WO2010069166A1 - 快速指纹搜索方法及快速指纹搜索系统 - Google Patents

快速指纹搜索方法及快速指纹搜索系统 Download PDF

Info

Publication number
WO2010069166A1
WO2010069166A1 PCT/CN2009/071257 CN2009071257W WO2010069166A1 WO 2010069166 A1 WO2010069166 A1 WO 2010069166A1 CN 2009071257 W CN2009071257 W CN 2009071257W WO 2010069166 A1 WO2010069166 A1 WO 2010069166A1
Authority
WO
WIPO (PCT)
Prior art keywords
fingerprint
template
feature
site
templates
Prior art date
Application number
PCT/CN2009/071257
Other languages
English (en)
French (fr)
Inventor
刘中秋
李健
吕虹晓
侯艳芹
陈高曙
蒋文琦
Original Assignee
杭州中正生物认证技术有限公司
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 杭州中正生物认证技术有限公司 filed Critical 杭州中正生物认证技术有限公司
Priority to EP09832848A priority Critical patent/EP2360619A4/en
Publication of WO2010069166A1 publication Critical patent/WO2010069166A1/zh

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/12Fingerprints or palmprints
    • G06V40/1365Matching; Classification
    • G06V40/1371Matching features related to minutiae or pores
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/12Fingerprints or palmprints
    • G06V40/1365Matching; Classification
    • G06V40/1376Matching features related to ridge properties or fingerprint texture

Definitions

  • the invention belongs to biometric identification technology, and particularly relates to a fast fingerprint searching method and a fast fingerprint searching system.
  • Biometric technology is a technology that uses human biometrics to determine an individual's identity and can be widely used in criminal investigation, security, banking, and other fields.
  • fingerprint authentication technology the most mature and widely used biometric authentication technology. Because fingerprints vary from person to person, from finger to finger, easy to collect and identify, and fingerprint authentication is easy to use and highly secure, fingerprints can be seen in the security field, financial field and people's life. Products, such as fingerprint access control, fingerprint notebook, fingerprint bank refusal machine, fingerprint bank POS machine, fingerprint U disk, etc.
  • Fingerprint authentication mainly includes two parts: fingerprint collection and fingerprint recognition, namely: pre-collecting the user's fingerprint as the fingerprint template of the fingerprint template library; during subsequent authentication, comparing the user fingerprint collected by the fingerprint with the fingerprint template in the fingerprint template library , find the matching fingerprint template.
  • a fingerprint identification method which forms an early mainstream mode of the fingerprint data search algorithm.
  • the specific search steps include: (1), on-site fingerprint feature and all fingerprint templates in the fingerprint template library. : 1 quick comparison, get their similarity.
  • (2) According to the similarity of the fingerprint template, the first N templates with higher similarity are extracted, and the exact comparison is performed to obtain the comparison result.
  • the fast alignment is a simplified mode of exact alignment, which compares all fingerprint templates according to the detail point topology.
  • the fingerprint search takes a relatively long time according to the method described in the invention patent.
  • Patent No. 200610065297.5 entitled “Fingerprint Identification Method and System”
  • fingerprint search process which mainly relies on the similarity of the detail point topology map, combining the singular point and the average ridge density. And block pattern features. For most images, it is still necessary to calculate the similarity of the detail point topology, and the calculation time is long.
  • you simply compare the bits of the singular point Set, direction and type the obtained similarity is added to the global feature similarity, the calculation accuracy of the singular point is high, and the performance is only slightly improved, but the search takes a long time. At the same time, the utilization rate of singular points is low.
  • the stored fingerprint template library is a massive database.
  • the fingerprint of the user collected will be collected from a large amount.
  • the main object of the present invention is to provide a fast fingerprint searching method, which solves the technical problem that the large-capacity data in the prior art can not quickly complete the comparison between the fingerprint of the on-site fingerprint and the fingerprint database on the basis of ensuring accuracy. .
  • Another object of the present invention is to provide a fast fingerprint search system.
  • the present invention provides a fast fingerprint searching method, which divides a fingerprint template in a fingerprint template library into a center point template and a centerless point template, and the method includes:
  • the fingerprint pattern that needs to be matched is collected to complete the on-site fingerprint feature extraction
  • the extracted on-the-spot fingerprint feature has a central point on-site fingerprint feature or a no-center on-site fingerprint feature. If there is a central point on-site fingerprint feature, the on-site fingerprint feature is compared with all the central point templates in the fingerprint template library. If there is no center point on-site fingerprint feature, the on-site fingerprint feature is compared with all the non-center point templates in the fingerprint template library.
  • the method further includes:
  • the on-site fingerprint feature is compared with all the non-centered point templates in the fingerprint template library.
  • the method further includes:
  • the on-site fingerprint feature is compared with all the central point templates in the fingerprint template library.
  • the method further includes:
  • the result of the comparison between the on-site fingerprint feature and the template is that when the template is successfully matched, the on-site fingerprint feature is determined to be verified by fingerprint; otherwise, the on-site fingerprint feature is determined to be an illegal fingerprint.
  • the comparing the on-site fingerprint feature with all the central point templates in the fingerprint template library is specifically:
  • the on-site fingerprint feature is accurately compared with all the templates in the set M x respectively.
  • the comparing the on-site fingerprint feature with all the non-center point templates in the fingerprint template library is specifically:
  • the center point on-site fingerprint feature is compared with all the templates in the set My, respectively.
  • the invention also provides a fast fingerprint search system, comprising:
  • the fingerprint template storage unit is configured to store a fingerprint template library, wherein the fingerprint template in the fingerprint template library is divided into a center point template and a centerless point template;
  • a fingerprint collection device for collecting a live fingerprint pattern to be compared
  • a feature point extracting device configured to extract feature points from the fingerprint pattern collected by the fingerprint collecting device
  • the feature judging unit is configured to determine that the on-the-spot fingerprint feature extracted by the feature point extracting device is a central point on-site fingerprint feature or a no-center spot fingerprint feature;
  • the fingerprint matching unit is configured to determine, according to the judgment result of the feature judging unit, if the on-site fingerprint feature has a central point on-site fingerprint feature, the center fingerprint feature and the fingerprint template storage unit stored in the fingerprint template storage unit have all centers The point templates are separately compared; if there is no center point on-site fingerprint feature, the scene fingerprint features are compared with all the centerless point templates in the fingerprint template library stored by the fingerprint template library storage unit.
  • the fingerprint comparison unit is also used for:
  • the on-site fingerprint feature is compared with all the non-centered point templates in the fingerprint template library;
  • the live fingerprint feature is compared with all the center point templates in the fingerprint template library.
  • the fingerprint comparison unit includes:
  • a ridge width histogram comparison device configured to compare the on-site fingerprint feature with a template of a preset range in all fingerprint template libraries, respectively, and find a comparison result that is not less than a predetermined ridge width A set of templates of histogram similarity thresholds;
  • a central point quick comparison device configured to quickly compare the on-the-spot fingerprint feature with all templates of a preset range, and find a set of templates with the highest pre-set number of presets;
  • the device is configured to accurately compare the on-site fingerprint feature with all templates of the preset range, and find a template that is accurately matched to be a matching fingerprint of the on-site fingerprint;
  • a detail point quick comparison device configured to quickly compare the on-the-spot fingerprint feature with a detail point of all templates of a preset range, and find a template set of the first set number with the highest similarity;
  • the device is configured to accurately compare the on-the-spot fingerprint feature with all the templates of the preset range, and find a template that is accurately matched to be a matching fingerprint of the live fingerprint.
  • the invention also provides a fast fingerprint search system, comprising: a fingerprint processing subsystem and a plurality of terminals, wherein the terminal is connected to the fingerprint processing subsystem through a dedicated line or a network, and the fingerprint processing subsystem further comprises Database and server; among them,
  • the terminal is configured to collect a fingerprint pattern to be compared, and complete on-site fingerprint feature extraction
  • the database is used to store the homepage template library, wherein the template in the template library is divided into a center point template and a center point template;
  • the server is configured to determine that the on-site fingerprint feature extracted by the terminal is a central point on-site fingerprint feature or a no-center on-site fingerprint feature. If there is a central point on-site fingerprint feature, the on-site fingerprint feature and the fingerprint template library are all The center point templates are separately compared; if there is no center point on-site fingerprint feature, the scene fingerprint features are compared with all the centerless point templates in the fingerprint template library.
  • the server is further configured to:
  • the on-site fingerprint feature is compared with all the non-centered point templates in the fingerprint template library;
  • the live fingerprint feature is compared with all the center point templates in the fingerprint template library.
  • the present invention has the following advantages:
  • the fingerprint search algorithm of the present invention can quickly filter the data in the fingerprint template library by strategically improving the existing fingerprint template search method, especially the classification and step-by-step search process, which greatly improves the data. calculating speed;
  • the fingerprint search algorithm of the present invention improves the search accuracy by making a more reasonable search and comparison process, so that it is closer to the accurate comparison index on the correctness indicator;
  • the fingerprint search algorithm of the present invention increases the utilization of the center point, so that the advantage of the majority of the center fingerprint features in the search alignment is fully expanded.
  • the ridge width histogram method is also added to the fingerprint search algorithm of the present invention, and the fingerprint search algorithm is further improved.
  • the search speed bottleneck caused by the expansion of fingerprint data under the network is broken, which opens up a new space for the rapid development of the fingerprint identification device in the networking direction.
  • DRAWINGS 1 is a relationship diagram of classification processing in a fingerprint template library and an on-site fingerprint feature in the fast fingerprint search method of the present invention
  • FIG. 2 is a schematic flow chart of the first step of the on-site fingerprint feature with a central point in the fast fingerprint search method of the present invention
  • FIG. 3 is a schematic flowchart of the second step of the on-site fingerprint feature with a central point in the fast fingerprint search method according to the present invention
  • FIG. 4 is a schematic flow chart of the first step of performing a classified search in the fast fingerprint searching method of the present invention
  • FIG. 5 is a schematic flowchart of the second step of performing a classified search in the fast fingerprint searching method of the present invention.
  • FIG. 6 is a schematic structural diagram of a fast fingerprint search system of the present invention.
  • FIG. 7 is a schematic diagram of another principle structure of the fast fingerprint search system of the present invention.
  • the fast fingerprint searching method and the fast fingerprint searching system of the present invention are specifically described below with reference to the accompanying drawings.
  • the fingerprint search of the present invention can quickly filter the data in the fingerprint template library by classifying step-by-step search, which greatly improves the operation speed. Before the present invention is specifically described, the following terms will be explained in advance so that the examples of the present invention can be clarified.
  • Fingerprint feature A fingerprint image is processed to generate encoded data representing the characteristics of the fingerprint image.
  • Fingerprint template A fingerprint feature stored in the medium for fingerprint verification and search.
  • Center point The center point is the apex of the innermost ridgeline of the fingerprint, which describes the overall characteristics of the fingerprint.
  • Center point exact comparison According to the central point information, accurately calculate the similarity between the fingerprint feature and the fingerprint template, and make a process of whether the two are from the same finger judgment.
  • All fingerprint template sets described in the present invention are divided into a fingerprint template set Q having a center point and a fingerprint template set having no center point as T.
  • the on-site fingerprint feature only has a central point or no central point.
  • FIG. 1 is a flowchart of the first step of the fast fingerprint searching method of the present invention. It includes the following steps:
  • the fingerprint template in the template library is divided into a center point template and a center point template.
  • the fingerprint template library collects fingerprint templates, the fingerprints need to be divided into a center point template and a centerless template.
  • Most of the templates in the entire template library are templates with a central point.
  • the following is a processing method of a fingerprint template disclosed in Patent No. 200610065297.5.
  • the fingerprint template processing methods of different companies may be different.
  • the present invention needs to be described that the present embodiment can be processed by a fingerprint template processing method disclosed in 200610065297.5, but is not limited thereto, and is for illustrative purposes only. To avoid the problem that the examiner believes that the content is not fully disclosed.
  • the processing method of a fingerprint template disclosed in Patent No. 200610065297.5 is as follows: When the fingerprint template library collects the fingerprint template, the fingerprint template needs to be pre-processed and normalized, the block pattern is calculated to extract the singular points, the calculation direction is calculated, and the background area is divided. And refine the singular points; image filtering and enhancement; calculate the ridge density; binarize the image and refine it, extract the minutiae points, verify the minutiae points, delete the pseudo-detail points, determine whether there is a center point, and extract the original according to the Poincare Index method.
  • the center point is processed according to the relationship between the original center point and the pattern, and the pseudo center point is removed.
  • the center point information includes position, orientation, and quality information. Fingerprint minutiae, singularity, average ridge density, and block directional features are eventually compressed into live fingerprint features or templates stored in the fingerprint template library.
  • the fingerprint template library can be pre-divided into two sub-libraries, a central point template library and a no-center point template library. After the fingerprint template is pre-processed and normalized, it is determined in which library the fingerprint template is placed for storage according to whether it has a center point feature.
  • the present invention can also set an identifier for indicating the presence or absence of a center point feature in the data of each fingerprint template. For example, if the fingerprint template has a center point feature, the identifier is marked as "1", and if the fingerprint template has no center point Characteristic, then the identifier is marked as "0".
  • the on-site fingerprint feature is characterized by a central point on-site fingerprint feature or a centerless on-site fingerprint feature.
  • the user's fingerprint is collected on the spot, that is, the fingerprint pattern to be compared is collected, the on-site fingerprint feature extraction is completed, and the extracted on-site fingerprint feature is determined to have a central point on-site fingerprint feature or There is no central point on-site fingerprint feature, and most of the on-site fingerprint features are centered.
  • the process of extracting the fingerprint feature of the site is similar to the process of extracting the fingerprint template involved in step S010, and will not be described here. These are all prior art and will not be described in detail.
  • the search process of the on-site fingerprint feature having the central point is entered, and the processing steps shown in FIG. 2 are performed; if the on-site fingerprint is to be compared, the feature is not centered. Then, the search process of the on-site fingerprint feature without the center point is entered, and the processing steps shown in FIG. 4 are performed.
  • step S110 is performed, and if not, the centerless point search process shown in Figs. 4 and 5 is performed.
  • An example of the present invention may also be to end the search process when the live fingerprints to be compared do not have a center point feature, such an example providing a fingerprint alignment with center point features.
  • the ridge width histogram represents the geometry of the ridgeline width distribution of a fingerprint image, in which the abscissa represents the ridgeline width value and the ordinate represents the number of occurrences of the ridgeline width value.
  • the extraction of the ridge width histogram is a well-known technique and will not be described in detail herein.
  • the ridge line width histogram similarity calculation formula is as follows:
  • the ridge width histogram similarity between the template and the feature is:
  • the invention defines a ridge width histogram similarity threshold value, which is used for determining whether the histogram of the fingerprint feature and the histogram of the fingerprint template are similar;
  • the ridge width histogram similarity threshold is preset, or it can be determined by the technician according to the situation.
  • the ridge width histogram similarity is the value.
  • the center point template which is found in the example is not less than the predetermined ridge width histogram similarity threshold, and constitutes the set Q l. According to the comparison result, the fingerprint template is placed in two Different collections.
  • the present invention can also set in advance an identifier for indicating whether the comparison result is not smaller than or smaller than a predetermined ridge width histogram similarity threshold value in the data of each fingerprint template, for example, if the fingerprint template has a comparison result not less than a predetermined value If the ridge width histogram similarity is large, the label is marked as "1". If the fingerprint template has a comparison result smaller than the predetermined ridge width histogram similarity threshold, the label is marked as "0". Each time the fingerprint to be compared is compared with the fingerprint template to perform a ridge width histogram comparison, the corresponding identifier of the fingerprint template is modified.
  • S130 Perform accurate comparison of the center point on-site fingerprint feature with all templates in the set, and find a template that is accurately matched as the matching fingerprint corresponding to the center point on-site fingerprint feature.
  • the present invention defines an exact comparison score for the value of the score: determining whether the on-site feature fingerprint and the template are from the same finger's score threshold. Wherein, when the exact comparison between the fingerprint feature of the scene and the template is not less than the preset If the exact comparison score is wide, the fingerprint of the scene fingerprint and the fingerprint of the template are from the same finger; when the accurate comparison of the fingerprint feature of the scene and the template is smaller than the preset accurate comparison score, the fingerprint of the on-site fingerprint feature and the template comes from Different fingers.
  • center point fast alignment algorithm and the center point exact alignment algorithm may be well known algorithms and will not be described in detail herein.
  • the center point fingerprint feature is centered with all the templates in the set to accurately compare the center points, and the similarity score is calculated. Once a template with a similarity score that is not less than a given exact alignment score is present, the feature is considered to be verified by fingerprinting, ending the search process.
  • the search process of the second step is performed, as shown in FIG.
  • most of the on-site fingerprint features with a central point will end the search process here and get the correct template. Therefore, if the accuracy is not high, only the step flow shown in Fig. 2 can be performed.
  • S210 comparing the ridge line width histogram with the center point fingerprint feature of the center point and all the non-center point templates in the fingerprint template library, and finding a template whose comparison result is not less than a predetermined ridge width histogram similarity width value, and is composed of Collection ⁇ .
  • S220 Quickly compare the feature points of the central point fingerprint feature with the template in the set ⁇ , and find the previous template composition set with the highest similarity ⁇ 2 , and ⁇ 2 is a natural number.
  • S230 Exactly compare the feature points of the center point on-site fingerprint feature with all the templates in the set ⁇ 2 , and calculate a similarity score, and if the similarity score is not less than a template of a given exact comparison score, It is considered that the fingerprint feature of the central point is verified by fingerprinting, and the search process is ended, otherwise it is judged as an illegal on-site fingerprint feature.
  • S310 comparing the non-center point on-site fingerprint feature with all the non-center point templates in the fingerprint template library, respectively, and comparing the ridge width histograms, and finding a template whose comparison result is not less than a similar ridge width histogram similarity threshold, Collection T 2 .
  • S320 The core point of the scene fingerprint feature of the template set ⁇ 2 minutia fast comparison to find the highest similarity ⁇ 3 before the template to form a set ⁇ 3; ⁇ 3 is a natural number.
  • the on-site fingerprint feature without a center point the vast majority will end the search process here and get the correct template. Therefore, if the accuracy is not high, the on-site fingerprint feature without the center point can only be carried out to this step.
  • Second step When the center point fingerprint feature is accurately compared with all the templates in the set ⁇ 3 , and the template for accurate comparison is not found, the step flow shown in FIG. 5 can be performed. Second step
  • S410 Accurately compare the feature points of the non-center point on-site fingerprint feature with all the templates in the set ⁇ 3 , and if no template is successfully found, the center fingerprint feature and the fingerprint template library have all the centers.
  • the point template performs a ridge line histogram comparison, respectively, and finds a template whose comparison result is not less than a predetermined ridge width histogram similarity threshold, and constitutes a set Q 2 .
  • S420 The core point of the fingerprint feature site template with the set of Q 2, respectively minutia fast comparison, calculating a similarity score, the highest similarity score obtained before the templates set M4; is a natural number.
  • S430 Perform exact comparison between the centerless point fingerprint feature and all templates in the set M4, and calculate a similarity score. Once the similarity score is not less than a template of a given exact ratio score, the template is similar. It is considered that the fingerprint feature of the centerless point is verified by fingerprinting, and the search process is ended. Otherwise, it is determined as an illegal on-site fingerprint feature.
  • the fingerprint search algorithm of the present invention can quickly filter the data in the fingerprint template library by strategically improving the existing fingerprint template search method, especially the classification and step-by-step search process, thereby greatly improving the operation speed. .
  • FIG. 6 is a schematic structural diagram of the first fast fingerprint search system of the present invention. It includes a number of terminals 11 including a fingerprint collector and a fingerprint processing subsystem 12, which is connected to the fingerprint processing subsystem 12 via a dedicated line or network.
  • the fingerprint processing subsystem 12 further includes a database 121 and a server 122:
  • the database 121 is configured to store the fingerprint template library, wherein the fingerprint template in the template library is divided into a center point template and a centerless point template.
  • the database 121 at least includes: a fingerprint template library storage unit for storing a fingerprint template library and a flow storage unit including a save processing flow.
  • the server 122 is configured to determine that the on-site fingerprint feature extracted by the terminal is a central point on-site fingerprint feature or a no-center on-site fingerprint feature. If there is a central point on-site fingerprint feature, the on-site fingerprint feature and the fingerprint template library are All the center point templates are compared separately; if there is no center point on-site fingerprint feature, the scene fingerprint features are compared with all the centerless point templates in the fingerprint template library.
  • the server 122 is further configured to: when the comparison between the on-site fingerprint feature and the all the center point templates is that there is no successful template, the scene fingerprint feature and the fingerprint template library are all absent.
  • the center point template is separately compared; when the field fingerprint feature is compared with all the centerless point templates, the result is that there is no successful template, and the scene fingerprint feature and the fingerprint template library are all The center point templates are compared separately.
  • the server 122 further includes:
  • Ridge width histogram comparison device for comparing the on-site fingerprint feature with a template of a preset range in all fingerprint template libraries, respectively, and comparing the ridge width histograms, and finding the comparison result is not less than a predetermined ridge width A set of templates of histogram similarity thresholds;
  • Center point fast comparison device for quickly comparing the on-site fingerprint feature with all templates of a preset range, and finding a set of templates with the highest pre-set number of presets; Pair device: for accurately comparing the on-site fingerprint feature with a center point of all templates of a preset range, and finding a template for accurate comparison as a matching finger of the on-site fingerprint Pattern.
  • the server 122 may further include: a detail point quick comparison device: configured to quickly compare the scene fingerprint feature with all templates of the preset range, and find the pre-set number with the highest similarity Template composition set;
  • Detail point precision comparison device used to accurately compare the scene fingerprint feature with all the templates of the preset range, and find a template that accurately compares the success as the matching fingerprint of the scene fingerprint.
  • the ridge width histogram comparison device, the center point quick comparison device, the center point precision comparison device, the minutiae point quick comparison device, and the minutiae precise alignment device are logic units.
  • the terminal 11 is configured to collect a fingerprint pattern to be compared, and complete on-site fingerprint feature extraction.
  • the terminal 11 further includes:
  • Fingerprint collection device used to collect live fingerprint images to be compared
  • Feature point extraction device It is connected to the fingerprint collection device for extracting feature points from the collected fingerprint image.
  • the fingerprint collection device is usually a fingerprint collector, and the feature point extraction device is usually implemented by a processor in a software manner, which is a logic unit.
  • FIG. 7 is a schematic structural diagram of another fast fingerprint search system according to the present invention. it includes:
  • the fingerprint template storage unit 21 is configured to store a fingerprint template library, wherein the fingerprint templates are classified according to the presence or absence of a center point;
  • Fingerprint collection device 22 used to collect fingerprint images to be compared
  • Feature point extraction device 23 It is connected to the fingerprint collection device for extracting feature points from the collected fingerprint image;
  • the feature judging unit 29 is configured to determine that the feature fingerprint feature extracted by the feature point extracting device 23 is a center point on-site fingerprint feature or a no-center spot fingerprint feature;
  • the fingerprint matching unit 30 is configured to: according to the judgment result of the feature determining unit 29, if the on-site fingerprint feature has a central point on-site fingerprint feature, the on-site fingerprint feature and the fingerprint template library stored in the fingerprint template storage unit 21 are All the center point templates are separately compared; if there is no center point on-site fingerprint feature, the on-site fingerprint feature and the fingerprint template storage unit 21 are stored All the non-centered point templates in the pattern template library are compared separately.
  • the fingerprint comparison unit 30 further includes:
  • Ridge width histogram comparison device 24 for comparing the field fingerprint feature with a template of a preset range in all fingerprint template libraries, respectively, and comparing the ridge width histograms, and finding the comparison result is not less than a predetermined ridge line a set of templates of width histogram similarity thresholds;
  • Center point fast comparison device 25 for quickly comparing the on-site fingerprint feature with all templates of a preset range, and finding the template with the highest degree of pre-preset number to form a central point precision ratio
  • the device 26 is configured to accurately compare the on-site fingerprint feature with all the templates of the preset range, and find a template that is accurately matched as a matching fingerprint of the live fingerprint.
  • the fingerprint matching unit 30 may further include: a detail point quick comparison device 27: for quickly comparing the scene fingerprint feature with a detailed point of all templates of a preset range, and finding a pre-set with the highest similarity Number of templates to form a collection;
  • the detail point precision comparison device 28 is configured to accurately compare the scene fingerprint feature with a preset point of all templates of the preset range, and find a template that accurately compares the success as a matching fingerprint of the scene fingerprint.
  • This example can also have a process storage unit that holds the processing flow.

Landscapes

  • Engineering & Computer Science (AREA)
  • Human Computer Interaction (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Collating Specific Patterns (AREA)

Description

快速指纹搜索方法及快速指纹搜索系统 本申请要求于 2008 年 12 月 19 日提交中国专利局、 申请号为 200810207425.4、 发明名称为 "快速指紋搜索方法及快速指紋搜索系统" 的中 国专利申请的优先权, 其全部内容通过引用结合在本申请中。
技术领域
本发明属于生物特征识别技术,特别涉及一种快速指紋搜索方法及快速指 紋搜索系统。
背景技术
生物认证技术是一项利用人体生物特征来确定个人身份的技术,可广泛应 用于刑侦、 安全、 银行等领域。 目前, 生物认证技术中发展最为成熟、 应用最 为广泛的是指紋认证技术。 由于指紋因人而异、 因指而异、 容易釆集和识别, 且指紋认证方便使用、 安全性高等特点, 在安全领域、 金融领域和民众生活领 域都能看到使用指紋进行身份识别的技术产品, 比如指紋门禁、 指紋笔记本、 指紋银行拒员机、 指紋银行 POS机、 指紋 U盘等。
指紋认证主要包括指紋釆集和指紋识别两部分, 即: 预先釆集用户的指紋 作为指紋模板库的指紋模板; 后续认证时,对釆集到的用户指紋与指紋模板库 中的指紋模板进行对比, 找到与之匹配的指紋模板。
在专利号为 02110873.0 中公开了一种指紋识别方法, 该专利形成了指紋 数据搜索算法的早期主流模式, 具体搜索步骤包括: (1 )、 现场指紋特征与指 紋模板库中的所有指紋模板进行 1 : 1的快速比对, 得到各自的相似度。 (2 )、 根据指紋模板的相似度提取相似度较高的前 N枚模板, 进行精确比对, 得到 比对结果。 在该专利中, 快速比对是精确比对的简化模式, 是根据细节点拓朴 图对所有的指紋模板进行比对。按照该发明专利所述的方法进行指紋搜索所需 时间比较长。
在专利号为 200610065297.5中公开的另一个专利, 发明名称为"指紋识别 方法与系统", 也提到了指紋搜索过程, 它主要依靠细节点拓朴图的相似度, 结合了奇异点、 平均脊密度和块方向图特征。 对于大部分的图像, 仍然需要计 算细节点拓朴图的相似度, 计算时间长。 另外, 若简单地两两比对奇异点的位 置、 方向和类型, 得到的相似度累加到全局特征相似度中, 则对奇异点的计算 准确度要求高, 对性能也仅有微弱改善, 却造成搜索所需时间比较长。 同时对 奇异点的利用率较低。
目前, 随着计算机网络的普及, 提供异地的身份识别成为可能。 以金融系 统为例, 若整一银行系统釆用指紋识别来进行身份认证, 则存储的指紋模板库 是一个海量数据库, 当需要认证某一用户时,将釆集到的该用户的指紋从海量 数据库中找到与之匹配的指紋模板, 按照上述的搜索方法, 消耗时间太长, 根 本无法满足该些需求。 也就是说, 面对大库容的数据量, 以及随后的计算机硬 件系统的不断更新, 如果沿用现有的技术, 仍将大大降低指紋比对消耗时间, 甚至影响比对准确性。 如何更快完成指紋的快速搜索, 并保证结果的准确性, 是当今指紋识别技术的一个瓶颈, 更是对搜索算法的一个挑战。
发明内容
本发明的主要目的在于提供一种快速指紋搜索方法,以解决现有技术中面 对大容量数据,不能在保证准确性的基础上快速完成现场指紋与指紋数据库中 的指紋模板的对比的技术问题。
本发明的另一目的在于提供一种快速指紋搜索系统。
为了达到上述目的, 本发明提供一种快速指紋搜索方法, 将指紋模板库中 的指紋模板分为有中心点模板和无中心点模板, 该方法包括:
釆集需比对的指紋图案, 完成现场指紋特征提取;
判断提取到的现场指紋特征为有中心点现场指紋特征或无中心点现场指 紋特征,如果为有中心点现场指紋特征,将所述现场指紋特征与指紋模板库中 所有有中心点模板分别进行比对; 如果为无中心点现场指紋特征, 则将所述现 场指紋特征与指紋模板库中所有无中心点模板分别进行比对。
其中,当所述现场指紋特征与所述所有有中心点模板进行比对的结果为不 存在比对成功的模板时, 该方法进一步包括:
将所述现场指紋特征与指紋模板库中所有无中心点模板分别进行比对。 当所述现场指紋特征与所述所有无中心点模板进行比对的结果为不存在 比对成功的模板时, 该方法进一步包括:
将所述现场指紋特征与指紋模板库中所有有中心点模板分别进行比对。 该方法进一步包括:
所述现场指紋特征与所述模板的比对结果为存在比对成功的模板时,确定 所述现场指紋特征通过指紋验证; 否则, 确定所述现场指紋特征为非法指紋。
所述将所述现场指紋特征与指紋模板库中所有有中心点模板分别进行比 对具体为:
将所述现场指紋特征与指紋模板库中所有有中心点模板分别进行脊线宽 度直方图比对,从中找到比对结果不小于预定脊线宽度直方图相似度阔值的模 板, 组成集合 Qx;
将所述现场指紋特征与所述集合 Qx中的所述模板分别进行中心点快速比 对, 获取相似度最高的前 ¾个模板组成集合 Mx, nx为自然数;
将所述现场指紋特征与集合 Mx中的所有模板分别进行中心点精确比对。 所述将所述现场指紋特征与指紋模板库中所有无中心点模板分别进行比 对具体为:
将所述现场指紋特征与指紋模板库中所有无中心点模板分别进行脊线宽 度直方图比对,从中找到比对结果不小于预定脊线宽度直方图相似度阔值的模 板, 组成集合 Ty;
将所述有中心点现场指紋特征与所述集合 Ty中的所述模板分别进行细节 点快速比对, 找到相似度最高的前 ny个模板组成集合 My, ny为自然数;
将所述有中心点现场指紋特征与集合 My中的所有模板分别进行细节点精 确比对。
本发明同时提供一种快速指紋搜索系统, 包括:
指紋模板库存储单元, 用于存储指紋模板库, 其中, 指紋模板库中的指紋 模板分为有中心点模板和无中心点模板;
指紋釆集装置, 用于釆集需比对的现场指紋图案;
特征点提取装置, 用于对指紋釆集装置釆集到的现场指紋图案提取特征 点;
特征判断单元,用于判断特征点提取装置提取到的现场指紋特征为有中心 点现场指紋特征或无中心点现场指紋特征; 指紋比对单元, 用于根据特征判断单元的判断结果,如果现场指紋特征为 有中心点现场指紋特征 ,则将所述现场指紋特征与指紋模板库存储单元所存储 的指紋模板库中所有有中心点模板分别进行比对;如果为无中心点现场指紋特 征 ,则将所述现场指紋特征与指紋模板库存储单元所存储的指紋模板库中所有 无中心点模板分别进行比对。
其中, 指紋比对单元还用于:
当所述现场指紋特征与所述所有有中心点模板进行比对的结果为不存在 比对成功的模板时,将所述现场指紋特征与指紋模板库中所有无中心点模板分 别进行比对;
当所述现场指紋特征与所述所有无中心点模板进行比对的结果为不存在 比对成功的模板时,将所述现场指紋特征与指紋模板库中所有有中心点模板分 别进行比对。
指紋比对单元包括:
脊线宽度直方图比对装置,用于将所述现场指紋特征与所有指紋模板库中 预先设定范围的模板分别进行脊线宽度直方图比对,从中找到比对结果不小于 预定脊线宽度直方图相似度阔值的模板组成集合;
中心点快速比对装置,用于将所述现场指紋特征与预先设定范围的所有模 板进行中心点快速比对, 找到相似度最高的前预先设定个数的模板组成集合; 中心点精确比对装置 ,用于将所述现场指紋特征与预先设定范围的所有模 板进行中心点精确比对,寻找精确比对成功的模板作为所述现场指紋的匹配指 紋;
细节点快速比对装置,用于将所述现场指紋特征与将预先设定范围的所有 模板进行细节点快速比对, 找到相似度最高的前设定个数的模板组成集合; 细节点精确比对装置 ,用于将所述现场指紋特征与预先设定范围的所有模 板进行细节点精确比对,寻找精确比对成功的模板作为所述现场指紋的匹配指 紋。
本发明还提供一种快速指紋搜索系统,包括:指紋处理子系统和若干终端, 所述终端通过专线或网络连接至指紋处理子系统,指紋处理子系统进一步包括 数据库和服务器; 其中,
终端, 用于釆集需比对的指紋图案, 完成现场指紋特征提取;
数据库, 用于存储指乡文模板库, 其中, 指故模板库中的指故模板分为有中 心点模板和无中心点模板;
服务器,用于判断终端提取到的现场指紋特征为有中心点现场指紋特征或 无中心点现场指紋特征, 如果为有中心点现场指紋特征, 则将所述现场指紋特 征与指紋模板库中所有有中心点模板分别进行比对;如果为无中心点现场指紋 特征, 则将所述现场指紋特征与指紋模板库中所有无中心点模板分别进行比 对。
其中, 所述服务器还用于:
当所述现场指紋特征与所述所有有中心点模板进行比对的结果为不存在 比对成功的模板时,将所述现场指紋特征与指紋模板库中所有无中心点模板分 别进行比对;
当所述现场指紋特征与所述所有无中心点模板进行比对的结果为不存在 比对成功的模板时,将所述现场指紋特征与指紋模板库中所有有中心点模板分 别进行比对。
相对与现有技术, 本发明具有以下优点:
1.本发明所述的指紋搜索算法通过对现有指紋模板搜索方法进行策略性 的改进措施,尤其是分类和分步的搜索流程,能快速筛选指紋模板库里的数据, 极大的提高了运算速度;
2. 本发明指紋搜索算法通过更加合理的搜索对比流程, 使其在正确性指 标上向精确比对的指标靠拢, 提高了搜索准确性;
3. 本发明指紋搜索算法增加了对中心点的利用率, 使现场指紋特征中占 绝大多数的中心点在搜索比对中的优势被充分扩展。
4. 本发明指紋搜索算法中还增加了脊线宽度直方图方法, 进一步完善了 指紋搜索算法。
5. 通过本发明指紋搜索算法, 突破了联网下指紋数据膨胀引发的搜索速 度瓶颈, 给指紋识别设备往联网方向快速发展开辟了崭新空间。
附图说明 图 1 为本发明快速指紋搜索方法中指紋模板库和现场指紋特征中进行分 类处理的关系图;
图 2 为本发明快速指紋搜索方法中有中心点的现场指紋特征在进行分类 搜索时第一步的流程示意图;
图 3 为本发明快速指紋搜索方法中有中心点的现场指紋特征在进行分类 搜索时第二步的流程示意图;
图 4 为本发明快速指紋搜索方法中无中心点的现场指紋特征在进行分类 搜索时第一步的流程示意图;
图 5 为本发明快速指紋搜索方法中无中心点的现场指紋特征在进行分类 搜索时第二步的流程示意图;
图 6为本发明快速指紋搜索系统的一原理结构示意图;
图 7为本发明快速指紋搜索系统的另一原理结构示意图。
具体实施方式
以下结合附图, 具体说明本发明快速指紋搜索方法及快速指紋搜索系统。 本发明的指紋搜索主要通过分类分步的搜索,能快速筛选指紋模板库中数 据, 极大提高了运算速度。 在对本发明进行具体说明之前, 预先对后续谈到的 名词做一下解释说明, 以便能将本发明的实例说清楚。
1、 指紋特征: 一幅指紋图像经过处理后产生的能表示该指紋图像特征的 编码数据。
2、 指紋模板: 贮储在介质中用于指紋验证和搜索的指紋特征。
3、 中心点: 中心点是指紋最内层脊线的顶点, 它描述了指紋的整体特征。
4、 中心点快速比对: 根据中心点信息, 粗略计算指紋特征和指紋模板相 似度的过程。
5、 中心点精确比对: 根据中心点信息, 精确计算指紋特征和指紋模板相 似度, 并做出两者是否来自相同手指判断的过程。
6、 细节点快速比对: 根据细节点信息, 粗略计算指紋特征和指紋模板相 似度的过程。
7、 细节点精确比对: 根据细节点信息, 精确计算指紋特征和指紋模板相 似度, 并做出两者是否来自相同手指判断的过程。
本发明所述的所有指紋模板集合分为有中心点的指紋模板集合 Q和无中 心点的指紋模板集合为 T 。 现场指紋特征只存在有中心点或无中心点两种情 况。
请参阅图 1 , 其为本发明快速指紋搜索方法第一步的流程图。 它包括以下 步骤:
S010: 将指故模板库中的指故模板分为有中心点模板和无中心点模板。 指紋模板库釆集指紋模板时 ,需要将指紋分为有中心点模板和无中心点模 板。 其中, 整个模板库中绝大部分模板都为带中心点的模板。
以下是专利号 200610065297.5公开的一种指紋模板的处理方式。 各家公 司指紋模板处理方式可能会不相同, 本发明需要说明的是本实例可釆用 200610065297.5 公开的一种指紋模板的处理方式来处理, 但是并不是局限于 此, 此仅是为说明之用, 以避免审查员认为内容公开不充分之问题。
专利号 200610065297.5公开的一种指紋模板的处理方式为: 指紋模板库 釆集指紋模板时, 需要对指紋模板进行预处理和规格化, 计算分块方向图提取 奇异点, 计算方向图、 分割背景区域并细化奇异点; 图像滤波与增强; 计算脊 线密度; 二值化图像并细化, 提取细节点, 细节点验证, 删除伪细节点, 判断 是否有中心点, 根据 Poincare Index方法提取到原始中心点, 根据原始中心点 与方向图的关系进行后续处理, 去掉伪中心点等。 中心点信息包括位置、 方向 和质量信息。 指紋细节点、 奇异点、 平均脊密度和块方向图特征等最终被压缩 成为现场指紋特征或模板在指紋模板库中存储。
本发明还需要说明的是,指紋模板库可预先分为两个子库,有中心点模板 库和无中心点模板库。釆集到的指紋模板预处理和规格化后,根据其是否有中 心点特征来决定将指紋模板放置在哪一个库中进行保存。本发明还可以在每个 指紋模板的数据里设置一个用于表明有无中心点特征的标识, 比如, 若指紋模 板有中心点特征, 则该标识标为 "1", 若指紋模板无中心点特征, 则该标识标 为" 0"。
S020: 釆集需比对的指紋图案, 完成现场指紋特征提取, 并判断提取到 的所述现场指紋特征为有中心点现场指紋特征或无中心点现场指紋特征。
当某一用户需要进行身份认证时,现场釆集该用户的指紋, 即釆集需比对 的指紋图案, 完成现场指紋特征提取, 并判断提取到的现场指紋特征为有中心 点现场指紋特征或无中心点现场指紋特征, 其中, 绝大部分现场指紋特征都为 有中心点的。
现场指紋特征提取过程与步骤 S010 中涉及到的指紋模板的提取过程类 似, 在此就不再赘述, 这些都是现有技术, 也不再详述。
若需比对的现场指紋是有中心点特征的,则进入有中心点的现场指紋特征 的搜索流程,执行如图 2所示的处理步骤; 若需比对的现场指紋是无中心点特 征的, 则进入无中心点的现场指紋特征的搜索流程,执行如图 4所示的处理步 骤。 第一步
如图 2所示, 包括如下步骤:
S110: 若需比对的现场指紋是有中心点特征的,则将所述有中心点现场指 紋特征与指紋模板库中所有的有中心点模板分别进行脊线宽度直方图比对,从 中找到比对结果不小于预定脊线宽度直方图相似度阔值的有中心点模板,组成 集合 Qi。
判断需比对的现场指紋特征是否有中心点特征, 若有, 则进行步骤 S110, 若无,则进行图 4和图 5所示的无中心点搜索过程。本发明的一实例也可以是, 当需比对的现场指紋未有中心点特征时, 结束搜索过程, 这种实例即提供有中 心点特征的指紋比对。
脊线宽度直方图表示一幅指紋图像的脊线宽度分布情况的几何图形,图中 横坐标表示脊线宽度值, 纵坐标表示该脊线宽度值出现的次数。脊线宽度直方 图的提取是现有的公知技术, 这里不再详述。
脊线宽度直方图相似度计算公式如下: 模板和特征的脊线宽度直方图定义为离散函数 H ¾ ) =!¾ 、¾( 其中, 是第 i级脊线宽度, !!^和!^表示脊线宽度等级为 的像素个数。 它 们分别记录在模板数据和特征数据中,则模板和特征的脊线宽度直方图相似度 为:
Figure imgf000011_0001
^imn = a ' e SimH的值在 0和 100之间。 本发明定义脊线宽度直方图相似度阔值,用于判定现场指紋特征和指紋模 板的直方图是否相似的分值阔值;
脊线宽度直方图相似度阔值是预先设定的,也可以是技术人员根据现场情 况选择确定的, 脊线宽度直方图相似度阔值是数值。
另夕卜,本实例中将找到的比对结果不小于预定脊线宽度直方图相似度阔值 的有中心点模板, 组成集合 Ql 本发明可以根据比对结果, 将指紋模板放置 在两个不同的集合中。
本发明也可以预先在每个指紋模板的数据里设置一个用于表明比对结果 不小于还是小于预定脊线宽度直方图相似度阔值的标识, 比如, 若指紋模板有 比对结果不小于预定脊线宽度直方图相似度阔值, 则该标识表标为 "1", 若指 紋模板有比对结果小于预定脊线宽度直方图相似度阔值,则该标识表标为 "0"。 每次需比对的指紋与指紋模板进行脊线宽度直方图比对时,修改该指紋模板的 该对应的标识。
S120: 将有中心点现场指紋特征与集合 中的所述模板进行中心点快速 比对, 找到相似度最高的前!^个模板组成集合 Ml ηι为自然数。
S130: 将所述有中心点现场指紋特征与集合 中的所有模板进行中心点 精确比对,找到精确比对成功的模板作为所述有中心点现场指紋特征对应的匹 配指紋。
本发明定义精确比对得分阔值为:判定现场特征指紋和模板是否来自相同 手指的得分阔值。 其中, 当现场指紋特征与模板的精确比对得分不小于预设的 精确比对得分阔值, 则现场指紋特征和模板的指紋来自相同手指; 当现场指紋 特征与模板的精确比对得分小于预设的精确比对得分阔值,则现场指紋特征和 模板的指紋来自不同手指。
中心点快速比对算法、中心点精确比对算法可以是公知的算法,在此就不 再详说。
将有中心点现场指紋特征与集合 中的所有模板进行中心点精确比对, 并计算相似度得分。 一旦出现相似度得分不小于给定精确比对得分阔值的模 板, 则认为该特征通过指紋验证, 结束搜索过程。
当所述有中心点现场指紋特征与集合 中的所有模板进行中心点精确比 对后, 未找到精确比对成功的模板, 则进行第二步的搜索过程, 如图 3所示。 但是, 绝大多数有中心点的现场指紋特征将在这儿结束搜索过程, 并得到正确 的模板。 因此, 若对精确度要求不高的情况下, 可只进行图 2所示的步骤流 程。
第二步
当所述有中心点现场指紋特征与集合 中的所有模板进行中心点精确比 对后, 未找到精确比对成功的模板, 执行图 3所示的步骤流程:
S210: 将有中心点现场指紋特征与指紋模板库中所有无中心点模板分别 进行脊线宽度直方图比对,从中找到比对结果不小于预定脊线宽度直方图相似 度阔值的模板, 组成集合 Ί\。
S220: 将有中心点现场指紋特征与集合 Ί\中的所述模板进行细节点快速 比对, 找到相似度最高的前 个模板组成集合 Μ2, η2为自然数。
S230: 将所述有中心点现场指紋特征与集合 Μ2中的所有模板进行细节点 精确比对, 并计算相似度得分, 一旦出现相似度得分不小于给定精确比对得分 阔值的模板, 则认为该有中心点现场指紋特征通过指紋验证, 结束搜索过程, 否则判为非法现场指紋特征。
关于细节点快速比对算法和细节点精确比对算法是现有技术,本发明就不 再赘述。 若需比对的现场指紋特征是无中心点现场指紋特征,则执行图 4所示的步 骤流程:
第一步
S310: 将无中心点现场指紋特征与指紋模板库中所有无中心点模板分别 进行脊线宽度直方图比对,从中找到比对结果不小于预定脊线宽度直方图相似 度阔值的模板, 组成集合 T2
S320: 将无中心点现场指紋特征与集合 Τ2中的所述模板进行细节点快速 比对, 找到相似度最高的前 η3个模板组成集合 Μ3; η3为自然数。
S330: 将无中心点现场指紋特征与集合 Μ3中的所有模板进行细节点精确 比对, 找到精确比对成功的模板作为所述需比对指紋的匹配指紋。
对于无中心点的现场指紋特征,其绝大多数将在这儿结束搜索过程,并得 到正确的模板。 因此, 若对精确度要求不高的情况下, 对无中心点的现场指紋 特征可只进行到该步骤。
当将所述无中心点现场指紋特征与集合 Μ3中的所有模板进行细节点精确 比对, 未找到精确比对成功的模板时, 则可执行图 5所示的步骤流程。 第二步
S410: 将无中心点现场指紋特征与集合 Μ3中的所有模板进行细节点精确 比对, 未找到精确比对成功的模板时, 则将无中心点现场指紋特征与指紋模板 库中所有有中心点模板分别进行脊线宽度直方图比对,从中找到比对结果不小 于预定脊线宽度直方图相似度阔值的模板, 组成集合 Q2
S420: 将所述无中心点现场指紋特征与集合 Q2中的模板分别进行细节点 快速比对, 计算相似度得分, 得到相似度得分最高的前 个模板集合 M4 ; 为自然数。
S430: 将所述无中心点现场指紋特征与集合 M4中的所有模板进行细节点 精确比对, 并计算相似度得分, 一旦出现相似度得分不小于给定精确比对得分 阔值的模板, 则认为该无中心点现场指紋特征通过指紋验证, 结束搜索过程, 否则, 判为非法现场指紋特征。 本发明所述的指紋搜索算法通过对现有指紋模板搜索方法进行策略性的 改进措施, 尤其是分类和分步的搜索流程, 能快速筛选指紋模板库里的数据, 极大的提高了运算速度。
请参阅图 6, 其为本发明第一种快速指紋搜索系统的原理结构示意图。 它 包括若干包括指紋釆集器的终端 11和指紋处理子系统 12 , 终端 11通过专线 或网络连接至指紋处理子系统 12。
指紋处理子系统 12进一步包括数据库 121和服务器 122:
所述数据库 121用于存储指故模板库, 其中,指故模板库中的指故模板分 为有中心点模板和无中心点模板。 所述数据库 121至少包括: 用于存储指紋模 板库的指紋模板库存储单元和包括保存处理流程的流程存储单元。
所述服务器 122 用于判断终端提取到的现场指紋特征为有中心点现场指 紋特征或无中心点现场指紋特征,如果为有中心点现场指紋特征, 则将所述现 场指紋特征与指紋模板库中所有有中心点模板分别进行比对;如果为无中心点 现场指紋特征,则将所述现场指紋特征与指紋模板库中所有无中心点模板分别 进行比对。
所述服务器 122还用于:当所述现场指紋特征与所述所有有中心点模板进 行比对的结果为不存在比对成功的模板时,将所述现场指紋特征与指紋模板库 中所有无中心点模板分别进行比对;当所述现场指紋特征与所述所有无中心点 模板进行比对的结果为不存在比对成功的模板时,将所述现场指紋特征与指紋 模板库中所有有中心点模板分别进行比对。
所述服务器 122进一步包括:
脊线宽度直方图比对装置:用于将所述现场指紋特征与所有指紋模板库中 预先设定范围的模板分别进行脊线宽度直方图比对,从中找到比对结果不小于 预定脊线宽度直方图相似度阔值的模板组成集合;
中心点快速比对装置:用于将所述现场指紋特征与预先设定范围的所有模 板进行中心点快速比对, 找到相似度最高的前预先设定个数的模板组成集合; 中心点精确比对装置:用于将所述现场指紋特征与预先设定范围的所有模 板进行中心点精确比对,找到精确比对成功的模板作为所述现场指紋的匹配指 紋。
所述服务器 122还可以再包括: 细节点快速比对装置: 用于将所述现场指 紋特征与预先设定范围的所有模板进行细节点快速比对,找到相似度最高的前 预先设定个数的模板组成集合;
细节点精确比对装置:用于将所述现场指紋特征与预先设定范围的所有模 板进行细节点精确比对,找到精确比对成功的模板作为所述现场指紋的匹配指 紋。
脊线宽度直方图比对装置、 中心点快速比对装置、 中心点精确比对装置、 细节点快速比对装置和细节点精确比对装置为逻辑单元。
所述终端 11 , 用于釆集需比对的指紋图案, 完成现场指紋特征提取。 终 端 11进一步包括:
指紋釆集装置: 用于釆集要比对的现场指紋图像;
特征点提取装置: 它与指紋釆集装置相连,用于对釆集到的现场指紋图像 提取特征点。指紋釆集装置通常为指紋釆集器,特征点提取装置通常是处理器 通过软件的方式来实现, 其为逻辑单元。
请参阅图 7 , 其为本发明另一种快速指紋搜索系统的原理结构示意图。 它 包括:
指紋模板库存储单元 21 : 用于存储指紋模板库, 其中指紋模板都按有无 中心点进行分类;
指紋釆集装置 22: 用于釆集要比对的指紋图像;
特征点提取装置 23 : 它与指紋釆集装置相连, 用于对釆集到的指紋图像 提取特征点;
特征判断单元 29: 用于判断特征点提取装置 23提取到的现场指紋特征为 有中心点现场指紋特征或无中心点现场指紋特征;
指紋比对单元 30: 用于根据特征判断单元 29的判断结果, 如果现场指紋 特征为有中心点现场指紋特征 ,则将所述现场指紋特征与指紋模板库存储单元 21 所存储的指紋模板库中所有有中心点模板分别进行比对; 如果为无中心点 现场指紋特征, 则将所述现场指紋特征与指紋模板库存储单元 21所存储的指 紋模板库中所有无中心点模板分别进行比对。
具体的, 指紋比对单元 30进一步包括:
脊线宽度直方图比对装置 24: 用于将所述现场指紋特征与所有指紋模板 库中预先设定范围的模板分别进行脊线宽度直方图比对,从中找到比对结果不 小于预定脊线宽度直方图相似度阔值的模板组成集合;
中心点快速比对装置 25: 用于将所述现场指紋特征与将预先设定范围的 所有模板进行中心点快速比对,找到相似度最高的前预先设定个数的模板组成 中心点精确比对装置 26: 用于将所述现场指紋特征与预先设定范围的所 有模板进行中心点精确比对,找到精确比对成功的模板作为所述现场指紋的匹 配指紋。
指紋比对单元 30还可以再包括: 细节点快速比对装置 27: 用于将所述现 场指紋特征与预先设定范围的所有模板进行细节点快速比对 ,找到相似度最高 的前预先设定个数模板组成集合;
细节点精确比对装置 28: 用于将所述现场指紋特征与预先设定范围的所 有模板进行细节点精确比对,找到精确比对成功的模板作为所述现场指紋的匹 配指紋。
本实例还可以有一个保存处理流程的流程存储单元。
以上公开的仅为本发明的几个具体实施例,但本发明并非局限于此,任何 本领域的技术人员能思之的变化, 都应落在本发明的保护范围内。

Claims

权 利 要 求
1、 一种快速指紋搜索方法, 其特征在于, 将指紋模板库中的指紋模板分 为有中心点模板和无中心点模板, 该方法包括:
釆集需比对的指紋图案, 完成现场指紋特征提取;
判断提取到的现场指紋特征为有中心点现场指紋特征或无中心点现场指 紋特征,如果为有中心点现场指紋特征,将所述现场指紋特征与指紋模板库中 所有有中心点模板分别进行比对; 如果为无中心点现场指紋特征, 则将所述现 场指紋特征与指紋模板库中所有无中心点模板分别进行比对。
2、 根据权利要求 1所述的方法, 其特征在于, 当所述现场指紋特征与所 述所有有中心点模板进行比对的结果为不存在比对成功的模板时,该方法进一 步包括:
将所述现场指紋特征与指紋模板库中所有无中心点模板分别进行比对。
3、 根据权利要求 1所述的方法, 其特征在于, 当所述现场指紋特征与所 述所有无中心点模板进行比对的结果为不存在比对成功的模板时,该方法进一 步包括:
将所述现场指紋特征与指紋模板库中所有有中心点模板分别进行比对。
4、 根据权利要求 1至 3任一项所述的方法, 其特征在于, 该方法进一步 包括:
所述现场指紋特征与所述模板的比对结果为存在比对成功的模板时,确定 所述现场指紋特征通过指紋验证; 否则, 确定所述现场指紋特征为非法指紋。
5、 根据权利要求 1至 3任一项所述的方法, 其特征在于, 所述将所述现 场指紋特征与指紋模板库中所有有中心点模板分别进行比对具体为:
将所述现场指紋特征与指紋模板库中所有有中心点模板分别进行脊线宽 度直方图比对,从中找到比对结果不小于预定脊线宽度直方图相似度阔值的模 板, 组成集合 Qx;
将所述现场指紋特征与所述集合 Qx中的所述模板分别进行中心点快速比 对, 获取相似度最高的前 nx个模板组成集合 Mx, nx为自然数;
将所述现场指紋特征与集合 Mx中的所有模板分别进行中心点精确比对。
6、 根据权利要求 1至 3任一项所述的方法, 其特征在于, 所述将所述现 场指紋特征与指紋模板库中所有无中心点模板分别进行比对具体为:
将所述现场指紋特征与指紋模板库中所有无中心点模板分别进行脊线宽 度直方图比对,从中找到比对结果不小于预定脊线宽度直方图相似度阔值的模 板, 组成集合 Ty;
将所述有中心点现场指紋特征与所述集合 Ty中的所述模板分别进行细节 点快速比对, 找到相似度最高的前 ny个模板组成集合 My, ny为自然数; 将所述有中心点现场指紋特征与集合 My中的所有模板分别进行细节点精 确比对。
7、 一种快速指紋搜索系统, 其特征在于, 包括:
指紋模板库存储单元, 用于存储指紋模板库, 其中, 指紋模板库中的指紋 模板分为有中心点模板和无中心点模板;
指紋釆集装置, 用于釆集需比对的现场指紋图案;
特征点提取装置, 用于对指紋釆集装置釆集到的现场指紋图案提取特征 点;
特征判断单元,用于判断特征点提取装置提取到的现场指紋特征为有中心 点现场指紋特征或无中心点现场指紋特征;
指紋比对单元, 用于根据特征判断单元的判断结果,如果现场指紋特征为 有中心点现场指紋特征 ,则将所述现场指紋特征与指紋模板库存储单元所存储 的指紋模板库中所有有中心点模板分别进行比对;如果为无中心点现场指紋特 征 ,则将所述现场指紋特征与指紋模板库存储单元所存储的指紋模板库中所有 无中心点模板分别进行比对。
8、 根据权利要求 7所述的系统, 其特征在于, 指紋比对单元还用于: 当所述现场指紋特征与所述所有有中心点模板进行比对的结果为不存在 比对成功的模板时,将所述现场指紋特征与指紋模板库中所有无中心点模板分 别进行比对;
当所述现场指紋特征与所述所有无中心点模板进行比对的结果为不存在 比对成功的模板时,将所述现场指紋特征与指紋模板库中所有有中心点模板分 别进行比对。
9、 根据权利要求 7或 8所述的系统, 其特征在于, 指紋比对单元包括: 脊线宽度直方图比对装置,用于将所述现场指紋特征与所有指紋模板库中 预先设定范围的模板分别进行脊线宽度直方图比对,从中找到比对结果不小于 预定脊线宽度直方图相似度阔值的模板组成集合;
中心点快速比对装置,用于将所述现场指紋特征与预先设定范围的所有模 板进行中心点快速比对, 找到相似度最高的前预先设定个数的模板组成集合; 中心点精确比对装置 ,用于将所述现场指紋特征与预先设定范围的所有模 板进行中心点精确比对,寻找精确比对成功的模板作为所述现场指紋的匹配指 紋;
细节点快速比对装置,用于将所述现场指紋特征与将预先设定范围的所有 模板进行细节点快速比对, 找到相似度最高的前设定个数的模板组成集合; 细节点精确比对装置 ,用于将所述现场指紋特征与预先设定范围的所有模 板进行细节点精确比对,寻找精确比对成功的模板作为所述现场指紋的匹配指 紋。
10、 一种快速指紋搜索系统, 其特征在于, 包括: 指紋处理子系统和若干 终端, 所述终端通过专线或网络连接至指紋处理子系统, 指紋处理子系统进一 步包括数据库和服务器; 其中,
终端, 用于釆集需比对的指紋图案, 完成现场指紋特征提取;
数据库, 用于存储指乡文模板库, 其中, 指故模板库中的指故模板分为有中 心点模板和无中心点模板;
服务器,用于判断终端提取到的现场指紋特征为有中心点现场指紋特征或 无中心点现场指紋特征, 如果为有中心点现场指紋特征, 则将所述现场指紋特 征与指紋模板库中所有有中心点模板分别进行比对;如果为无中心点现场指紋 特征, 则将所述现场指紋特征与指紋模板库中所有无中心点模板分别进行比 对。
11、 根据权利要求 10所述的系统, 其特征在于, 所述服务器还用于: 当所述现场指紋特征与所述所有有中心点模板进行比对的结果为不存在 比对成功的模板时,将所述现场指紋特征与指紋模板库中所有无中心点模板分 别进行比对;
当所述现场指紋特征与所述所有无中心点模板进行比对的结果为不存在 比对成功的模板时,将所述现场指紋特征与指紋模板库中所有有中心点模板分 别进行比对。
PCT/CN2009/071257 2008-12-19 2009-04-14 快速指纹搜索方法及快速指纹搜索系统 WO2010069166A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
EP09832848A EP2360619A4 (en) 2008-12-19 2009-04-14 FAST FINGERPRINT SEARCH PROCESS AND FAST FINGERPRINT SEARCH SYSTEM

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN2008102074254A CN101751550B (zh) 2008-12-19 2008-12-19 快速指纹搜索方法及快速指纹搜索系统
CN200810207425.4 2008-12-19

Publications (1)

Publication Number Publication Date
WO2010069166A1 true WO2010069166A1 (zh) 2010-06-24

Family

ID=42268284

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2009/071257 WO2010069166A1 (zh) 2008-12-19 2009-04-14 快速指纹搜索方法及快速指纹搜索系统

Country Status (3)

Country Link
EP (1) EP2360619A4 (zh)
CN (1) CN101751550B (zh)
WO (1) WO2010069166A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102999476A (zh) * 2011-09-09 2013-03-27 中国航天科工集团第三研究院第八三五七研究所 一种二维矩阵匹配方法及电路
WO2017067269A1 (zh) * 2015-10-19 2017-04-27 广东欧珀移动通信有限公司 一种指纹的识别方法、装置以及移动终端

Families Citing this family (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101996318B (zh) * 2010-11-25 2013-01-30 杭州晟元芯片技术有限公司 一种快速计算指纹相似度的方法
CN106326819B (zh) * 2015-06-30 2019-11-12 宇龙计算机通信科技(深圳)有限公司 一种指纹识别方法、装置和终端
CN106055958B (zh) * 2016-05-31 2019-02-05 Oppo广东移动通信有限公司 一种解锁方法及装置
CN106096361A (zh) * 2016-05-31 2016-11-09 广东欧珀移动通信有限公司 一种指纹解锁方法及移动终端
CN107545215A (zh) * 2016-06-23 2018-01-05 杭州海康威视数字技术股份有限公司 一种指纹识别方法及装置
CN106548133B (zh) * 2016-10-17 2019-04-23 歌尔科技有限公司 一种模板匹配方法和装置以及手势识别方法和装置
CN108304289B (zh) * 2018-02-07 2020-09-01 深圳壹账通智能科技有限公司 终端指纹测试方法、装置、计算机设备及存储介质
WO2020172773A1 (zh) * 2019-02-25 2020-09-03 深圳市汇顶科技股份有限公司 指纹识别的方法、装置和电子设备
CN117132925B (zh) * 2023-10-26 2024-02-06 成都索贝数码科技股份有限公司 一种体育赛事的智能场记方法及装置

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1506903A (zh) * 2002-12-06 2004-06-23 中国科学院自动化研究所 基于模板学习的自动指纹识别系统和方法
CN1521684A (zh) * 2003-02-13 2004-08-18 合肥工业大学 协同指纹识别方法
WO2006062378A1 (en) * 2004-12-09 2006-06-15 Lg Electronics Inc. Fingerprint recognition method

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
AU2987700A (en) * 1999-02-08 2000-08-25 Amy S. Zelson Fingerprint analysis method
JP4405656B2 (ja) * 2000-10-20 2010-01-27 富士通株式会社 指紋情報を用いた個人認証システム及び同システム用登録・認証方法
US20070248249A1 (en) * 2006-04-20 2007-10-25 Bioscrypt Inc. Fingerprint identification system for access control

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1506903A (zh) * 2002-12-06 2004-06-23 中国科学院自动化研究所 基于模板学习的自动指纹识别系统和方法
CN1521684A (zh) * 2003-02-13 2004-08-18 合肥工业大学 协同指纹识别方法
WO2006062378A1 (en) * 2004-12-09 2006-06-15 Lg Electronics Inc. Fingerprint recognition method

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
See also references of EP2360619A4
ZHANG YING: "Research of Fingerprint Recognition Algorithm Based on Large-Scale", CNKI CHINA MASTER'S THESES FULL-TEXT DATABASE, 19 November 2008 (2008-11-19), pages 6 - 7,24-29,37-38, XP008155130 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102999476A (zh) * 2011-09-09 2013-03-27 中国航天科工集团第三研究院第八三五七研究所 一种二维矩阵匹配方法及电路
WO2017067269A1 (zh) * 2015-10-19 2017-04-27 广东欧珀移动通信有限公司 一种指纹的识别方法、装置以及移动终端

Also Published As

Publication number Publication date
CN101751550A (zh) 2010-06-23
CN101751550B (zh) 2012-02-01
EP2360619A1 (en) 2011-08-24
EP2360619A4 (en) 2012-05-30

Similar Documents

Publication Publication Date Title
WO2010069166A1 (zh) 快速指纹搜索方法及快速指纹搜索系统
US20100080425A1 (en) Minutiae-based template synthesis and matching
CN100414558C (zh) 基于模板学习的自动指纹识别系统和方法
Hemalatha A systematic review on Fingerprint based Biometric Authentication System
CN109934114B (zh) 一种手指静脉模板生成与更新算法及系统
Zanganeh et al. Partial fingerprint matching through region-based similarity
US9251398B2 (en) Method of matching, biometric matching apparatus, and computer program
WO2020070821A1 (ja) 生体認証装置、生体認証方法、及び生体認証プログラム
CN111178203B (zh) 签名审核方法、装置、计算机设备和存储介质
George et al. An efficient system for palm print recognition using ridges
Kong et al. A hierarchical classification method for finger knuckle print recognition
TWI325568B (en) A method for face varification
KR101639404B1 (ko) 지문 분할 인식 기반의 사용자 인증 장치 및 방법
Kanjan et al. A comparative study of fingerprint matching algorithms
CN111723595A (zh) 人员身份识别方法和系统
Agrawal et al. A hybrid partial fingerprint matching algorithm for estimation of equal error rate
Seng et al. An improved fingerprint image matching and multi-view fingerprint recognition algorithm.
Hendre et al. Utility of quality metrics in partial fingerprint recognition
Jeong et al. Effect of Smaller Fingerprint Sensors on the Security of Fingerprint Authentication
Agarwal et al. A utility of ridge contour points in minutiae-based fingerprint matching
Jin et al. Comparative assessment of fingerprint sample quality measures based on minutiae-based matching performance
Ayodele et al. Current practices in information fusion for multimodal biometrics
Sharma et al. Fingerprint matching Using Minutiae Extraction Techniques
CN116959075B (zh) 基于深度学习的身份识别机器人迭代优化方法
Cao et al. Fingerprint matching based on neighboring information and penalized logistic regression

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: 09832848

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 2009832848

Country of ref document: EP

NENP Non-entry into the national phase

Ref country code: DE