, second largest angle 2, minimum angle 3, maximum angular vertex abscissa X p1 , maximum angle
  • the vertex ordinate Y p1 the vector from the maximum angular apex to the sub-large apex, the vector from the largest angular apex to the minimum apex, the attribute of the largest angular apex p1, the attribute of the next largest apex p2, and the attribute of the smallest angular vertice p3.
  • the geometric data of the second non-isosceles triangle includes: maximum side length l1', second largest side length l2', minimum side length l3', maximum angle 1', second largest angle 2', minimum angle 3', maximum angle apex horizontal Coordinate X p1 ⁇ , maximum angular vertex ordinate Y p1 ⁇ , vector of maximum angular apex to sub-large apex, vector of maximum angular apex to minimum apex, attribute of maximum angular apex p1 ⁇ , attribute of sub-large apex p2', The property of the smallest angular vertex p3 ⁇ .
  • the triangle constructed by the fingerprint template is matched with the triangle constructed by the on-site fingerprint. If two triangles satisfy the following conditions at the same time, the two triangles are considered to match:
  • the length of the three sides is within 3 (better threshold, the invention is not limited)
  • the three angle difference is within 3 (better threshold, the invention is not limited)
  • the translation distance includes:
  • the rotation angle and translation distance between the 40 pairs of matching triangles are counted, and the highest frequency is taken as the rotation coefficient and translation coefficient of the scene fingerprint, and the scene fingerprint is rotated and translated.
  • Step 3 On the basis of step 2, the number of feature points matching between the on-site fingerprint and the template fingerprint may be calculated.
  • step 4 If the number of matching points is greater than 9 (better threshold, the invention is not limited), proceed to step 4.
  • the number of matching points is not more than 9 (preferably the threshold, the invention is not limited), it is considered that the two fingerprints do not match.
  • the 8bits angle feature value of the feature point information of the matched field fingerprint is quantized into a 6bits angle feature value, and substituted into a polynomial, which is equal to the polynomial value of the fingerprint template. At least the following equations can be combined to solve the polynomial coefficients:
  • C 0 to C 8 obtained from the feature information of the field fingerprint can be obtained.
  • the C 0 to C 3 obtained from the feature information of the on-site fingerprint are hashed to generate a 128-bit sequence, and the first 80 bits of the sequence are divided into 5 groups, each group of 16 bits, and the corresponding C 4 is obtained in turn.
  • C 8 The C 4 to C 8 obtained in this way are consistent with the C 4 to C 8 obtained by solving the equations before, and the fingerprint of the scene is matched with the template fingerprint, and the two fingerprints are compared, otherwise the two fingerprints do not match.
  • the flowcharts of the above steps 2 to 4 can be simultaneously referred to FIG. 7.
  • the main principle of the present invention is to verify whether the relationship between the solved coefficients is consistent with a predetermined setting.
  • the invention has the advantages that the fingerprint template does not store complete fingerprint feature point information, only the coordinate information and attribute information of the feature point, and the value of a polynomial are recorded. This information is not enough to restore the fingerprint image, nor is it enough to be used for fingerprint comparison alone, thus protecting the privacy and security of the user.
  • the traditional fuzzy algorithm performs polynomial mapping, the comparison information appears on the template, which blurs the alignment information by adding more than 10 times of the hash points.
  • the alignment information does not appear in the template, and therefore no hash points are used.
  • the fingerprint template is incomplete feature point information, and only the coordinate information and attribute information of the feature point and the value of a polynomial are recorded. This information is not enough to restore the fingerprint image, nor is it enough to use the fingerprint comparison alone.
  • the feature point information of the on-site fingerprint into the polynomial solution coefficient and judging whether the coefficient satisfies the predetermined rule, it is judged whether the on-site fingerprint matches the template fingerprint. . Therefore, the privacy and security of the user are guaranteed.

Landscapes

  • Engineering & Computer Science (AREA)
  • Human Computer Interaction (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Collating Specific Patterns (AREA)
  • Image Analysis (AREA)

Abstract

公开了一种基于模糊特征点信息的指纹模板及指纹识别方法，指纹模板为若干不完整的特征点信息，包括配准信息和比对信息，其中仅记录了特征点的坐标信息和属性信息，以及一个多项式的值。多项式P(X i)=C0+C1Xi+ C2Xi^2…CnXi^n中，n为多项式的阶数，Xi为特征点信息中除配准信息以外的一个或多个特征点信息所量化得到的数值，C0~Cn为具有一定预定规则的系数。通过将现场指纹的对应特征点信息代入多项式求得C0~Cn，判断求得的C0~Cn是否符合预定规则，从而判断现场指纹是否与模板指纹匹配。

Description

一种基于模糊特征点信息的指纹模板及指纹识别方法 技术领域
本发明涉及指纹识别领域，特别涉及一种基于模糊特征点信息的指纹模板及指纹识别方法。
背景技术
指纹识别，是通过比较一个人的现场指纹和他预先保存的指纹模板，以验证其真实身份的识别方式。
所谓的指纹模板，记录了指纹特征点的完整信息，其包括特征点的坐标、属性、角度，甚至脊线宽度等信息。一个典型的指纹模板建立过程如下：通过指纹读取设备采集人体指纹图像，对采集到的原始图像进行处理，使之更加清晰。然后提取指纹特征点(特征点一般为指纹的端点、叉点、中心点等)，获取特征点的几何意义上的数据描述(坐标、属性、角度等，即特征点信息)，最后保存形成指纹模板。现场指纹图像在被采集以后，也会通过处理，提取特征点，获取特征点信息。进而，通过对现场指纹特征点信息和指纹模板特征点信息进行比较，从而完成指纹识别过程。因此，指纹模板，是指纹识别方法里的一个核心数据结构。
目前，各指纹识别方法的建立都基于完整特征点信息的指纹模板，即，指纹模板记录了特征点的完整特征值信息，包括特征点的坐标、属性(即上述的端点、叉点、中心点等类型)、角度，甚至脊线宽度等其他数据。一个典型的现有的指纹特征点信息的结构示意图，如图1所示，其中的各项数据的长度可根据实际情况进行设置，比如角度的数据长度在图1中为8bits，在实际情况中，可以设置为9bits或者7bits，这里只是举例说明。现有的这种指纹模板，存在着一个其自身无法克服的缺陷：由于指纹模板里记录了特征点的完整特征点信息的值，那么一旦指纹模板泄露，就可能会被恶意之人利用，根据指纹模板的信息还原出指纹图像来，或者制作出含有相同特征点的指纹图像。这样对用户的安全和隐私都构成了危险。当前应用的各种指纹识别方法都存在着这样的问题。
由此，诞生了一种在指纹模板中加入杂凑点的方法来对指纹模版中的信息进行保护，然而这种做法的缺陷在于：首先，指纹模板不可以更新，因为由此方法产生的模板真实点是不变的，杂凑点随机产生，当攻击者取得两个以上模板，很 容易得到真实点；第二，指纹模板中需要加入大量的杂凑点(一般为真实的指纹特征点的10倍以上)，造成数据量增加，对比过程的算法难度也会相应增加，而且指纹模板中仍然有完整的特征点信息，不能完全克服特征点信息被还原成指纹的缺陷。
发明内容
本发明针对现有技术存在的上述不足，提供了一种基于模糊特征点信息的指纹模板，本发明通过以下技术方案实现：
一种基于模糊特征点信息的指纹模板，基于模糊特征点信息的指纹模板包括：根据一模板指纹的部分特征点信息建立的若干模糊特征点信息，若干模糊特征点信息不足以还原模板指纹；
模糊特征点信息包括配准信息和比对信息，配准信息包括模糊特征点信息的横坐标、纵坐标和属性，比对信息包括一多项式P(Xi)＝C0+C1Xi+C2Xi 2+…….+CnXi n的值，n为多项式的阶数，Xi为模糊特征点信息中除配准信息以外的一个或多个特征点信息所量化得到的数值，C0～Cn为具有一预定规则的系数。
较佳的，预定规则包括：C0～Cn包括第一部分和第二部分，第一部分的值为随机产生，第二部分的值为将第一部分的值按照预定规则进行计算而获取的值。
通过本发明提供的基于模糊特征点信息的指纹模板建立方法，指纹模版为不完整的特征点信息，仅记录了特征点的坐标信息和属性信息，以及一个多项式的值。这些信息不足以用来还原指纹图像，也不足以单独用来进行指纹比对，因此，保障了用户的隐私和安全。
本发明另提供一种基于模糊特征点信息的指纹识别方法，通过以下技术方案实现：
一种基于模糊特征点信息的指纹识别方法，包括步骤：S1、获取一基于模糊特征点信息的指纹模板以及现场指纹；S4、将现场指纹与基于模糊特征点信息的指纹模板进行比对；
基于模糊特征点信息的指纹模板包括：根据一模板指纹的部分特征点信息建立的若干模糊特征点信息，若干模糊特征点信息不足以还原模板指纹；
模糊特征点信息包括配准信息和比对信息，配准信息包括模糊特征点信息的横坐标、纵坐标和属性，比对信息包括一多项式P(Xi)＝C0+C1Xi+C2Xi 2+…….+CnXi n的值，n为多项式的阶数，Xi为模糊特征点信息中除配准信息以外的一个或多个特征点信息所量化得到的数值，C0～Cn为具有 一预定规则的系数；
步骤S4包括：将现场指纹的对应特征点信息代入多项式，计算C0～Cn的值并判断计算得到的C0～Cn是否满足预定规则，若判断结果为是，则现场指纹与基于模糊特征点信息的指纹模板匹配，若判断结果为否，则现场指纹与基于模糊特征点信息的指纹模板不匹配。
较佳的，步骤S1与步骤S4之间还包括：S2、将现场指纹与基于模糊特征点信息的指纹模板进行指纹配准；
步骤S2包括步骤：
a、选取三个基于模糊特征点信息的指纹模板的模糊特征点信息的横坐标、纵坐标作为顶点，构建一个第一非等腰三角形，并记录第一非等腰三角形的几何数据；
b、选取三个现场指纹的特征点信息的横坐标、纵坐标作为顶点，构建第二非等腰三角形，并记录第二非等腰三角形的几何数据；
c、将第一非等腰三角形与第二非等腰三角形进行匹配；
d、重复执行步骤a至步骤c，若可获取大于第一阈值对匹配的非等腰三角形，则执行e；若无法获取大于第一阈值对匹配的非等腰三角形，则现场指纹与基于模糊特征点信息的指纹模板不匹配；
e、计算现场指纹相较于基于模糊特征点信息的指纹模板的旋转角度和平移距离，并根据计算结果对现场指纹进行旋转和/或平移，以与基于模糊特征点信息的指纹模板配准。
较佳的，几何数据包括：最大边长、次大边长、最小边长、最大角度、次大角度、最小角度、最大角顶点横坐标、最大角顶点纵坐标、最大角顶点到次大角顶点的向量、最大角顶点到最小角顶点的向量、最大角顶点的属性、次大角顶点的属性以及最小角顶点的属性。
较佳的，步骤e的计算现场指纹相较于基于模糊特征点信息的指纹模板的旋转角度和平移距离，并根据计算结果对现场指纹进行旋转和/或平移包括：以现场指纹的最大角顶点为基准点，分别计算第一阈值对匹配的非等腰三角形的旋转角度和平移距离，以计算得到的频率最高的数值作为现场指纹的旋转系数和平移系数进行旋转和平移。
较佳的，步骤c中匹配成功包括以下条件：
现场指纹与基于模糊特征点信息的指纹模板对应的三对顶点的属性一致；
现场指纹与基于模糊特征点信息的指纹模板对应的三对边的长度差值在第二阈值内；
现场指纹与基于模糊特征点信息的指纹模板对应的三对角的角度差值在第三阈值内。
较佳的，在步骤S2与步骤S4之间还包括步骤：
S3、计算经过旋转和平移后的现场指纹的横坐标、纵坐标以及属性与基于模糊特征点信息的指纹模板的配准信息的特征点匹配个数，若计算结果不大于第四阈值，则现场指纹与基于模糊特征点信息的指纹模板不匹配，若计算结果大于第四阈值，则执行步骤S4。
较佳的，C0～Cn包括第一部分和第二部分，第一部分的值为随机产生，第二部分的值为将第一部分的值按照预定规则进行计算而获取的值。
较佳的，步骤S4中的将现场指纹的特征点信息代入多项式，计算C0～Cn的值并判断计算得到的C0～Cn是否满足预定规则包括：
将现场指纹的特征点信息代入多项式，计算C0～Cn的值，获取C0～Cn中第一部分的值，将第一部分的值按照预定规则进行计算得到第二部分的值，判断计算得到的第二部分的值与模板指纹中的第二部分的值是否相同。
通过本发明提供的基于模糊特征点信息的指纹识别方法，指纹模版为不完整的特征点信息，仅记录了特征点的坐标信息和属性信息，以及一个多项式的值。这些信息不足以用来还原指纹图像，也不足以单独用来进行指纹比对，通过将现场指纹的特征点信息代入多项式求解系数，判断系数是否满足预定的规律来判断现场指纹是否与模版指纹匹配。因此，保障了用户的隐私和安全。
附图说明
图1所示的是一现有的指纹特征点信息的结构示意图；
图2所是的是本发明的模糊特征点信息的结构示意图；
图3所示的是本发明一实施例中的基于模糊特征点信息的指纹模板结构及其数据长度示意图；
图4所示的是本发明根据基于模糊特征点信息的指纹模板中的三个特征点信息所建立的一非等腰三角形示意图；
图5所示的是本发明根据现场指纹的三个特征点信息所建立的一非等腰三角形示意图；
图6所示的是本发明中非等腰三角形示意图的几何数据示意图；
图7所示的是本发明一实施例中的基于模糊特征点信息的指纹识别方法。
具体实施方式
以下将结合本发明的附图，对本发明实施例中的技术方案进行清楚、完整的描述和讨论，显然，这里所描述的仅仅是本发明的一部分实例，并不是全部的实例，基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动的前提下所获得的所有其他实施例，都属于本发明的保护范围。
为了便于对本发明实施例的理解，下面将结合附图以具体实施例为例作进一步的解释说明，且本实施例的各步骤不构成对本发明实施例的限定。
本发明提供的基于模糊特征点信息的指纹模板包括若干模糊特征点信息组成，其结构如图2所示，包括配准信息和比对信息。配准信息为指纹的特征点的部分特征信息的值，这些信息不足以还原指纹图像，也不足以单独用来进行比对，但可以用来配准。比对信息为除配准信息外的特征点信息的其它部分特征点信息的值，可以是其中的一个特征点信息的值也可以是多个特征点信息的值，其在形成模板时经过模糊处理，无法恢复原来的特征，在本发明中用来比对。
配准信息，包含特征点的横坐标x、纵坐标y和属性。属性是指特征点的类型，即端点、叉点、中心点等。比对信息，包含一个多项式的值P(Xi)。所述配准信息外的一个或多个特征值量化为一个数值Xi，基于Xi构造多项式P(Xi)＝C0+C1Xi+C2Xi 2+…….+CnXi n。系数C0～Cn相互之间存在着一种预定的规则。在求得多项式的值以后，将P(Xi)的值代入指纹模板，系数C0～Cn并不进入指纹模板。n为多项式的阶数。较佳的，n一般取8或大于8的整数,它意味着在指纹模板和现场指纹之间至少存在9个以上量化相等的特征点，本发明不做限制。C0～Cn，总共有n+1个系数，那么在后续匹配过程中，需至少在现场指纹中找出n+1个匹配点，将其代入多项式方能解出C0～Cn。Xi为特征点信息中除配准信息以外的一个或多个特征点信息所量化得到的数值。
在指纹的比对过程中，只要将采集到的现场指纹中的相关特征点信息代入多项式中，即可求得C0～Cn的值，通过判断所求得的C0～Cn的规则是否与先前的指纹模板中计算多项式时的C0～Cn的规则一致，若一致则现场指纹与指纹模板匹配，反之则不匹配。
为了便于技术人员理解，以下提供一具体的实施例：
步骤一、一个较佳的基于模糊特征点信息的指纹模板的建立。
通过指纹读取设备采集模板指纹，对采集到的模板指纹进行处理，包括图像的细化、增强、去噪等手段，然后提取部分的指纹特征点(端点、叉点、中心点等)，获取特征点信息，包括特征点的坐标、属性和角度信息，建立成若干模糊特征点信息以构成指纹模板。由于是提取部分的指纹特征点建立成若干模糊特征点信息，且模糊特征点信息并不是完整的特征点信息，因此不足以还原成模板指纹。
请参考图3，这里指纹模板包含至少27个模糊特征点信息，每个模糊特征点信息的数据结构包括8bits的横坐标x、8bits的纵坐标y、1bits的特征点属性、16bits的一个多项式的值P(Xi)＝C0+C1Xi+C2Xi 2+…….+C8Xi 8。其中，Xi采用特征点信息中的角度特征值，将其8bits角度特征值量化为6bits角度特征值，且取n＝8。量化的目的是为了容忍一定的测量误差。
C0～C3均为16位的随机数，将C0～C3通过hash运算产生128位的数列，取数列前80位，分成5组，每组16位，即可依次获得C4～C8。系数C0～C8不进入指纹模板，但指纹模板中保存了多项式的值，且C0～C8相互之间已经建立了一定的规则。
通过上述步骤建立的指纹模板，只记载了模糊的特征点信息，并且由于C0～C3随机产生的关系，即便同一枚指纹的两次采集，生成的指纹模板也不一样。因此能更有效的保护用户的隐私和安全。
步骤二、在将现场指纹的特征点信息代入多项式进行比对之前，本实施例还设置了一次现场指纹与指纹模板的配准过程。
根据上述的指纹模板中的三个模糊特征点信息中的横坐标和纵坐标，可构建一个如图4所示的第一非等腰三角形(即不包括等腰三角形和等边三角形的其他三角形)，并根据现场指纹的三个特征点信息，可构建一个如图5所示的第二非等腰三角形，分别记录这两个非等腰三角形的几何数据，如图6所示。
第一非等腰三角形的几何数据包括：最大边长l1、次大边长l2、最小边长l3、最大角度1、次大角度2、最小角度3、最大角顶点横坐标Xp1、最大角顶点纵坐标Yp1、最大角顶点到次大角顶点的向量、最大角顶点到最小角顶点的向量、最大角顶点p1的属性、次大角顶点p2的属性、最小角顶点p3的属性。
第二非等腰三角形的几何数据包括：最大边长l1`、次大边长l2`、最小边长l3`、最大角度1`、次大角度2`、最小角度3`、最大角顶点横坐标Xp1`、最大角顶点纵坐 标Yp1`、最大角顶点到次大角顶点的向量、最大角顶点到最小角顶点的向量、最大角顶点p1`的属性、次大角顶点p2`的属性、最小角顶点p3`的属性。
将指纹模板所构筑的三角形与现场指纹所构筑的三角形进行匹配。如果两个三角形同时满足如下条件，则认为这两个三角形匹配：
1、三顶点的属性一致
2、三边长度差值在3内(较佳阈值，本发明不做限制)
3、三角度差值在3内(较佳阈值，本发明不做限制)
重复上述过程，直至获取40对(较佳阈值，本发明不做限制)匹配的三角形。如果无法获取到40对匹配的三角形，则认为两枚指纹不匹配。
在获取40对匹配的三角形的条件下，以现场指纹所构筑的三角形的最大角顶点为基准点，计算现场指纹所构筑的三角形相较于纹模板所构筑的三角形的旋转角度和平移距离。平移距离包括：
水平位移＝Xp1-Xp1`
垂直位移＝Yp1-Yp1`
统计这40对匹配三角形之间的旋转角度和平移距离，分别取出现频率最高的数值作为现场指纹的旋转系数和平移系数，并对现场指纹进行旋转和平移。
步骤三、在步骤二的基础上，可再对现场指纹与模板指纹进行特征点匹配数量的计算。
对现场指纹和指纹模板的特征点信息进行比较。如果两个特征点的坐标相同、属性一致，则认为两个特征点匹配。统计匹配点的个数：
若匹配点个数大于9(较佳阈值，本发明不做限制)，进入步骤四。
若匹配点个数不大于9(较佳阈值，本发明不做限制)，则认为两枚指纹不匹配。
四、将现场指纹与指纹模板进行比对。
将匹配的现场指纹的特征点信息的8bits角度特征值量化为6bits角度特征值，代入多项式，令其等于指纹模板的多项式值。至少可以联立如下方程组求解多项式系数：
P(Xi)＝C0+C0Xi+C0Xi 2+…….+C0Xi 8
P(Xi)＝C1+C1Xi+C1Xi 2+…….+C1Xi 8
P(Xi)＝C2+C2Xi+C2Xi 2+…….+C2Xi 8
......
P(Xi)＝C8+C8Xi+C8Xi 2+…….+C8Xi 8
将量化后的6bits角度特征值代入多项式，即可求得根据现场指纹的特征信息求得的C0～C8。
为简化计算，实际求解方程组的运算在2^16的Galois域内进行。
最后，将根据现场指纹的特征信息求得的C0～C3，进行hash运算产生128位的数列，取数列前80位，分成5组，每组16位，依次获得其对应的C4～C8。比较如此得到的C4～C8与之前解方程组得到的C4～C8是否一致，是则现场指纹与模板指纹匹配，两枚指纹比对通过，否则两枚指纹不匹配。上述步骤二至四的流程图可同时参考图7所示。
本发明的主要原理即是验证解出的系数之间关系是否和事先设定的一致。本发明的有益之处：指纹模板不存储完整的指纹特征点信息，仅记录了特征点的坐标信息和属性信息，以及一个多项式的值。这些信息不足以用来还原指纹图像，也不足以单独用来进行指纹比对，因此，保障了用户的隐私和安全。传统的模糊算法在进行多项式映射时，比对信息出现在模板上,它通过增加10倍以上的杂凑点来模糊比对信息。本发明的算法中比对信息不出现在模板，因此也不用杂凑点。
工业实用性
本发明所述的基于模糊特征点信息的指纹识别方法，指纹模版为不完整的特征点信息，仅记录了特征点的坐标信息和属性信息，以及一个多项式的值。这些信息不足以用来还原指纹图像，也不足以单独用来进行指纹比对，通过将现场指纹的特征点信息代入多项式求解系数，判断系数是否满足预定的规律来判断现场指纹是否与模版指纹匹配。因此，保障了用户的隐私和安全。
以上所述，仅为本发明较佳的具体实施方式，但本发明的保护范围并不局限于此，任何熟悉本技术领域的技术人员在本发明揭露的技术范围内，可轻易想到的变化或替换，都应涵盖在本发明的保护范围之内。因此，本发明的保护范围应该以权利要求的保护范围为准。

Claims (10)

  1. 一种基于模糊特征点信息的指纹模板，其特征在于，所述基于模糊特征点信息的指纹模板包括：根据一模板指纹的部分特征点信息建立的若干模糊特征点信息，所述若干模糊特征点信息不足以还原所述模板指纹；
    模糊特征点信息包括配准信息和比对信息，所述配准信息包括所述模糊特征点信息的横坐标、纵坐标和属性，所述比对信息包括一多项式P(Xi)＝C0+C1Xi+C2Xi 2+…….+CnXi n的值，n为多项式的阶数，所述Xi为所述模糊特征点信息中除所述配准信息以外的一个或多个特征点信息所量化得到的数值，所述C0～Cn为具有一预定规则的系数。
  2. 根据权利要求1所述的基于模糊特征点信息的指纹模板，其特征在于，所述预定规则包括：
    所述C0～Cn包括第一部分和第二部分，所述第一部分的值为随机产生，所述第二部分的值为将所述第一部分的值按照所述预定规则进行计算而获取的值。
  3. 一种基于模糊特征点信息的指纹识别方法，其特征在于，包括步骤：S1、获取一基于模糊特征点信息的指纹模板以及现场指纹；S4、将所述现场指纹与所述基于模糊特征点信息的指纹模板进行比对；
    所述基于模糊特征点信息的指纹模板包括：根据一模板指纹的部分特征点信息建立的若干模糊特征点信息，所述若干模糊特征点信息不足以还原所述模板指纹；
    模糊特征点信息包括配准信息和比对信息，所述配准信息包括所述模糊特征点信息的横坐标、纵坐标和属性，所述比对信息包括一多项式P(Xi)＝C0+C1Xi+C2Xi 2+…….+CnXi n的值，n为多项式的阶数，所述Xi为所述模糊特征点信息中除所述配准信息以外的一个或多个特征点信息所量化得到的数值，所述C0～Cn为具有一预定规则的系数；
    所述步骤S4包括：将所述现场指纹的对应特征点信息代入所述多项式，计算所述C0～Cn的值并判断计算得到的所述C0～Cn是否满足所述预定规则，若判断结果为是，则所述现场指纹与所述基于模糊特征点信息的指纹模板匹配，若判断结果为否，则所述现场指纹与所述基于模糊特征点信息的指纹模板不匹配。
  4. 根据权利要求3所述的基于模糊特征点信息的指纹识别方法，其特征在于，所述步骤S1与步骤S4之间还包括：S2、将所述现场指纹与所述基于模糊特征点信息的指纹模板进行指纹配准；
    所述步骤S2包括步骤：
    a、选取三个所述基于模糊特征点信息的指纹模板的模糊特征点信息的横坐标、纵坐标作为顶点，构建一个第一非等腰三角形，并记录所述第一非等腰三角形的几何数据；
    b、选取三个现场指纹的特征点信息的横坐标、纵坐标作为顶点，构建一个第二非等腰三角形，并记录所述第二非等腰三角形的几何数据；
    c、将所述第一非等腰三角形与所述第二非等腰三角形进行匹配；
    d、重复执行步骤a至步骤c，若可获取大于第一阈值对匹配的非等腰三角形，则执行e；若无法获取大于所述第一阈值对匹配的非等腰三角形，则所述现场指纹与所述基于模糊特征点信息的指纹模板不匹配；
    e、计算所述现场指纹相较于所述基于模糊特征点信息的指纹模板的旋转角度和平移距离，并根据计算结果对所述现场指纹进行旋转和/或平移，以与所述基于模糊特征点信息的指纹模板配准。
  5. 根据权利要求4所述的基于模糊特征点信息的指纹识别方法，其特征在于，所述几何数据包括：最大边长、次大边长、最小边长、最大角度、次大角度、最小角度、最大角顶点横坐标、最大角顶点纵坐标、最大角顶点到次大角顶点的向量、最大角顶点到最小角顶点的向量、最大角顶点的属性、次大角顶点的属性以及最小角顶点的属性。
  6. 根据权利要求4述的基于模糊特征点信息的指纹识别方法，其特征在于，步骤e所述的计算所述现场指纹相较于所述基于模糊特征点信息的指纹模板的旋转角度和平移距离，并根据计算结果对所述现场指纹进行旋转和/或平移包括：以所述现场指纹的最大角顶点为基准点，分别计算所述第一阈值对匹配的非等腰三角形的所述旋转角度和平移距离，以计算得到的频率最高的数值作为所述现场指纹的旋转系数和平移系数进行旋转和平移。
  7. 根据权利要求4所述的基于模糊特征点信息的指纹识别方法，其特征在于，步骤c中匹配成功包括以下条件：
    所述现场指纹与所述基于模糊特征点信息的指纹模板对应的三对顶点的属性一致；
    所述现场指纹与所述基于模糊特征点信息的指纹模板对应的三对边的长度差值在第二阈值内；
    所述现场指纹与所述基于模糊特征点信息的指纹模板对应的三对角的角度差值在第三阈值内。
  8. 根据权利要求7所述的基于模糊特征点信息的指纹识别方法，其特征在于，在所述步骤S2与所述步骤S4之间还包括步骤：
    S3、计算经过旋转和平移后的所述现场指纹的横坐标、纵坐标以及属性与所述基于模糊特征点信息的指纹模板的配准信息的特征点匹配个数，若计算结果不大于第四阈值，则所述现场指纹与所述基于模糊特征点信息的指纹模板不匹配，若计算结果大于所述第四阈值，则执行步骤S4。
  9. 根据权利要求3所述的基于模糊特征点信息的指纹识别方法，其特征在于，所述C0～Cn包括第一部分和第二部分，所述第一部分的值为随机产生，所述第二部分的值为将所述第一部分的值按照所述预定规则进行计算而获取的值。
  10. 根据权利要求9所述的基于模糊特征点信息的指纹识别方法，其特征在于，所述步骤S4中的所述将现场指纹的特征点信息代入所述多项式，计算所述C0～Cn的值并判断计算得到的所述C0～Cn是否满足所述预定规则包括：
    将现场指纹的特征点信息代入所述多项式，计算所述C0～Cn的值，获取所述C0～Cn中所述第一部分的值，将所述第一部分的值按照所述预定规则进行计算得到第二部分的值，判断所述按规则计算得到的第二部分的值与所述按解多项式方程组所得的第二部分的值是否相同。
PCT/CN2014/094781 2013-12-31 2014-12-24 一种基于模糊特征点信息的指纹模板及指纹识别方法 WO2015101197A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US14/902,288 US9886619B2 (en) 2013-12-31 2014-12-24 Fingerprint template based on fuzzy feature point information and fingerprint identification method

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201310754835.1A CN104751112B (zh) 2013-12-31 2013-12-31 一种基于模糊特征点信息的指纹模板及指纹识别方法
CN201310754835.1 2013-12-31

Publications (1)

Publication Number Publication Date
WO2015101197A1 true WO2015101197A1 (zh) 2015-07-09

Family

ID=53493195

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2014/094781 WO2015101197A1 (zh) 2013-12-31 2014-12-24 一种基于模糊特征点信息的指纹模板及指纹识别方法

Country Status (3)

Country Link
US (1) US9886619B2 (zh)
CN (1) CN104751112B (zh)
WO (1) WO2015101197A1 (zh)

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105447437B (zh) * 2015-02-13 2017-05-03 比亚迪股份有限公司 指纹识别方法和装置
CN106330454B (zh) * 2016-08-16 2019-08-06 优云合一（北京）科技有限公司 一种数字证书的生成方法及验证方法
CN108960039B (zh) * 2018-05-07 2021-08-06 西安电子科技大学 一种基于对称hash的不可逆指纹模板加密方法
CN110472459B (zh) * 2018-05-11 2022-12-27 华为技术有限公司 提取特征点的方法和装置
IL281410B2 (en) * 2018-09-15 2024-02-01 Quantum Star Tech Inc Bit-level data generation and artificial intelligence techniques and architectures for data protection
CN113239137B (zh) * 2021-05-20 2023-04-14 北京智慧图科技有限责任公司 一种Wi-Fi SLAM进行指纹生成及维护的方法
CN113409259B (zh) * 2021-06-09 2022-04-19 电子科技大学 基于图像特征信息的精密工件台倾角检测方法
CN115527244B (zh) * 2022-11-28 2023-03-21 深圳市航顺芯片技术研发有限公司 一种指纹图像匹配方法、装置、计算机设备及存储介质

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101013943A (zh) * 2007-02-14 2007-08-08 北京邮电大学 一种利用指纹细节特征点绑定/恢复密钥的方法
US20080013804A1 (en) * 2005-12-09 2008-01-17 Electronics And Telecommunications Research Institute Method and apparatus for recognizing fingerprint by hiding minutiae
CN101777115A (zh) * 2009-11-25 2010-07-14 中国科学院自动化研究所 一种安全的指纹认证方法及系统

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1671260B1 (en) * 2003-10-01 2014-06-11 Authentec, Inc. Methods for finger biometric processing and associated finger biometric sensors
JP4886371B2 (ja) * 2006-06-07 2012-02-29 株式会社日立情報制御ソリューションズ 生体認証方法及びシステム
US9042606B2 (en) * 2006-06-16 2015-05-26 Board Of Regents Of The Nevada System Of Higher Education Hand-based biometric analysis
JP5050644B2 (ja) * 2007-05-15 2012-10-17 ソニー株式会社 登録装置、照合装置、プログラム及びデータ構造
JP5050642B2 (ja) * 2007-05-15 2012-10-17 ソニー株式会社 登録装置、照合装置、プログラム及びデータ構造
KR100996466B1 (ko) * 2008-10-09 2010-11-25 조선대학교산학협력단 비밀분산 기법을 이용한 지문정보 저장 장치, 비밀분산 기법을 이용한 지문 인증 시스템 및 비밀분산 기법을 이용한 지문 인증 방법
US10445555B2 (en) * 2009-01-27 2019-10-15 Sciometrics, Llc Systems and methods for ridge-based fingerprint analysis
KR101255555B1 (ko) * 2009-11-24 2013-04-17 한국전자통신연구원 보안성이 강화된 지문인식 방법 및 장치
CN101853382B (zh) * 2010-05-18 2012-08-15 清华大学 一种求取现场指纹方向的方法和装置
CN103984929B (zh) * 2014-05-20 2017-04-19 清华大学 扭曲指纹的校正方法及系统

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080013804A1 (en) * 2005-12-09 2008-01-17 Electronics And Telecommunications Research Institute Method and apparatus for recognizing fingerprint by hiding minutiae
CN101013943A (zh) * 2007-02-14 2007-08-08 北京邮电大学 一种利用指纹细节特征点绑定/恢复密钥的方法
CN101777115A (zh) * 2009-11-25 2010-07-14 中国科学院自动化研究所 一种安全的指纹认证方法及系统

Also Published As

Publication number Publication date
CN104751112B (zh) 2018-05-04
US20160371532A1 (en) 2016-12-22
CN104751112A (zh) 2015-07-01
US9886619B2 (en) 2018-02-06

Similar Documents

Publication Publication Date Title
WO2015101197A1 (zh) 一种基于模糊特征点信息的指纹模板及指纹识别方法
Qi et al. A singular-value-based semi-fragile watermarking scheme for image content authentication with tamper localization
JP6353910B2 (ja) バイオメトリックテンプレートセキュリティおよびキー生成
US8699799B2 (en) Fingerprint verification method and apparatus with high security
JP5431362B2 (ja) 画像識別のための特徴ベースの識別特性（ｓｉｇｎａｔｕｒｅ）
CN104166955B (zh) 基于保角变换图像哈希生成及图像篡改检测定位方法
Wojtowicz et al. Digital images authentication scheme based on bimodal biometric watermarking in an independent domain
CN108182401B (zh) 一种基于聚合块信息的安全虹膜识别方法
CN108550166B (zh) 一种空间目标图像匹配方法
JP2008287428A5 (zh)
JP5534411B2 (ja) 画像処理装置
Bacchuwar et al. A jump patch-block match algorithm for multiple forgery detection
CN115168210A (zh) 一种联邦学习中在黑盒场景下基于对抗样本的鲁棒水印遗忘验证方法
Nguyen et al. A fingerprint fuzzy vault scheme using a fast chaff point generation algorithm
CN111553241A (zh) 掌纹的误匹配点剔除方法、装置、设备及存储介质
Fadl et al. Copy-rotate-move forgery detection based on spatial domain
CN106656506A (zh) 一种手指静脉加密方法
CN116051349A (zh) 一种基于brisk-freak的医学图像水印方法
Zhang et al. Global resynchronization-based image watermarking resilient to geometric attacks
Vidya et al. Image steganography using ken ken puzzle for secure data hiding
US9208352B2 (en) LFSR watermark system
CN108009464B (zh) 一种指纹识别方法和装置
Hooda et al. Novel chaff generation for fingerprint fuzzy vault
JP4257264B2 (ja) パターン照合装置、画像品質検証方法及び画像品質検証プログラム
Mall et al. Non-blind method of detection and localization of structural tampering using robust hash-like function and similarity metric for digital images

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 14876815

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 14902288

Country of ref document: US

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 14876815

Country of ref document: EP

Kind code of ref document: A1