WO2019153739A1 - 基于人脸识别的身份认证方法、装置、设备和存储介质 - Google Patents

基于人脸识别的身份认证方法、装置、设备和存储介质 Download PDF

Info

Publication number
WO2019153739A1
WO2019153739A1 PCT/CN2018/105173 CN2018105173W WO2019153739A1 WO 2019153739 A1 WO2019153739 A1 WO 2019153739A1 CN 2018105173 W CN2018105173 W CN 2018105173W WO 2019153739 A1 WO2019153739 A1 WO 2019153739A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
information
card
classified
precise positioning
Prior art date
Application number
PCT/CN2018/105173
Other languages
English (en)
French (fr)
Inventor
瞿康敏
洪光宝
Original Assignee
深圳壹账通智能科技有限公司
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 深圳壹账通智能科技有限公司 filed Critical 深圳壹账通智能科技有限公司
Publication of WO2019153739A1 publication Critical patent/WO2019153739A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q40/00Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/03Credit; Loans; Processing thereof
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q40/00Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/02Banking, e.g. interest calculation or account maintenance
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/20Image preprocessing
    • G06V10/26Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G06V10/267Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/40Document-oriented image-based pattern recognition
    • 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
    • G06V40/16Human faces, e.g. facial parts, sketches or expressions
    • G06V40/172Classification, e.g. identification
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q40/00Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/08Insurance

Definitions

  • the present application relates to the field of information security, and in particular, to an identity authentication method, apparatus, device, and storage medium based on face recognition.
  • online credit is also becoming a trend. It can complete the steps of loan application without leaving home, including understanding the application conditions of various types of loans, preparing application materials, and submitting loan applications. Efficient completion on the Internet.
  • the verification of the customer's identity cannot be provided by the customer with the identity document as the traditional business, and the business management personnel perform the approximate comparison by the naked eye.
  • Face recognition based on face recognition becomes inevitable.
  • the prior art has the following disadvantages: the face recognition rate is low, and the face shape is unstable, especially when the face sample is covered by facial expressions, external illumination, and face coverage.
  • the object and other factors affect the face recognition it will increase the difficulty of face recognition.
  • ID card for identity verification it is impossible to quickly filter the personal information in the ID card and read the text information quickly, and the comparison recognition speed is slow.
  • a method for identity authentication based on face recognition comprising: acquiring an image captured by a client camera in real time, obtaining a facial image from the image; extracting identity card information, capturing a front image of the ID card and a reverse image of the ID card, and identifying the identity
  • the front image and the reverse image of the ID card are used for row positioning, binarization, noise removal, tilt correction, layout analysis, and character segmentation to obtain valid text information of a single ID card;
  • the valid text information of the ID card is by name,
  • the address, the nationality, and the ID number are classified, and the classified ID information is output;
  • the facial image and the classified ID information are compared and verified by the network, and the network comparison check is to check the connection citizenship information.
  • the system verifies the facial image and the classified ID card information, and determines the consistency thereof; when the networked verification is performed through the classified ID card information, user information exists; and the facial image and the network are connected When the image of the returned ID face is the same, the ID card information is true; otherwise, the ID card information is false; Check the results and output.
  • An identity recognition device based on face recognition comprising: a face acquisition unit configured to acquire an image captured by a client camera in real time, obtain a face image from the image; and an identity information acquisition unit configured to extract identity card information Capture the front image of the ID card and the reverse image of the ID card, and perform line positioning, binarization, noise removal, tilt correction, layout analysis, character segmentation on the front image of the ID card and the reverse image of the ID card to obtain valid text for the single ID card.
  • the identity information categorizing unit is configured to classify the valid text information of the ID card by name, address, nationality, and ID number, and output the classified ID information
  • the comparison verification unit is set to The face image and the classified ID card information are checked for network comparison, and the networked comparison check verifies the facial image and the classified ID card information for the connected citizenship information verification system, and determines that the same Sex; when the network check is performed through the classified ID information, user information exists; and the face When the same image as the face and the return of people online verification ID card, ID information is true; otherwise, false identity information
  • result output unit configured to obtain comparative results of the verification and output.
  • a computer device comprising a memory and a processor, the memory storing computer readable instructions, the computer readable instructions being executed by the processor, causing the processor to perform the above-described face recognition based identity authentication The steps of the method.
  • a storage medium storing computer readable instructions that, when executed by one or more processors, cause one or more processors to perform the steps of the above-described face recognition based identity authentication method.
  • the above-mentioned face recognition-based identity authentication method, device, device and storage medium acquire the image captured by the client camera in real time, obtain a facial image from the image, extract the identity card information, and capture the front image and the ID card of the ID card.
  • the text information is classified according to the name, address, ethnicity, and ID number, and the classified ID information is output; the facial image and the classified ID information are compared and verified by the network, and the network comparison check is
  • the connection citizenship information verification system verifies the facial image and the classified ID card information, and determines the consistency thereof; when the networked verification is performed through the classified ID card information, the user information exists, and the When the face image is consistent with the ID face image returned by the network check, the ID card information is true; otherwise, the identity The information is false;
  • FIG. 1 is a flowchart of a method for identity authentication based on face recognition in an embodiment of the present application
  • FIG. 2 is a method for performing line positioning, binarization, noise removal, tilt correction, layout analysis, and character segmentation on the front image of the ID card and the reverse image of the ID card in one embodiment of the present application, and obtaining a valid text information method for a single ID card. flow chart;
  • FIG. 3 is a structural block diagram of an identity recognition device based on face recognition according to an embodiment of the present application.
  • an identity authentication method based on face recognition is proposed.
  • the method for identifying a face recognition based identity may specifically include:
  • the terminal device of the client may be a mobile phone, an IPAD, etc.
  • the cameras of these devices collect images in real time, and the processor obtains the image from the camera in real time, where the processor obtains an image that is not a photo taken by the camera, but only a real-time image in the image area of the camera.
  • the real-time image such as facial contour data and facial feature data, including eyes, nose, and mouth, but no specific recognition of the face is required, this step only needs to recognize the face from the background. Easy for the next step.
  • Frame calibration of the client's camera that is, set a red border of 580*580 at the center position of each frame captured by the camera, which simplifies the algorithm, speeds up the processing progress, and improves the recognition efficiency due to the front image of the ID card.
  • Including face information a large number of face images are selected, and these face images are marked as positive samples, non-face images are labeled as negative samples, and the face image and identity of the user's ID card are distinguished by detecting whether the face image is included in the model.
  • the negative image of the card that is, the face image can be detected as the front image of the ID card, and vice versa.
  • the problem of extracting the positive information of the ID card is solved simply and effectively, and the positive image of the ID card and the reverse image of the ID card are respectively processed to obtain the valid text of the ID card. information.
  • the location of the ID card information in the ID card image is fixed. Therefore, the identification of the positive image of the ID card and the reverse image of the ID card are used to identify the positive image of the ID card and the valid image in the reverse image of the ID card. Preprocessing such as removal, tilt correction, layout analysis, character segmentation, etc.
  • Image binarization is to set the gray value of the pixel on the image to 0 or 255, that is, the process of presenting the entire image with a distinct black and white effect.
  • the binarization of the image greatly reduces the amount of data in the image, which can highlight the contour of the target, which is beneficial for further processing of the image, and then the image is subjected to rotational tilt correction, image denoising, image enhancement.
  • Such as pre-processing, layout analysis, information column positioning, line segmentation and character segmentation Chinese characters consist of a stroke, each stroke is composed of pixels in some connected domains, the connected domain of the square Chinese characters is Characteristic, when the inter-character strokes are stuck, multiple characters form a connected domain, which needs to be divided according to the average word width.
  • S103 classify the valid text information of the ID card by name, address, nationality, and ID number, and output the classified ID information;
  • the ID card information is identified and classified according to the name, address, ethnicity, and ID number, and the corresponding fields in the population database of the Public Security Bureau are compared on the condition of the classified ID card information for comparison.
  • S104 Perform network comparison check on the facial image and the classified ID card information, where the networked comparison check is performed by the connected citizenship information verification system on the facial image and the classified ID information. Verification, judging the consistency; when the networked verification is performed by the classified ID information, the user information exists; and the facial image is consistent with the ID image of the ID returned by the network verification, the ID information is true; Otherwise, the ID card information is false;
  • the name and ID card number are automatically entered into the citizenship information network verification system. If the name and ID number exist in the library, the system returns the name and identity.
  • the certificate information with the same number of licenses includes: name, ID number, address, ethnicity, date of birth, face photo, etc.
  • the face recognition technology is used to compare the face image on the ID card with the face image of the ID card in the online verification to determine the consistency.
  • the check result is checked, and the check type includes the response result check, the database result check, and the log result check, obtain interface data corresponding to the check type, and match the interface data with the preset correction result data to obtain Whether the matching verification result is completed, the system can automatically complete the verification of the result, improve the efficiency of the result verification and reduce the false detection rate, and output the result of the verification.
  • the method before acquiring the image captured by the client camera in real time, the method further comprises: performing pre-detection on the captured face pose, angle, illumination, and occlusion influence factors, and issuing a prompt to enable the user to cooperate. .
  • the video is captured by the camera, and then the face is positioned to perform a live detection on the face, and the scene pre-detection of the captured face pose, angle, illumination, and occlusion factors is performed, and the voice prompter and the display prompt are prompted. User cooperation.
  • the method before acquiring the image captured by the client camera in real time, the method further includes: receiving, by the output adjustment button of the client, the first instruction input to adjust the predetermined area, and adjusting the predetermined area according to the first instruction.
  • the first command is a control command of the photographer, for example, the photographer clicks the adjustment knob of the terminal device to perform adjustment of the predetermined area, and the adjustment of the predetermined area includes size adjustment and position adjustment, and when the adjustment knob is used for the display image of the terminal device
  • the photographer can directly drag the ring structure on the display screen for position adjustment and size adjustment.
  • the predetermined area can be re-determined according to the above adjustment of the photographer, and the subsequent operations are performed. follow the newly created reservation area.
  • S101 further includes: detecting a position of the facial image in the image, and prompting the face to return to the predetermined area by issuing a prompt when the facial image is located outside the predetermined area in the image.
  • the predetermined area is preset, and it can be passively set. For example, if the default is 3 ⁇ 3 inches in the middle of the screen of the terminal mobile device, the area is predetermined, and the size and area of the user can be actively set.
  • the face is recognized from the image, and it is detected whether the face is located in a predetermined area in the image, when it is outside the predetermined area, but when it is detected that the face is not located in the predetermined area, that is, control
  • the rotary drive mechanism drives the handset and prompts the face to return to the predetermined area by issuing a prompt.
  • Face recognition is a common knowledge and common technical means in the field of communication, such as artificial neural network method, support vector machine method, PCA recognition, elastic matching method, feature face method based on KL transform, integral image feature method, etc. The examples are not described one by one.
  • the valid text information of the card includes:
  • S201 Perform rough positioning on the valid information according to the position of the valid text information in the ID card
  • the effective information can be roughly positioned according to the position of the valid text information in the ID card, that is, the coarse positioning, for example, if the width and height of the entire ID card image are 0.9, The origin is in the upper left corner, and the ID number is generally located in the upper left corner (0.38, 0.83), the width is 0.48, and the height is 0.145.
  • the valid text information of the ID card is obtained.
  • S202 Perform precise positioning on the coarsely positioned text information according to the characteristic that the gray level average of the pixels of the line where the boundary of the valid text information line is smaller than the background, and generate a pixel matrix to be filtered, and the pixel matrix is accurately positioned.
  • the pixel points in each single-word image area are processed separately, the useful information is extracted, the seed origin is selected according to the partial features of the pixel points, and the region growing algorithm based on the pixel point distribution feature is used from the seed origin to gradually advance and read to the adjacent pixel points. Take until all pixels have been processed.
  • the region growing algorithm based on the distribution feature of pixel points traverses the pixel points to obtain the character structure information, and identifies the determined points. Some of the relatively dark or relatively white points in the image are easily considered as foreground or background points. In order to determine the point, the determination point needs to be identified first in the initialization phase to reduce the computational time.
  • T is the main gradation level, that is, the maximum gradation value corresponding to the grading operation
  • r is the empirical value
  • r 0.6
  • the lower boundary LT T*r
  • HT 255-(255-T)*r
  • the point where the gray value is less than or equal to LT is a relatively dark point, marked as the front spot
  • the point where the gray value is greater than or equal to the upper boundary HT is a relatively white point, marked as a background point, to prevent a certain difference
  • the background line is regarded as the boundary of precise positioning, and the upper boundary, the left boundary and the right boundary of the precisely positioned area are obtained in the same way.
  • Gaussian filtering is performed on the precise positioning area by using a Gaussian filter template of 3 pixels ⁇ 3 pixels, wherein the Gaussian filtering is performed by using a template to scan each pixel in the image, and the weighted average gray value of the pixels in the neighborhood determined by the template is used. Replacing the value of the center pixel of the template, reducing the interference of background information and removing noise;
  • Gaussian filtering is performed to attenuate the interference of background information and remove noise.
  • Gaussian filtering is a process of weighted averaging of the entire image. The value of each pixel is obtained by weighted averaging of itself and other pixel values in the neighborhood. The specific operation of Gaussian filtering is to scan each pixel in the image with a template (or convolution, mask), and replace the value of the center pixel of the template with the weighted average gray value of the pixels in the neighborhood determined by the template.
  • S204 Using a dynamic threshold to binarize the precise positioning area, obtain a binary image of the precise positioning area, and perform single word segmentation on the binary image of the precise positioning area to obtain a single valid text information.
  • the dynamic binarization threshold of the image pixel is used to binarize the smoothed grayscale image, and further denoising improves the accuracy of the later matching.
  • an identity recognition device based on face recognition includes:
  • a face acquisition unit configured to acquire an image acquired by a client camera in real time, and obtain a facial image from the image
  • the identity information obtaining unit is configured to extract the identity card information, capture the front image of the ID card and the reverse image of the ID card, perform line positioning, binarization, noise removal, tilt correction, layout analysis on the front image of the ID card and the reverse image of the ID card. Character segmentation to obtain valid text information for a single ID card;
  • the identity information classification unit is configured to classify the valid text information of the identity card by name, address, ethnicity, and ID number, and output the classified ID information;
  • a comparison verification unit configured to perform a network comparison check on the facial image and the classified identity card information, wherein the networked comparison check is a connection citizenship information verification system for the facial image and the classified
  • the ID card information is verified and judged to be consistent; when the networked verification is performed through the classified ID card information, the user information exists; and the face image is consistent with the ID face image returned by the network verification, the ID card The information is true; otherwise, the ID card information is false;
  • the result output unit is set to get the comparison check result and output.
  • the facial acquisition unit is further configured to perform pre-detection of the captured scene pose, angle, illumination, and occlusion influence factors, and issue a prompt to enable the user to cooperate.
  • the face obtaining unit is further configured to receive a first command input to adjust a predetermined area through an output adjustment button of the client, adjust the predetermined area according to the first instruction, and detect the face The portion image is located at a position in the image, and when the face image is outside a predetermined area in the image, the face is returned to the predetermined area by issuing a prompt.
  • the facial acquisition unit is further configured to perform pre-detection of the screenshot environment for the face pose, angle, illumination, and occlusion influence factors in the captured image, and issue a prompt to enable the user to cooperate.
  • the identity information acquiring unit is further configured to perform coarse positioning on the valid information according to the position of the valid text information in the ID card; according to the grayscale average ratio of the pixels in the row of the valid text information line To be small, the coarsely positioned text information is accurately positioned to generate a pixel matrix to be filtered.
  • the pixel matrix is a precise positioning area, and the formula is: the gray value in the image is less than or equal to the lower boundary LT.
  • the Gaussian filter template performs Gaussian filtering on the precise positioning area to attenuate interference of background information and remove noise.
  • the Gaussian filter scans each pixel in the image with a template.
  • the weighted average gray value of the pixels in the neighborhood determined by the template replaces the value of the center pixel of the template; the dynamic threshold is used to binarize the precise positioning area to obtain a binary image of the precise positioning area, and the binary value of the precise positioning area is obtained.
  • the image is divided into words to obtain a single valid text message.
  • a computer apparatus comprising a memory and a processor, the memory storing computer readable instructions that, when executed by the processor, cause the processor to execute
  • the readable instructions implement the steps in the face recognition based identity authentication method in the above embodiments.
  • a storage medium storing computer readable instructions, when executed by one or more processors, causes one or more processors to perform a person based implementation in the above embodiments Steps in the identity authentication method for face recognition.
  • the storage medium may be a non-volatile storage medium.
  • the program may be stored in a computer readable storage medium, and the storage medium may include: Read Only Memory (ROM), Random Access Memory (RAM), disk or optical disk.
  • ROM Read Only Memory
  • RAM Random Access Memory

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Business, Economics & Management (AREA)
  • Accounting & Taxation (AREA)
  • Finance (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Economics (AREA)
  • General Business, Economics & Management (AREA)
  • Technology Law (AREA)
  • Strategic Management (AREA)
  • Artificial Intelligence (AREA)
  • Marketing (AREA)
  • Development Economics (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Human Computer Interaction (AREA)
  • Collating Specific Patterns (AREA)
  • Image Analysis (AREA)

Abstract

一种基于人脸识别的身份认证方法、装置、设备和存储介质,包括实时获取客户端摄像头采集的图像,从所述图像中获取脸部图像;提取身份证信息,捕获身份证正反面图像,对其处理得到有效文字信息;将有效文字信息按姓名、地址、民族、身份证号归类,输出归类的身份证信息;对脸部图像和归类身份证信息进行联网对比核查;当用户信息存在且脸部图像与联网核查返回的身份证人脸图像一致时,身份证信息为真;否则为假;得到对比核查结果并输出。上述方法提高了身份证信息的安全性,采用智能对比方式,快速确定待识别对象与身份证中个人信息的一致性,完成身份认证。

Description

基于人脸识别的身份认证方法、装置、设备和存储介质
本申请要求于2018年02月09日提交中国专利局、申请号为201810131476.7、发明名称为“基于人脸识别的身份认证方法、装置、设备和存储介质”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
本申请涉及信息安全领域,尤其涉及基于人脸识别的身份认证方法、装置、设备和存储介质。
背景技术
借助互联网的优势,网上信贷也正在成为一种趋势,可以足不出户的完成贷款申请的各项步骤,包括了解各类贷款的申请条件,准备申请材料,一直到递交贷款申请,都可以在互联网上高效的完成。网络信贷业务中,对客户身份的核查不能像传统业务一样通过客户提供身份证件,业务管理人员通过肉眼进行大致的比对来完成。
基于人脸识别的身份认证成为必然,然而现有技术存在以下不足,对人脸图像识别率较低,当人脸外形不稳定,尤其当脸部采样受脸部表情、外部光照、脸部遮盖物等多方面因素的影响时,将增加脸部识别难度,配合身份证进行身份验证时,无法快速过滤身份证中的个人信息和快读读取文字信息,对比识别速度慢。
发明内容
基于此,有必要针对现有技术的不足,提供一种基于人脸识别的身份认证 方法、装置、设备和存储介质。
一种基于人脸识别的身份认证方法,包括:实时获取客户端摄像头采集的图像,从所述图像中获取脸部图像;提取身份证信息,捕获身份证正面图像和身份证反面图像,对身份证正面图像和身份证反面图像进行行定位、二值化、噪声去除、倾斜校正、版面分析、字符分割,得到单个的身份证的有效文字信息;将所述身份证的有效文字信息按姓名、地址、民族、身份证号进行归类,输出归类的身份证信息;对所述脸部图像和所述归类的身份证信息进行联网对比核查,所述联网对比核查为连接公民身份信息核查系统对所述脸部图像和所述归类的身份证信息进行验证,判断其一致性;当通过所述归类的身份证信息进行联网核查,用户信息存在;且所述脸部图像与联网核查返回的身份证人脸图像一致时,身份证信息为真;否则,身份证信息为假;得到对比核查结果并输出。
一种基于人脸识别的身份认证装置,包括:脸部获取单元,设置为实时获取客户端摄像头采集的图像,从所述图像中获取脸部图像;身份信息获取单元,设置为提取身份证信息,捕获身份证正面图像和身份证反面图像,对身份证正面图像和身份证反面图像进行行定位、二值化、噪声去除、倾斜校正、版面分析、字符分割,得到单个的身份证的有效文字信息;身份信息归类单元,设置为将所述身份证的有效文字信息按姓名、地址、民族、身份证号进行归类,输出归类的身份证信息;对比核实单元,设置为对所述脸部图像和所述归类的身份证信息进行联网对比核查,所述联网对比核查为连接公民身份信息核查系统对所述脸部图像和所述归类的身份证信息进行验证,判断其一致性;当通过所述归类的身份证信息进行联网核查,用户信息存在;且所述脸部图像与联网核查返回的身份证人脸图像一致时,身份证信息为真;否则,身份证信息为假; 结果输出单元,设置为得到对比核查结果并输出。
一种计算机设备,包括存储器和处理器,所述存储器中存储有计算机可读指令,所述计算机可读指令被所述处理器执行时,使得所述处理器执行上述基于人脸识别的身份认证方法的步骤。
一种存储有计算机可读指令的存储介质,所述计算机可读指令被一个或多个处理器执行时,使得一个或多个处理器执行上述基于人脸识别的身份认证方法的步骤。
上述基于人脸识别的身份认证方法、装置、设备和存储介质,通过实时获取客户端摄像头采集的图像,从所述图像中获取脸部图像,提取身份证信息,捕获身份证正面图像和身份证反面图像,对身份证正面图像和身份证反面图像进行行定位、二值化、噪声去除、倾斜校正、版面分析、字符分割,得到单个的身份证的有效文字信息,将所述身份证的有效文字信息按姓名、地址、民族、身份证号进行归类,输出归类的身份证信息;对所述脸部图像和所述归类的身份证信息进行联网对比核查,所述联网对比核查为连接公民身份信息核查系统对所述脸部图像和所述归类的身份证信息进行验证,判断其一致性;当通过所述归类的身份证信息进行联网核查,用户信息存在,且所述脸部图像与联网核查返回的身份证人脸图像一致时,身份证信息为真;否则,身份证信息为假;得到对比核查结果并输出,减少了使用身份证个人信息手工录入及肉眼比对的出错率,提高了身份证信息的安全性,采用智能对比方式,快速确定待识别对象与身份证中个人信息的一致性,完成身份认证。
附图说明
通过阅读下文优选实施方式的详细描述,各种其他的优点和益处对于本领 域普通技术人员将变得清楚明了。附图仅用于示出优选实施方式的目的,而并不认为是对本申请的限制。
图1为本申请一个实施例中基于人脸识别的身份认证方法的流程图;
图2为本申请一个实施例中对身份证正面图像和身份证反面图像进行行定位、二值化、噪声去除、倾斜校正、版面分析、字符分割,得到单个的身份证的有效文字信息方法的流程图;
图3为本申请一个实施例中基于人脸识别的身份认证装置的结构框图。
具体实施方式
为了使本申请的目的、技术方案及优点更加清楚明白,以下结合附图及实施例,对本申请进行进一步详细说明。应当理解,此处所描述的具体实施例仅仅用以解释本申请,并不用于限定本申请。
本技术领域技术人员可以理解,除非特意声明,这里使用的单数形式“一”、“一个”、“所述”和“该”也可包括复数形式。应该进一步理解的是,本申请的说明书中使用的措辞“包括”是指存在所述特征、整数、步骤、操作、元件和/或组件,但是并不排除存在或添加一个或多个其他特征、整数、步骤、操作、元件、组件和/或它们的组。
如图1所示,在一个实施例中,提出了一种基于人脸识别的身份认证方法,该基于人脸识别的身份认证方法具体可以包括:
S101,实时获取客户端摄像头采集的图像,从图像中获取脸部图像;
客户端的终端设备可以是手机、IPAD等,这些设备的摄像头实时采集图像,处理器实时从摄像头中获取到该图像,这里的处理器获取图像并非摄像头拍摄的照片,只是摄像头影像区域内的实时图像,获取到实时图像后,如脸部轮廓 数据、脸部特征数据,包括眼睛、鼻子、嘴巴,但不需要对人脸进行具体识别,此步骤只需要将人脸从背景中识别出来即可,便于进行下一步处理。
S102,提取身份证信息,捕获身份证正面图像和身份证反面图像,对身份证正面图像和身份证反面图像进行行定位、二值化、噪声去除、倾斜校正、版面分析、字符分割,得到单个的身份证的有效文字信息;
对客户端的摄像头进行边框标定,即在摄像头捕捉的每一帧图像在中心位置设定一个大小为580*580的红色边框,这样可以简化算法,加快处理进度,提高识别效率,由于身份证正面图像包含人脸信息,故选取大量的人脸图像,并将这些人脸图像标定成正样本,非人脸图像标定成负样本,通过模型检测是否含有人脸图像来区分用户的身份证正面图像和身份证反面图像,即能检测到人脸图像即为身份证正面图像,反之为身份证反面图像。
以区分用户的身份证正面图像和身份证反面图像,简单有效的解决了对身份证正面信息提取的问题,分别对身份证正面图像和身份证反面图像进行相关处理,以获取身份证的有效文字信息。身份证图像中身份证信息位置是固定的,因而对身份证正面图像和身份证的反面图像中的有效信息进行识别时采用对身份证正面图像和身份证反面图像进行定位、二值化、噪声去除、倾斜校正、版面分析、字符分割等预处理,图像二值化就是将图像上的像素点的灰度值设置为0或255,也就是将整个图像呈现出明显的黑白效果的过程,在数字图像处理中,图像的二值化使图像中数据量大为减少,从而能凸显出目标的轮廓,有利于在对图像做进一步处理,然后对图像进行旋转倾斜校正,图像去噪,图像增强等预处理,版面分析,信息栏目定位,行分割和字符分割,汉字由一笔一划构成,每一笔每一划都是由一些连通域内的像素点构成的,方正的汉字的连通域是有特征的,字符间笔画有粘连时,多个字符形成一个连通域,需要根据平 均字宽,通过投影分析法进行切割,得到切分的结果,来得到单个的身份证的有效文字信息,即获得字符结构信息,待识别的字。
S103,将所述身份证的有效文字信息按姓名、地址、民族、身份证号进行归类,输出归类的身份证信息;
将身份证信息按姓名、地址、民族、身份证号识别并归类,并以归类的身份证信息为条件查询公安局人口数据库中对应的字段进行比对。
S104,对所述脸部图像和所述归类的身份证信息进行联网对比核查,所述联网对比核查为连接公民身份信息核查系统对所述脸部图像和所述归类的身份证信息进行验证,判断其一致性;当通过所述归类的身份证信息进行联网核查,用户信息存在;且所述脸部图像与联网核查返回的身份证人脸图像一致时,身份证信息为真;否则,身份证信息为假;
将身份证表面的字符信息识别出来后,自动将姓名和身份证号码输入到公民身份信息联网核查系统中查询,如果查询到库中存在该姓名和身份证号码,系统则返回与此姓名和身份证号码一致的证件信息,具体的包括:姓名、身份证号、住址、民族、出生年月日、人脸照片等。并利用人脸识别技术将身份证上的人脸图像与联网核查中身份证人脸图像进行比对,判断其一致性。
S105,得到对比核查结果并输出。
对比查询结果进行校验,校验类型包括响应结果校验、数据库结果校验和日志结果校验,获取与校验类型相应的接口数据,将接口数据与预设的校正结果数据进行匹配,得到是否匹配的校验结果,由系统可自动完成结果的校验,提高结果校验的效率和降低误查率,并输出校验后的结果。
在一个实施例中,S101中,在实时获取客户端摄像头采集的图像之前还包括:对所采集的人脸姿势、角度、光照、遮挡影响因素进行截图环境预检测, 并发出提示以使用户配合。
通过摄像头调取视频,然后进行人脸定位,对人脸进行活体化检测,对所采集的人脸姿势、角度、光照、遮挡影响因素进行截图环境预检测,并通过语音提示器、显示屏提示用户配合。
在一个实施例中,S101中,在实时获取客户端摄像头采集的图像之前还包括:通过客户端的输出调节按钮,接收第一指令输入以调节预定区域,根据第一指令调节预定区域。
第一指令即为拍照者的控制指令,如拍摄者点击终端设备的调节旋钮以进行预定区域的调节,预定区域的调节包括大小调节和位置调节,当调节旋钮为终端设备的显示屏图像上用于展现预定区域的圈状结构时,拍摄者可以直接在显示屏上拖拉圈状结构以进行位置调节和大小调节,具体地,可以根据拍摄者上述的调节后重新确定预定区域,以后的操作均按照新设的预定区域进行。
在一个实施例中,S101之后还包括:检测脸部图像位于图像中的位置,当脸部图像位于图像中的预定区域以外时,通过发出提示以使得脸部回到预定区域中。
预定区域为预先设置,其可以被动设置,举例说明,如默认为终端移动设备屏幕中间3X3英寸的区域为预定区域,也可以由使用者主动设置其大小和区域。获取到实时图像后,从图像中识别出人脸,并检测人脸是否位于图像中的预定区域内,当处于预定区域之外时,但当检测到人脸不是位于预定区域内时,即控制转动驱动机构驱动手机,通过发出提示以使得脸部回到预定区域中。人脸识别是通信领域的公知常识和惯用技术手段,如人工神经网络法、支持向量机法、PCA识别、弹性匹配方法、基于KL变换的特征脸法、基于积分图像特征法等等,本实施例不一一赘述。
如图2所示,在一个实施例中,在S102中,对身份证正面图像和身份证反面图像进行行定位、二值化、噪声去除、倾斜校正、版面分析、字符分割,得到单个的身份证的有效文字信息包括:
S201,根据有效文字信息在身份证中的位置对有效信息进行粗定位;
由于身份证中有效文字信息位置是固定的,因此可以根据有效文字信息在身份证中的位置对有效信息进行大致定位,即粗定位,比如若整幅身份证图像的宽和高都为0.9,原点在左上角,则身份证编号一般位于(0.38,0.83)为左上角,宽0.48,高0.145的矩形内,获取身份证的有效文字信息。
S202,根据在有效文字信息行的边界所在行像素的灰度平均值比背景要小这一特性,对粗定位的文字信息进行精准定位,生成待进行滤波处理的像素矩阵,像素矩阵为精准定位区域,采用的公式为:将图像中灰度值小于等于下边界LT的点标记为前景点,灰度值大于等于上边界HT的点标记为背景点,其中LT=T*r,HT=255-(255-T)*r,T为切分操作所在层次的最大灰度值,r是经验值,r=0.6,以防止将存在一定差异的背景行当成精准定位的边界,同理得到精准定位区域的上边界、左边界、右边界;
分别处理各个单字图像区域内的像素点,提取出有用信息,根据像素点的分部特征选择种子原点,从种子原点出发采用基于像素点分布特征的区域增长算法,逐渐向邻近像素点前进并读取,直至处理完所有的像素点。基于像素点分布特征的区域增长算法遍历像素点获取字符结构信息,标识确定点,图像中存在一部分相对比较黑或者相对比较白的点很容易被认为是前景或者是背景点,这部分点被称为确定点,在初始化阶段需要先将确定点标识出来,以减少计算耗时。假定T是主层次级别即切分操作所在的层次所对应的最大灰度值,r是经验值,r=0.6,令下边界LT=T*r,HT=255-(255-T)*r,则灰度值小于等于 LT的点是相对较黑的点,标记为前景点,灰度值大于等于上边界HT的点是相对较白的点,标记为背景点,以防止将存在一定差异的背景行当成精准定位的边界,同理得到精准定位区域的上边界、左边界、右边界。
S203,采用3像素×3像素的高斯滤波模板对精准定位区域进行高斯滤波,所述高斯滤波是用一个模板扫描图像中的每一个像素,用模板确定的邻域内像素的加权平均灰度值去替代模板中心像素点的值,减弱背景信息的干扰及去除噪声;
针对仍然存在的一些孤立的笔画或者黑色区域,首先算出平均笔画宽度,如果这些孤立的笔画或者黑色区域的面积小于平均笔画宽度的平方,则认为它是噪音,如果小于两倍的平均笔画宽度的平方,且从它们的中心像素点向八个方向出发到边界的长度均大于平均笔画的宽度,则认为它也是噪音,将噪音去除,因此采用3像素×3像素的高斯滤波模板对精准定位区域进行高斯滤波,减弱背景信息的干扰及去除噪声,高斯滤波就是对整幅图像进行加权平均的过程,每一个像素点的值,都由其本身和邻域内的其他像素值经过加权平均后得到。高斯滤波的具体操作是用一个模板(或称卷积、掩模)扫描图像中的每一个像素,用模板确定的邻域内像素的加权平均灰度值去替代模板中心像素点的值。
S204,采用动态阈值对精准定位区域进行二值化,得到精准定位区域的二值图像,并对精准定位区域的二值图像进行单字分割,得到单个的有效文字信息。
动态的获取图像像素点的二值化阈值用于对平滑处理后的灰度图片进行二值化处理,进一步去噪提高后期匹配的精确性。
如图3所示,在一个实施例中,提供了一种基于人脸识别的身份认证装置, 该装置包括:
脸部获取单元,设置为实时获取客户端摄像头采集的图像,从所述图像中获取脸部图像;
身份信息获取单元,设置为提取身份证信息,捕获身份证正面图像和身份证反面图像,对身份证正面图像和身份证反面图像进行行定位、二值化、噪声去除、倾斜校正、版面分析、字符分割,得到单个的身份证的有效文字信息;
身份信息归类单元,设置为将所述身份证的有效文字信息按姓名、地址、民族、身份证号进行归类,输出归类的身份证信息;
对比核实单元,设置为对所述脸部图像和所述归类的身份证信息进行联网对比核查,所述联网对比核查为连接公民身份信息核查系统对所述脸部图像和所述归类的身份证信息进行验证,判断其一致性;当通过所述归类的身份证信息进行联网核查,用户信息存在;且所述脸部图像与联网核查返回的身份证人脸图像一致时,身份证信息为真;否则,身份证信息为假;
结果输出单元,设置为得到对比核查结果并输出。
在一个实施例中,所述脸部获取单元还设置为对所采集的人脸姿势、角度、光照、遮挡影响因素进行截图环境预检测,并发出提示以使用户配合。
在一个实施例中,所述脸部获取单元还设置为通过所述客户端的输出调节按钮,接收第一指令输入以调节预定区域,根据所述第一指令调节所述预定区域和检测所述脸部图像位于所述图像中的位置,当所述脸部图像位于所述图像中的预定区域以外时,通过发出提示以使得所述脸部回到预定区域中。
在一个实施例中,所述脸部获取单元还设置为对所采集的图像中,人脸姿势、角度、光照、遮挡影响因素进行截图环境预检测,并发出提示以使用户配合。
在一个实施例中,所述身份信息获取单元还设置为根据有效文字信息在身份证中的位置对有效信息进行粗定位;根据在有效文字信息行的边界所在行像素的灰度平均值比背景要小这一特性,对粗定位的文字信息进行精准定位,生成待进行滤波处理的像素矩阵,所述像素矩阵为精准定位区域,采用的公式为:将图像中灰度值小于等于下边界LT的点标记为前景点,灰度值大于等于上边界HT的点标记为背景点,其中LT=T*r,HT=255-(255-T)*r,T为切分操作所在层次的最大灰度值,r是经验值,r=0.6,以防止将存在一定差异的背景行当成精准定位的边界,同理得到精准定位区域的上边界、左边界、右边界;采用3像素×3像素的高斯滤波模板对所述精准定位区域进行高斯滤波,减弱背景信息的干扰及去除噪声,所述高斯滤波是用一个模板扫描图像中的每一个像素,用模板确定的邻域内像素的加权平均灰度值去替代模板中心像素点的值;采用动态阈值对精准定位区域进行二值化,得到精准定位区域的二值图像,并对精准定位区域的二值图像进行单字分割,得到单个的有效文字信息。
在一个实施例中,提出了一种计算机设备,包括存储器和处理器,所述存储器中存储有计算机可读指令,所述计算机可读指令被所述处理器执行时,使得所述处理器执行所述可读指令时实现上述各实施例中基于人脸识别的身份认证方法中的步骤。
在一个实施例中,提出了一种存储有计算机可读指令的存储介质,所述计算机可读指令被一个或多个处理器执行时,使得一个或多个处理器执行上述实施例中基于人脸识别的身份认证方法中的步骤。其中,存储介质可以为非易失性存储介质。
本领域普通技术人员可以理解上述实施例的各种方法中的全部或部分步骤是可以通过程序来指令相关的硬件来完成,该程序可以存储于一计算机可读存 储介质中,存储介质可以包括:只读存储器(ROM,Read Only Memory)、随机存取存储器(RAM,Random Access Memory)、磁盘或光盘等。
以上所述实施例的各技术特征可以进行任意的组合,为使描述简洁,未对上述实施例中的各个技术特征所有可能的组合都进行描述,然而,只要这些技术特征的组合不存在矛盾,都应当认为是本说明书记载的范围。
以上所述实施例仅表达了本申请一些示例性实施例,其描述较为具体和详细,但并不能因此而理解为对本申请专利范围的限制。应当指出的是,对于本领域的普通技术人员来说,在不脱离本申请构思的前提下,还可以做出若干变形和改进,这些都属于本申请的保护范围。因此,本申请专利的保护范围应以所附权利要求为准。

Claims (20)

  1. 一种基于人脸识别的身份认证方法,包括:
    实时获取客户端摄像头采集的图像,从所述图像中获取脸部图像;
    提取身份证信息,捕获身份证正面图像和身份证反面图像,对身份证正面图像和身份证反面图像进行行定位、二值化、噪声去除、倾斜校正、版面分析、字符分割,得到单个的身份证的有效文字信息;
    将所述身份证的有效文字信息按姓名、地址、民族、身份证号进行归类,输出归类的身份证信息;
    对所述脸部图像和所述归类的身份证信息进行联网对比核查,所述联网对比核查为连接公民身份信息核查系统对所述脸部图像和所述归类的身份证信息进行验证,判断其一致性;当通过所述归类的身份证信息进行联网核查,用户信息存在;且所述脸部图像与联网核查返回的身份证人脸图像一致时,身份证信息为真;否则,身份证信息为假;
    得到对比核查结果并输出。
  2. 根据权利要求1所述的方法,其中,所述实时获取客户端摄像头采集的图像之前还包括:
    对所采集的图像中,人脸姿势、角度、光照、遮挡影响因素进行截图环境预检测,并发出提示以使用户配合。
  3. 根据权利要求1所述的方法,其中,所述实时获取客户端摄像头采集的图像之前还包括:
    通过所述客户端的输出调节按钮,接收第一指令输入以调节预定区域,根据所述第一指令调节所述预定区域。
  4. 根据权利要求1所述的方法,其中,所述实时获取客户端摄像头采集的图像,从所述图像中获取脸部图像之后还包括:
    检测所述脸部图像位于所述图像中的位置,当所述脸部图像位于所述图像中的预定区域以外时,通过发出提示以使得所述脸部图像回到预定区域中。
  5. 根据权利要求4所述的方法,其中,所述对身份证正面图像和身份证反面图像进行行定位、二值化、噪声去除、倾斜校正、版面分析、字符分割,得到单个的身份证的有效文字信息包括:
    根据有效文字信息在身份证中的位置对有效信息进行粗定位;
    根据在有效文字信息行的边界所在行像素的灰度平均值比背景要小这一特性,对粗定位的文字信息进行精准定位,生成待进行滤波处理的像素矩阵,所述像素矩阵为精准定位区域,采用的公式为:将图像中灰度值小于等于下边界LT的点标记为前景点,灰度值大于等于上边界HT的点标记为背景点,其中LT=T*r,HT=255-(255-T)*r,T为切分操作所在层次的最大灰度值,r是经验值,r=0.6,以防止将存在一定差异的背景行当成精准定位的边界,同理得到精准定位区域的上边界、左边界、右边界;
    采用3像素×3像素的高斯滤波模板对所述精准定位区域进行高斯滤波,减弱背景信息的干扰及去除噪声,所述高斯滤波是用一个模板扫描图像中的每一个像素,用模板确定的邻域内像素的加权平均灰度值去替代模板中心像素点的值;
    采用动态阈值对精准定位区域进行二值化,得到精准定位区域的二值图像,并对精准定位区域的二值图像进行单字分割,得到单个的有效文字信息。
  6. 一种基于人脸识别的身份认证装置,包括:
    脸部获取单元,设置为实时获取客户端摄像头采集的图像,从所述图像中获取脸部图像;
    身份信息获取单元,设置为提取身份证信息,捕获身份证正面图像和身份 证反面图像,对身份证正面图像和身份证反面图像进行行定位、二值化、噪声去除、倾斜校正、版面分析、字符分割,得到单个的身份证的有效文字信息;
    身份信息归类单元,设置为将所述身份证的有效文字信息按姓名、地址、民族、身份证号进行归类,输出归类的身份证信息;
    对比核实单元,设置为对所述脸部图像和所述归类的身份证信息进行联网对比核查,所述联网对比核查为连接公民身份信息核查系统对所述脸部图像和所述归类的身份证信息进行验证,判断其一致性;当通过所述归类的身份证信息进行联网核查,用户信息存在;且所述脸部图像与联网核查返回的身份证人脸图像一致时,身份证信息为真;否则,身份证信息为假;
    结果输出单元,设置为得到对比核查结果并输出。
  7. 根据权利要求6所述的装置,其中,所述脸部获取单元还设置为对所采集的人脸姿势、角度、光照、遮挡影响因素进行截图环境预检测,并发出提示以使用户配合。
  8. 根据权利要求6所述的装置,其中,所述脸部获取单元还设置为通过所述客户端的输出调节按钮,接收第一指令输入以调节预定区域,根据所述第一指令调节所述预定区域和检测所述脸部图像位于所述图像中的位置,当所述脸部图像位于所述图像中的预定区域以外时,通过发出提示以使得所述脸部回到预定区域中。
  9. 根据权利要求6所述的装置,其中,所述脸部获取单元还设置为对所采集的图像中,人脸姿势、角度、光照、遮挡影响因素进行截图环境预检测,并发出提示以使用户配合。
  10. 根据权利要求6所述的装置,其中,所述身份信息获取单元还设置为根据有效文字信息在身份证中的位置对有效信息进行粗定位;根据在有效文字 信息行的边界所在行像素的灰度平均值比背景要小这一特性,对粗定位的文字信息进行精准定位,生成待进行滤波处理的像素矩阵,所述像素矩阵为精准定位区域,采用的公式为:将图像中灰度值小于等于下边界LT的点标记为前景点,灰度值大于等于上边界HT的点标记为背景点,其中LT=T*r,HT=255-(255-T)*r,T为切分操作所在层次的最大灰度值,r是经验值,r=0.6,以防止将存在一定差异的背景行当成精准定位的边界,同理得到精准定位区域的上边界、左边界、右边界;采用3像素×3像素的高斯滤波模板对所述精准定位区域进行高斯滤波,减弱背景信息的干扰及去除噪声,所述高斯滤波是用一个模板扫描图像中的每一个像素,用模板确定的邻域内像素的加权平均灰度值去替代模板中心像素点的值;采用动态阈值对精准定位区域进行二值化,得到精准定位区域的二值图像,并对精准定位区域的二值图像进行单字分割,得到单个的有效文字信息。
  11. 一种计算机设备,包括存储器和处理器,所述存储器中存储有计算机可读指令,所述计算机可读指令被所述处理器执行时,使得所述处理器执行以下步骤:
    实时获取客户端摄像头采集的图像,从所述图像中获取脸部图像;
    提取身份证信息,捕获身份证正面图像和身份证反面图像,对身份证正面图像和身份证反面图像进行行定位、二值化、噪声去除、倾斜校正、版面分析、字符分割,得到单个的身份证的有效文字信息;
    将所述身份证的有效文字信息按姓名、地址、民族、身份证号进行归类,输出归类的身份证信息;
    对所述脸部图像和所述归类的身份证信息进行联网对比核查,所述联网对比核查为连接公民身份信息核查系统对所述脸部图像和所述归类的身份证信息 进行验证,判断其一致性;当通过所述归类的身份证信息进行联网核查,用户信息存在;且所述脸部图像与联网核查返回的身份证人脸图像一致时,身份证信息为真;否则,身份证信息为假;
    得到对比核查结果并输出。
  12. 根据权利要求11所述的一种计算机设备,其中,所述实时获取客户端摄像头采集的图像之前还包括:对所采集的图像中,人脸姿势、角度、光照、遮挡影响因素进行截图环境预检测,并发出提示以使用户配合。
  13. 根据权利要求11所述的一种计算机设备,其中,所述实时获取客户端摄像头采集的图像之前还包括:通过所述客户端的输出调节按钮,接收第一指令输入以调节预定区域,根据所述第一指令调节所述预定区域。
  14. 根据权利要求11所述的一种计算机设备,其中,所述实时获取客户端摄像头采集的图像,从所述图像中获取脸部图像之后还包括:检测所述脸部图像位于所述图像中的位置,当所述脸部图像位于所述图像中的预定区域以外时,通过发出提示以使得所述脸部图像回到预定区域中。
  15. 根据权利要求11所述的一种计算机设备,其中,所述对身份证正面图像和身份证反面图像进行行定位、二值化、噪声去除、倾斜校正、版面分析、字符分割,得到单个的身份证的有效文字信息包括:根据有效文字信息在身份证中的位置对有效信息进行粗定位;根据在有效文字信息行的边界所在行像素的灰度平均值比背景要小这一特性,对粗定位的文字信息进行精准定位,生成待进行滤波处理的像素矩阵,所述像素矩阵为精准定位区域,采用的公式为:将图像中灰度值小于等于下边界LT的点标记为前景点,灰度值大于等于上边界HT的点标记为背景点,其中LT=T*r,HT=255-(255-T)*r,T为切分操作所在层次的最大灰度值,r是经验值,r=0.6,以防止将存在一定差异的背景行当成精 准定位的边界,同理得到精准定位区域的上边界、左边界、右边界;采用3像素×3像素的高斯滤波模板对所述精准定位区域进行高斯滤波,减弱背景信息的干扰及去除噪声,所述高斯滤波是用一个模板扫描图像中的每一个像素,用模板确定的邻域内像素的加权平均灰度值去替代模板中心像素点的值;采用动态阈值对精准定位区域进行二值化,得到精准定位区域的二值图像,并对精准定位区域的二值图像进行单字分割,得到单个的有效文字信息。
  16. 一种存储有计算机可读指令的存储介质,所述计算机可读指令被一个或多个处理器执行时,使得一个或多个处理器执行以下步骤:
    实时获取客户端摄像头采集的图像,从所述图像中获取脸部图像;
    提取身份证信息,捕获身份证正面图像和身份证反面图像,对身份证正面图像和身份证反面图像进行行定位、二值化、噪声去除、倾斜校正、版面分析、字符分割,得到单个的身份证的有效文字信息;
    将所述身份证的有效文字信息按姓名、地址、民族、身份证号进行归类,输出归类的身份证信息;
    对所述脸部图像和所述归类的身份证信息进行联网对比核查,所述联网对比核查为连接公民身份信息核查系统对所述脸部图像和所述归类的身份证信息进行验证,判断其一致性;当通过所述归类的身份证信息进行联网核查,用户信息存在;且所述脸部图像与联网核查返回的身份证人脸图像一致时,身份证信息为真;否则,身份证信息为假;
    得到对比核查结果并输出。
  17. 根据权利要求16所述的一种计算机设备,其中,所述实时获取客户端摄像头采集的图像之前还包括:对所采集的图像中,人脸姿势、角度、光照、遮挡影响因素进行截图环境预检测,并发出提示以使用户配合。
  18. 根据权利要求16所述的一种计算机设备,其中,所述实时获取客户端摄像头采集的图像之前还包括:通过所述客户端的输出调节按钮,接收第一指令输入以调节预定区域,根据所述第一指令调节所述预定区域。
  19. 根据权利要求16所述的一种计算机设备,其中,所述实时获取客户端摄像头采集的图像,从所述图像中获取脸部图像之后还包括:检测所述脸部图像位于所述图像中的位置,当所述脸部图像位于所述图像中的预定区域以外时,通过发出提示以使得所述脸部图像回到预定区域中。
  20. 根据权利要求16所述的一种计算机设备,其中,所述对身份证正面图像和身份证反面图像进行行定位、二值化、噪声去除、倾斜校正、版面分析、字符分割,得到单个的身份证的有效文字信息包括:根据有效文字信息在身份证中的位置对有效信息进行粗定位;根据在有效文字信息行的边界所在行像素的灰度平均值比背景要小这一特性,对粗定位的文字信息进行精准定位,生成待进行滤波处理的像素矩阵,所述像素矩阵为精准定位区域,采用的公式为:将图像中灰度值小于等于下边界LT的点标记为前景点,灰度值大于等于上边界HT的点标记为背景点,其中LT=T*r,HT=255-(255-T)*r,T为切分操作所在层次的最大灰度值,r是经验值,r=0.6,以防止将存在一定差异的背景行当成精准定位的边界,同理得到精准定位区域的上边界、左边界、右边界;采用3像素×3像素的高斯滤波模板对所述精准定位区域进行高斯滤波,减弱背景信息的干扰及去除噪声,所述高斯滤波是用一个模板扫描图像中的每一个像素,用模板确定的邻域内像素的加权平均灰度值去替代模板中心像素点的值;采用动态阈值对精准定位区域进行二值化,得到精准定位区域的二值图像,并对精准定位区域的二值图像进行单字分割,得到单个的有效文字信息。
PCT/CN2018/105173 2018-02-09 2018-09-12 基于人脸识别的身份认证方法、装置、设备和存储介质 WO2019153739A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201810131476.7 2018-02-09
CN201810131476.7A CN109410026A (zh) 2018-02-09 2018-02-09 基于人脸识别的身份认证方法、装置、设备和存储介质

Publications (1)

Publication Number Publication Date
WO2019153739A1 true WO2019153739A1 (zh) 2019-08-15

Family

ID=65463950

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/105173 WO2019153739A1 (zh) 2018-02-09 2018-09-12 基于人脸识别的身份认证方法、装置、设备和存储介质

Country Status (2)

Country Link
CN (1) CN109410026A (zh)
WO (1) WO2019153739A1 (zh)

Cited By (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110969143A (zh) * 2019-12-19 2020-04-07 深圳壹账通智能科技有限公司 基于图像识别的取证方法、系统、计算机设备及存储介质
CN111291711A (zh) * 2020-02-25 2020-06-16 山东超越数控电子股份有限公司 一种基于Python的深度学习人脸识别方法,设备及可读存储介质
CN111324878A (zh) * 2020-02-05 2020-06-23 重庆特斯联智慧科技股份有限公司 一种基于人脸识别的身份验证方法、装置、存储介质及终端
CN111461131A (zh) * 2020-04-16 2020-07-28 上海东普信息科技有限公司 身份证号码信息识别方法、装置、设备及存储介质
CN111768346A (zh) * 2020-05-12 2020-10-13 北京奇艺世纪科技有限公司 身份证背面图像的校正方法、装置、设备及存储介质
CN111860314A (zh) * 2020-07-20 2020-10-30 浪潮云信息技术股份公司 基于图像识别的电子证照验证方法、装置和系统
CN111898601A (zh) * 2020-07-14 2020-11-06 浙江大华技术股份有限公司 一种身份证要素提取方法及装置
CN111986794A (zh) * 2020-09-03 2020-11-24 平安国际智慧城市科技股份有限公司 基于人脸识别的防冒挂号方法、装置、计算机设备及介质
CN113129020A (zh) * 2021-05-18 2021-07-16 中国银行股份有限公司 一种基于物联网的5g消息身份认证系统及方法
CN113449663A (zh) * 2021-07-06 2021-09-28 深圳中智明科智能科技有限公司 基于多态拟合的协同智能安防方法及装置
CN113688815A (zh) * 2021-06-01 2021-11-23 无锡启凌科技有限公司 用于复杂光照环境的药品包装文字计算机识别算法及装置
CN113793196A (zh) * 2021-08-31 2021-12-14 宁夏百川电力股份有限公司 一种基于云计算的电医生售电管理系统
CN118097192A (zh) * 2024-04-23 2024-05-28 深圳妙月科技有限公司 一种基于云边协同的网关数据处理方法及系统

Families Citing this family (17)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112906741A (zh) * 2019-05-21 2021-06-04 北京嘀嘀无限科技发展有限公司 图像处理方法、装置、电子设备及存储介质
CN110443184B (zh) * 2019-07-31 2022-09-30 上海海事大学 身份证信息提取方法、装置及计算机存储介质
CN110533568A (zh) * 2019-10-09 2019-12-03 重庆特斯联智慧科技股份有限公司 一种基于双重身份认证的公安服务方法及其系统
CN110889326B (zh) * 2019-10-16 2022-07-01 中科南京人工智能创新研究院 基于人体检测的插队行为监测警告系统、方法、装置和存储介质
CN111405365B (zh) * 2020-03-12 2021-09-21 安徽文香科技有限公司 一种身份验证方法、装置、系统及存储介质
CN111553312A (zh) * 2020-05-13 2020-08-18 中国银行股份有限公司 业务办理方法及装置
CN111783757A (zh) * 2020-06-01 2020-10-16 成都科大极智科技有限公司 一种基于ocr技术的复杂场景下身份证识别方法
CN111784498A (zh) * 2020-06-22 2020-10-16 北京海益同展信息科技有限公司 身份认证方法、装置、电子设备及存储介质
CN111915423A (zh) * 2020-07-18 2020-11-10 湖南三湘银行股份有限公司 银行线上录像系统
CN111626274B (zh) * 2020-07-30 2020-10-27 四川骏逸富顿科技有限公司 一种社保卡识别方法
CN113449673A (zh) * 2021-07-09 2021-09-28 中国银行股份有限公司 一种证件识别核实方法和装置
CN113688362A (zh) * 2021-08-25 2021-11-23 中国工商银行股份有限公司 身份证信息安全处理方法及装置
CN114005160B (zh) * 2021-10-28 2022-05-17 建湖县公安局 一种基于身份二维码和人工智能的门禁控制系统及方法
CN115471919B (zh) * 2022-09-19 2023-09-12 江苏至真健康科技有限公司 一种基于便携式免散瞳眼底照相机的建档方法及系统
CN116189181B (zh) * 2022-11-11 2024-01-23 杭州海量信息技术有限公司 一种身份证ocr的图像规范化方法及系统
CN116281070A (zh) * 2023-03-30 2023-06-23 山东大舜医药物流有限公司 基于大数据的医药分拣系统及方法
CN116156426B (zh) * 2023-04-20 2023-06-30 环球数科集团有限公司 一种应用ldsw技术的人员组织管理系统

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103218599A (zh) * 2013-03-26 2013-07-24 苏州福丰科技有限公司 基于人脸识别第二代居民身份证的认证系统和认证方法
CN106991421A (zh) * 2017-03-22 2017-07-28 湖南联信科技有限公司 一种身份证信息提取系统
AU2017101064A4 (en) * 2017-08-04 2017-08-31 Id-Checkr Pty Ltd A system and method of extracting data (ie. text and/or image) from an identity card via a user’s mobile (smart) phone (or computer or tablet) for automatic verification of that user’s digital identity online for an agent (or another user) (“id-checkr”).
CN107145873A (zh) * 2017-05-12 2017-09-08 江苏鸿信系统集成有限公司 基于人脸识别和ocr的身份证图像识别审核方法及系统
CN107609529A (zh) * 2017-09-22 2018-01-19 芜湖星途机器人科技有限公司 机器人人脸识别方法

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104680161A (zh) * 2015-01-09 2015-06-03 安徽清新互联信息科技有限公司 一种身份证数字识别方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103218599A (zh) * 2013-03-26 2013-07-24 苏州福丰科技有限公司 基于人脸识别第二代居民身份证的认证系统和认证方法
CN106991421A (zh) * 2017-03-22 2017-07-28 湖南联信科技有限公司 一种身份证信息提取系统
CN107145873A (zh) * 2017-05-12 2017-09-08 江苏鸿信系统集成有限公司 基于人脸识别和ocr的身份证图像识别审核方法及系统
AU2017101064A4 (en) * 2017-08-04 2017-08-31 Id-Checkr Pty Ltd A system and method of extracting data (ie. text and/or image) from an identity card via a user’s mobile (smart) phone (or computer or tablet) for automatic verification of that user’s digital identity online for an agent (or another user) (“id-checkr”).
CN107609529A (zh) * 2017-09-22 2018-01-19 芜湖星途机器人科技有限公司 机器人人脸识别方法

Cited By (17)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110969143A (zh) * 2019-12-19 2020-04-07 深圳壹账通智能科技有限公司 基于图像识别的取证方法、系统、计算机设备及存储介质
CN111324878A (zh) * 2020-02-05 2020-06-23 重庆特斯联智慧科技股份有限公司 一种基于人脸识别的身份验证方法、装置、存储介质及终端
CN111291711A (zh) * 2020-02-25 2020-06-16 山东超越数控电子股份有限公司 一种基于Python的深度学习人脸识别方法,设备及可读存储介质
CN111461131B (zh) * 2020-04-16 2023-04-18 上海东普信息科技有限公司 身份证号码信息识别方法、装置、设备及存储介质
CN111461131A (zh) * 2020-04-16 2020-07-28 上海东普信息科技有限公司 身份证号码信息识别方法、装置、设备及存储介质
CN111768346A (zh) * 2020-05-12 2020-10-13 北京奇艺世纪科技有限公司 身份证背面图像的校正方法、装置、设备及存储介质
CN111768346B (zh) * 2020-05-12 2023-09-01 北京奇艺世纪科技有限公司 身份证背面图像的校正方法、装置、设备及存储介质
CN111898601A (zh) * 2020-07-14 2020-11-06 浙江大华技术股份有限公司 一种身份证要素提取方法及装置
CN111860314A (zh) * 2020-07-20 2020-10-30 浪潮云信息技术股份公司 基于图像识别的电子证照验证方法、装置和系统
CN111860314B (zh) * 2020-07-20 2024-05-17 浪潮云信息技术股份公司 基于图像识别的电子证照验证方法、装置和系统
CN111986794A (zh) * 2020-09-03 2020-11-24 平安国际智慧城市科技股份有限公司 基于人脸识别的防冒挂号方法、装置、计算机设备及介质
CN111986794B (zh) * 2020-09-03 2024-06-25 深圳平安智慧医健科技有限公司 基于人脸识别的防冒挂号方法、装置、计算机设备及介质
CN113129020A (zh) * 2021-05-18 2021-07-16 中国银行股份有限公司 一种基于物联网的5g消息身份认证系统及方法
CN113688815A (zh) * 2021-06-01 2021-11-23 无锡启凌科技有限公司 用于复杂光照环境的药品包装文字计算机识别算法及装置
CN113449663A (zh) * 2021-07-06 2021-09-28 深圳中智明科智能科技有限公司 基于多态拟合的协同智能安防方法及装置
CN113793196A (zh) * 2021-08-31 2021-12-14 宁夏百川电力股份有限公司 一种基于云计算的电医生售电管理系统
CN118097192A (zh) * 2024-04-23 2024-05-28 深圳妙月科技有限公司 一种基于云边协同的网关数据处理方法及系统

Also Published As

Publication number Publication date
CN109410026A (zh) 2019-03-01

Similar Documents

Publication Publication Date Title
WO2019153739A1 (zh) 基于人脸识别的身份认证方法、装置、设备和存储介质
US10726260B2 (en) Feature extraction and matching for biometric authentication
US11676285B1 (en) System, computing device, and method for document detection
US9922238B2 (en) Apparatuses, systems, and methods for confirming identity
TW202006602A (zh) 三維臉部活體檢測方法、臉部認證識別方法及裝置
US20200410074A1 (en) Identity authentication method and apparatus, electronic device, and storage medium
WO2018086543A1 (zh) 活体判别方法、身份认证方法、终端、服务器和存储介质
WO2020000908A1 (zh) 一种人脸活体检测方法及装置
WO2019223069A1 (zh) 基于直方图的虹膜图像增强方法、装置、设备及存储介质
JP6255486B2 (ja) 情報認識のための方法及びシステム
US6885766B2 (en) Automatic color defect correction
US7657086B2 (en) Method and apparatus for automatic eyeglasses detection using a nose ridge mask
WO2019061658A1 (zh) 眼镜定位方法、装置及存储介质
WO2016084072A1 (en) Anti-spoofing system and methods useful in conjunction therewith
US11244150B2 (en) Facial liveness detection
WO2019223068A1 (zh) 虹膜图像局部增强方法、装置、设备及存储介质
US20210157891A1 (en) Certificate verification
WO2017177616A1 (zh) 一种人脸识别方法及装置、画面显示方法及装置
CN113379713B (zh) 证件图像的检测方法及其装置
WO2019223066A1 (zh) 虹膜图像全局增强方法、装置、设备及存储介质
CN113822927A (zh) 一种适用弱质量图像的人脸检测方法、装置、介质及设备
WO2019223067A1 (zh) 基于多重处理的虹膜图像增强方法、装置、设备及介质
WO2021139447A1 (zh) 一种宫颈异常细胞检测装置及方法
CN113014914B (zh) 一种基于神经网络的单人换脸短视频的识别方法和系统
WO2017219562A1 (zh) 一种二维码生成方法及装置

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: 18904651

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 04/12/2020)

122 Ep: pct application non-entry in european phase

Ref document number: 18904651

Country of ref document: EP

Kind code of ref document: A1