WO2021227124A1 - 一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法 - Google Patents

一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法 Download PDF

Info

Publication number
WO2021227124A1
WO2021227124A1 PCT/CN2020/091718 CN2020091718W WO2021227124A1 WO 2021227124 A1 WO2021227124 A1 WO 2021227124A1 CN 2020091718 W CN2020091718 W CN 2020091718W WO 2021227124 A1 WO2021227124 A1 WO 2021227124A1
Authority
WO
WIPO (PCT)
Prior art keywords
face
thermal imaging
image
area
facial
Prior art date
Application number
PCT/CN2020/091718
Other languages
English (en)
French (fr)
Inventor
郑乐行
李霁
邹方玉
刘晋
陈科言
Original Assignee
南京中科道置智能科技有限公司
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 南京中科道置智能科技有限公司 filed Critical 南京中科道置智能科技有限公司
Publication of WO2021227124A1 publication Critical patent/WO2021227124A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/168—Feature extraction; Face representation
    • G06V40/171—Local features and components; Facial parts ; Occluding parts, e.g. glasses; Geometrical relationships
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
    • G01J5/00—Radiation pyrometry, e.g. infrared or optical thermometry
    • G01J5/0022—Radiation pyrometry, e.g. infrared or optical thermometry for sensing the radiation of moving bodies
    • G01J5/0025—Living bodies
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
    • G01J5/00—Radiation pyrometry, e.g. infrared or optical thermometry
    • G01J5/48—Thermography; Techniques using wholly visual means
    • G01J5/485—Temperature profile
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/172—Classification, e.g. identification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/18—Eye characteristics, e.g. of the iris
    • G06V40/193—Preprocessing; Feature extraction
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/18—Eye characteristics, e.g. of the iris
    • G06V40/197—Matching; Classification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/40—Spoof detection, e.g. liveness detection
    • G06V40/45—Detection of the body part being alive
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
    • G01J5/00—Radiation pyrometry, e.g. infrared or optical thermometry
    • G01J2005/0077—Imaging

Definitions

  • the invention provides a face recognition living body detection method based on facial iris recognition and thermal imaging technology, and belongs to the technical field of identity recognition.
  • Living body detection is a method to determine the true physiological characteristics of objects in some identity verification scenarios.
  • living body detection can use combined actions such as blinking, opening mouth, shaking head, and nodding, using technologies such as face key point positioning and face tracking.
  • face key point positioning and face tracking To verify whether the user is a real living person. It can effectively resist common attack methods such as photos, face changes, masks, occlusions, and screen remakes, so as to help users identify fraudulent behaviors and protect users' interests.
  • 3D printing and high-precision display technology there have been cases of using industrial-grade equipment to manufacture camouflage masks or using high-precision displays to display dynamic iris images to deceive biometric authentication systems.
  • the existing face recognition live body detection mainly uses the following three solutions: 1.
  • the user cooperates with the corresponding action to complete the live body detection, and the user needs to cooperate with a series of actions such as shaking his head, blinking, and opening his mouth, resulting in low user experience.
  • the algorithm is used to analyze and judge the video images collected by a single visible light camera. Because the video images collected by the camera are in two-dimensional space, it is difficult to distinguish whether the face facing the camera is a living face, which leads to recognition. The accuracy is low; 3.
  • the depth-of-field camera uses the depth-of-field camera to perform 3D modeling of the subject, and use the 3D model to determine whether it is a live human face, but the depth-of-field camera used in this method is extremely expensive, and the 3D modeling algorithm requires a lot of computer Computing power, very low computing speed.
  • the prior art still has problems such as the accuracy of living body detection that needs to be improved, it can only prevent photo attacks, but cannot prevent screen attacks, and has a single anti-attack method. Therefore, there is an urgent need to develop a living body detection method with high recognition accuracy, fast recognition speed, and accurate recognition results.
  • the present invention provides a face recognition living body detection method based on facial iris recognition and thermal imaging technology with high recognition accuracy, fast recognition speed, and accurate recognition results.
  • the present invention provides a face recognition living body detection method based on facial iris recognition and thermal imaging technology, including:
  • iris image uses the iris image to identify the relative distance between the iris, eye socket, eyeball and face of the tested target to form a point set of iris, eye socket, eyeball, and face;
  • the tested target is a living human face.
  • the lens parameters of the imaging sensor are consistent with the lens parameter settings of the face capture camera.
  • the thermal imaging image determined as abnormal body temperature is recorded and stored as a case thermal imaging image data sample library, which is used to assist the comparison and screening of thermal imaging image characteristics of epidemic patients who are expected to be monitored later.
  • the face images judged to be abnormal body temperature are recorded and stored as a sample library of visible light image data of the case, which is used for the tracking of epidemic patients in the later stage.
  • the specific steps to determine the temperature distribution are: (1) Segmentation of the face, eyes and cheeks, eye sockets, and eyeball regions on the visible light image obtained by shooting the human face to determine the forehead, cheeks, eye sockets and eyeballs The plane space occupied in the image; (2) This plane space is mapped to the infrared thermal imaging image; (3) The detection algorithm is used to collect the forehead body temperature with the highest temperature in the thermal imaging grayscale image, and the body temperature Compare the preset value of, to determine whether the body temperature is abnormal.
  • the method of segmentation judgment in step (1) is as follows: first, first identify the image space occupied by the face, and crop this part of the image into a new image to become a facial image; second, the facial image It is sent to the cascade classifier to obtain the orbital area, the upper part, the forehead area, the lower part, the left part, the right part, and the single cheek part.
  • the position of the eyeball and nose is identified by the classifier, the eyeball area expands 450%-550% of the eyeball area area and the eyeball area is removed as the orbital area; the line is connected by the middle point of the eyeballs of the two eyes Divide the face image into two parts: the upper part and the lower part; the upper part takes the center point, and expands 10-20% of the upper part outward as the forehead area; the lower part is made perpendicular to the middle of the eyeball with the middle point of the nose Click the connecting line segment and extend it, and divide it into the left part and the right part; in the left part and the right part, separately take the outermost position of the eyeball area to make a downward vertical line, and make the lowermost part of the nose area. Intersect the horizontal lines to obtain the position of the center point of the cheek, and expand 10-20% of the lower half of the area as a single cheek.
  • step (3) when the preset value of body temperature is set, the temperature of the eyeball area is the highest, which is regarded as the highest point of facial temperature; in the thermal imaging image, the median temperature is taken according to the area as a single The temperature of the area; the temperature difference between the forehead area and the highest point area of the face is -0.5 to -3 degrees Celsius; the temperature difference between the orbital area and the highest point area of the face should be -0.5 to -3 degrees Celsius; the orbital area and the highest point area of the face The temperature difference range should be -2 to -5 degrees Celsius; the temperature difference between the cheek area and the highest point area of the face should be -7 to -12 degrees Celsius.
  • the living body detection method provided by the present invention utilizes a combination of facial iris recognition and thermal imaging technology, which can not only quickly identify living bodies with high recognition accuracy, but also can record the temperature of the identified living body for epidemic screening.
  • using thermal imaging images as live verification before facial and iris feature inspection can effectively prevent attackers from using high-precision printed facial or iris images or three-dimensional facial models, eyeball models, dynamic image display and other methods to forge the recognition system. .
  • FIG. 1 is a schematic diagram of the segmented area of a face image in Embodiment 1.
  • FIG. 1 is a schematic diagram of the segmented area of a face image in Embodiment 1.
  • FIG. 2 is a temperature distribution diagram corresponding to the infrared thermal imaging image of the embodiment 1.
  • FIG. 2 is a temperature distribution diagram corresponding to the infrared thermal imaging image of the embodiment 1.
  • a face recognition living body detection method based on facial iris recognition and thermal imaging technology including: collecting visible light images taken in real time of the human face, and segmenting the image based on facial iris recognition;
  • the above method is adopted to set a camera and a thermal imaging sensor with the same imaging angle and position;
  • iris image uses the iris image to identify the relative distance between the iris, eye socket, eyeball and face of the tested target to form a point set of iris, eye socket, eyeball, and face;
  • the tested target is a living human face.
  • the lens parameters of the imaging sensor are consistent with the lens parameter settings of the face capture camera.
  • the highest regional temperature of the secondary image recognized by the thermal imaging image is recorded for epidemic screening.
  • the thermal imaging images judged as abnormal body temperature are recorded and stored as a case thermal imaging image data sample library, which is used to assist the comparison and screening of thermal imaging image characteristics of epidemic patients who are expected to be monitored later.
  • the face images judged to be abnormal body temperature are recorded and stored as a sample library of visible light image data of the case, which is used for the tracking of epidemic patients in the later stage.
  • the specific steps to determine the temperature distribution are: (1) Segment the human face, eyes and cheeks, eye sockets, and eyeball regions on the visible light image obtained by shooting the human face, and determine the forehead, cheeks, eye sockets and eyeballs occupied in the image (2) Map this flat space to infrared thermal imaging images; (3) Use detection algorithms to collect the highest temperature of the forehead body temperature in the thermal imaging grayscale image, and compare it with the preset value of body temperature , To determine whether the body temperature is abnormal.
  • the method of segmentation judgment in step (1) is as follows: first, first identify the image space occupied by the face, and crop this part of the image into a new image to become a face image; second, send the face image to the cascade classifier , Get the orbital area, upper part, forehead area, lower part, left part, right part, single cheek part.
  • the positions of the eyeballs and noses are identified by the classifier, the eyeball area expands 450%-550% of the eyeball area area, preferably 500%, and the eyeball area is removed as the orbital area; the eyeballs are connected by the middle point of the eyeballs.
  • the line divides the facial image into two parts: the upper part and the lower part; the upper part takes the center point, and the upper part is expanded by 10-20%.
  • 15% of the area is selected as the forehead area; the lower part is based on the nose
  • the middle point is made a line segment perpendicular to the middle point of the eyeball and extended, divided into the left part and the right part; in the left part and the right part, separately take the outermost position of the eyeball area to make a downward vertical line, and
  • the lowermost part of the nose area is intersected by a horizontal line to obtain the position of the center of the cheek, and expand the lower half 10-20% outwards.
  • 15% of the area is selected as the single cheek part.
  • the specific setting of the detection algorithm is that the temperature of the eyeball area is the highest, which is regarded as the highest point of the face temperature; the other parts are in the thermal imaging image, and the median temperature is taken according to the area as the temperature of a single area; among them, the forehead area
  • the temperature difference between the area with the highest temperature of the face is -0.5 to -3 degrees Celsius; the temperature difference between the orbital area and the highest point area of the face is -0.5 to -3 degrees Celsius; the temperature difference between the orbital area and the highest point area of the face should be -2 to -5 Celsius; the temperature difference between the cheek area and the highest point area of the face ranges from -7 to -12 degrees Celsius.
  • the camera and thermal imaging sensor with the same imaging angle and position are used to shoot to further obtain the face image and thermal imaging image.
  • segment the key areas of the human face and eyes to find the plane space occupied by the forehead and cheeks, eye sockets and eyeballs in the image, as shown in Figure 1.
  • This spatial relationship is mapped to the infrared thermal imaging image: the infrared recognition temperature of the eyeball part will be significantly higher than the orbital part, and there is an obvious oval high temperature area.
  • the detection algorithm can be used to capture; the cheek area temperature will be It is significantly lower than the forehead area, as shown in Figure 2.
  • the advantage is that it can be judged by the comparison of regional grayscale averages in the thermal imaging grayscale image, and can be compared with the preset temperature or preset standard grayscale, and the image at abnormal body temperature can be screened out.
  • the thermal imaging images determined as abnormal body temperature can be recorded and stored as a case thermal imaging image data sample library, which is used to assist the comparison and screening of thermal imaging image features of epidemic patients who are expected to be monitored later.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Human Computer Interaction (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Ophthalmology & Optometry (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
  • Measuring And Recording Apparatus For Diagnosis (AREA)

Abstract

一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法，包括采集人脸实时拍摄的可见光图像，基于面部虹膜识别对图像进行人脸区域分割；对分割后人脸区域，进行映射获得热成像图像；通过检测算法识别热成像图像采集分割后人脸区域的对应体温温度；将获得的体温温度，与体温的预设值进行对比，当分割后区域体温分布差异在预设值内，判定为正常；当分割后区域体温分布差异不在预设值内，判定为体温异常，发出报警信号，并对人脸图像样本进行数据库存储、后期追踪及分析。本活体检测方法利用面部虹膜识别和热成像技术组合，不仅能够快速对活体进行识别，识别精度高，而且可记录识别的活体温度，用于流行病筛查。

Description

一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法 技术领域
本发明提供了一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法，属于身份识别技术领域。
背景技术
随着人们生活质量的提高，生活住宅物联网化，办公高楼现代化，交通工具日益发达多样化等等，这些无疑都时时刻刻影响着人们的各个方面，从而人们对大到社会治安小到家庭住所的安全更加担忧；政府部门以及飞机场、高铁站、地铁站、海关等公共场所的人流控制、治安管理、潜在犯罪分析等等需求日益增长；大型公共场所比如体育场、足球场或者金融中心CBD等对人流及身份的监控，都需要用到活体检测技术。
活体检测是在一些身份验证场景确定对象真实生理特征的方法，在人脸识别应用中，活体检测能通过眨眼、张嘴、摇头、点头等组合动作，使用人脸关键点定位和人脸追踪等技术，验证用户是否为真实活体本人操作。可有效抵御照片、换脸、面具、遮挡以及屏幕翻拍等常见的攻击手段，从而帮助用户甄别欺诈行为，保障用户的利益。近些年来，随着3D打印和高精度显示技术的普及，国外有使用工业级设备制造伪装面具，或使用高精度显示器显示动态虹膜图像对生物认证系统进行欺骗攻击的案例。
现有的人脸识别活体检测主要采用以下三种方案：1、使用者配合做出相应动作完成活体检测，其使用者需要配合进行摇头、眨眼、张嘴等等一系列动作，导致用户体验性低；2、利用算法对单个可见光摄像头采集的视频图像进行分析判断，由于摄像头采集的视频图片是二维空间的，所以不管使用什么算法都很难区分面对摄像头的是否是活体人脸，导致识别准确度低；3、利用景深相机对被测者进行3D建模，通过3D模型来判断是否为活体人脸，但这种方法使用的景深相机成本极高，并且3D建模算法需要耗费大量计算机计算能力，运算速度极低。此外，现有技术还存在活体检测的精度有待提高，仅能防照片攻击，并不能防屏幕攻击，防攻击手段单一等问题。因此，亟需开发一种识别精度高，识别速度快，识别结果准确的活体检测方法。
发明内容
技术问题：为了解决现有技术的缺陷，本发明提供了一种识别精度高，识别速度快，识别结果准确的基于面部虹膜识别和热成像技术的人脸识别活体检测方法。
技术方案：本发明提供一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法，包括：
采集人脸实时拍摄的可见光图像，基于面部虹膜识别对图像进行人脸区域分割；
对分割后人脸区域，进行映射获得热成像图像；
通过检测算法对热成像图像，采集分割后人脸区域的对应体温温度；
将获得的体温温度，与体温的预设值进行对比，当分割后区域体温分布差异在预设值内，判定为正常；
当分割后区域体温分布差异在预设值内，判定为体温异常，发出报警信号，并对应人脸图像样本进行数据库存储、后期追踪及分析。
作为另一种改进，设置成像角度和位置一致的摄像头和热成像传感器；
利用摄像头拍摄人脸图像和虹膜图像，利用热成像传感器生成热成像图像；
利用虹膜图像识别被测试目标的虹膜、眼眶、眼球和面部之间的相对距离，形成虹膜、眼眶、眼球、面部的点集合；
利用热成像图像识别图像上温度；
在热成像图像上叠加虹膜、眼眶、眼球、面部的点集合，利用热成像图像判断被测试目标的虹膜、眼眶、眼球和面部之间是否存在温度分布差异；
当热成像图像识别到的图像温度高于警报值时，发出报警信号；
当虹膜、眼眶、眼球和面部之间温度分布差异低于预设值时，判定所述测试目标为非活体人脸；
当虹膜、眼眶、眼球和面部之间温度分布差异高于预设值时，判定所测试目标为活体人脸。
作为改进，成像传感器的镜头参数与面部捕捉摄像头的镜头参数设置一致。
作为另一种改进，记录热成像图像识别到的该副图的最高温度，用于流行病筛查。
作为另一种改进，将判定为体温异常的热成像图像进行记录处理，存储为病例热成像图像数据样本库，用于辅助后期待监测的待确诊流行病人的热成像图像特征对比、筛查。
作为另一种改进，将判定为体温异常的人脸图像进行记录处理，存储为病例可见光图像数据样本库，用于后期流行病病人的跟踪。
作为另一种改进，判断温度分布的具体步骤为：(1)对拍摄人脸获得的可见光图像， 进行人脸、眼部及脸颊、眼眶、眼球区域的分割，确定前额、脸颊、眼眶及眼球在该图像中所占据的平面空间；(2)将这一平面空间进行映射到红外热成像图像中；(3)利用检测算法在热成像灰度图像中，采集温度最高的前额体温，与体温的预设值进行对比，判定体温是否异常。
作为另一种改进，步骤(1)中进行分割判定的方法为：第一，先识别面部所占有的图像空间，将这部分图像裁剪为一个新的图像成为面部图像；第二，将面部图像送入级联分类器，获得眼眶区域、上半部分、前额区域、下半部分、左部分、右部分、单面脸颊部分。
作为另一种改进，其中，通过分类器识别其中的眼球、鼻子的位置，眼球区域向外扩张眼球区域面积的450％-550％并去除眼球区域作为眼眶区域；通过两眼眼球中间点连线将面部图像分划为上半部分、下两部分；上半部分取中心点，向外扩上半部分10-20％的面积作为前额区域；下半部分在以鼻子中间点做垂直于眼球中间点连线的线段并延长，分割为左部分、右部分；在左部分、右部分中，分别单独地取眼球区域最靠外的位置做向下的垂线，与鼻子区域最靠下部位做水平线相交，取得脸颊中心点位置，并向外扩张下半部分10-20％的面积作为单面脸颊部分。
作为另一种改进，步骤(3)中，体温的预设值设定时，眼球区域温度最高，作为面部温度最高点；其它部位在热成像图像中，按面积取温度中位值，作为单个区域的温度；其中前额区域与面部温度最高点区域温差范围为-0.5到-3摄氏度；眼眶区域与面部温度最高点区域温差范围应为-0.5到-3摄氏度；眼眶区域与面部温度最高点区域温差范围应为-2到-5摄氏度；脸颊区域与面部温度最高点区域温差范围应为-7到-12摄氏度。
有益效果：本发明提供的活体检测方法利用面部虹膜识别和热成像技术组合，不仅能够快速对活体进行识别，识别精度高，而且可记录识别的活体温度，用于流行病筛查。同时，使用热成像图像作为面部和虹膜特征检验之前的活体验证，可以有效避攻击者使用高精度打印的面部或虹膜图像或三维面部模型、眼球模型、动态图像显示等方法对识别系统进行伪造攻击。
另外，一方面，能够通过记录体温异常的人员可见光图像即人脸图像的面部特征图像，有利于后期进行传染源的跟踪、监督，减少传染源的扩散，避免人员伤亡。另一方面，能够通过记录体温异常的人员的热成像图像，进行病例数据库的建立，对后期待确诊流行病人的热成像图像特征对比、筛查时，提高了效率，提供了医学借鉴依据。
附图说明
图1为实施例1人脸图像分割区域示意图。
图2为实施例1对应红外热成像图片像进行温度分布图。
具体实施方式
下面对本发明作出进一步说明。
一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法，包括：采集人脸实时拍摄的可见光图像，基于面部虹膜识别对图像进行人脸区域分割；
对分割后人脸区域，进行映射获得热成像图像；
通过检测算法对热成像图像，采集分割后人脸区域的对应体温温度；
将获得的体温温度，与体温的预设值进行对比，当分割后区域体温分布差异在预设值内，判定为正常；
当分割后区域体温分布差异在预设值内，判定为体温异常，发出报警信号，并对应人脸图像样本进行数据库存储、后期追踪及分析。
作为本发明的具体实施方式，进上述方法，设置成像角度和位置一致的摄像头和热成像传感器；
利用摄像头拍摄人脸图像和虹膜图像，利用热成像传感器生成热成像图像；
利用虹膜图像识别被测试目标的虹膜、眼眶、眼球和面部之间的相对距离，形成虹膜、眼眶、眼球、面部的点集合；
利用热成像图像识别图像上温度；
在热成像图像上叠加虹膜、眼眶、眼球、面部的点集合，利用热成像图像判断被测试目标的虹膜、眼眶、眼球和面部之间是否存在温度分布差异；
当热成像图像识别到的图像温度高于警报值时，发出报警信号；
当虹膜、眼眶、眼球和面部之间温度分布差异低于预设值时，判定所述测试目标为非活体人脸；
当虹膜、眼眶、眼球和面部之间温度分布差异高于预设值时，判定所测试目标为活体人脸。
其中成像传感器的镜头参数与面部捕捉摄像头的镜头参数设置一致。作为本发明的具体实施方式，记录热成像图像识别到的该副图的最高区域温度，用于流行病筛查。
将判定为体温异常的热成像图像进行记录处理，存储为病例热成像图像数据样本 库，用于辅助后期待监测的待确诊流行病人的热成像图像特征对比、筛查。
将判定为体温异常的人脸图像进行记录处理，存储为病例可见光图像数据样本库，用于后期流行病病人的跟踪。
判断温度分布的具体步骤为：(1)对拍摄人脸获得的可见光图像，进行人脸、眼部及脸颊、眼眶、眼球区域的分割，确定前额、脸颊、眼眶及眼球在该图像中所占据的平面空间；(2)将这一平面空间进行映射到红外热成像图像中；(3)利用检测算法在热成像灰度图像中，采集温度最高的前额体温，与体温的预设值进行对比，判定体温是否异常。
步骤(1)中进行分割判定的方法为：第一，先识别面部所占有的图像空间，将这部分图像裁剪为一个新的图像成为面部图像；第二，将面部图像送入级联分类器，获得眼眶区域、上半部分、前额区域、下半部分、左部分、右部分、单面脸颊部分。
其中，通过分类器识别其中的眼球、鼻子的位置，眼球区域向外扩张眼球区域面积的450％-550％，优选地为500％，并去除眼球区域作为眼眶区域；通过两眼眼球中间点连线将面部图像分划为上半部分、下两部分；上半部分取中心点，向外扩上半部分10-20％，优选的选择15％的面积作为前额区域；下半部分在以鼻子中间点做垂直于眼球中间点连线的线段并延长，分割为左部分、右部分；在左部分、右部分中，分别单独地取眼球区域最靠外的位置做向下的垂线，与鼻子区域最靠下部位做水平线相交，取得脸颊中心点位置，并向外扩张下半部分10-20％，优选的选择15％的面积作为单面脸颊部分。
步骤(3)中，检测算法的具体设定为，眼球区域温度最高，作为面部温度最高点；其它部位在热成像图像中，按面积取温度中位值，作为单个区域的温度；其中前额区域与面部温度最高点区域温差范围为-0.5到-3摄氏度；眼眶区域与面部温度最高点区域温差范围为-0.5到-3摄氏度；眼眶区域与面部温度最高点区域温差范围应-2到-5摄氏度；脸颊区域与面部温度最高点区域温差范围为-7到-12摄氏度。
实施例1
利用设置成像角度和位置一致的摄像头和热成像传感器，进行拍摄，进一步地获得人脸图像和热成像图像。首先，在可见光图像中，对人脸和眼部重点区域进行分割，找到前额和脸颊，眼眶和眼球在图像中所占据的平面空间，见图1所示。将此空间关系映射到红外热成像图像中：其中眼球部分的红外识别温度会显著高于眼眶部分，存在明显的椭圆形高温区域，图像二值化后可利用检测算法进行捕捉；脸颊区域温度会明显低于 前额区域，见图2中。
上述测量方法中，优势为可在热成像灰度图像中通过区域灰度均值对比判断，进可实现与预设值温度或预设的标准灰度进行对比，筛选出异常体温下的图像。
然后，可以将判定为体温异常的热成像图像进行记录处理，存储为病例热成像图像数据样本库，用于辅助后期待监测的待确诊流行病人的热成像图像特征对比、筛查。
还可以将判定为体温异常的人脸图像进行记录处理，存储为病例可见光图像数据样本库，用于后期流行病病人的跟踪。
以上所述实施例仅表达了本发明的几种实施方式，其描述较为具体和详细，但并不能因此而理解为对发明专利范围的限制。应当指出的是，对于本领域的普通技术人员来说，在不脱离本发明构思的前提下，还可以做出若干变形和改进，这些都属于本发明的保护范围。因此，本发明专利的保护范围应以所附权利要求为准。

Claims (10)

  1. 一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法，其特征在于：包括：
    采集人脸实时拍摄的可见光图像，基于面部虹膜识别对图像进行人脸区域分割；
    对分割后人脸区域，进行映射获得热成像图像；
    通过检测算法对热成像图像，采集分割后人脸区域的对应体温温度；
    将获得的体温温度，与体温的预设值进行对比，当分割后区域体温分布差异在预设值内，判定为正常；
    当分割后区域体温分布差异在预设值内，判定为体温异常，发出报警信号，并对应人脸图像样本进行数据库存储、后期追踪及分析。
  2. 根据权利要求1所述一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法，其特征在于：
    设置成像角度和位置一致的摄像头和热成像传感器；
    利用摄像头拍摄人脸图像和虹膜图像，利用热成像传感器生成热成像图像；
    利用虹膜图像识别被测试目标的虹膜、眼眶、眼球和面部之间的相对距离，形成虹膜、眼眶、眼球、面部的点集合；
    利用热成像图像识别图像上温度；
    在热成像图像上叠加虹膜、眼眶、眼球、面部的点集合，利用热成像图像判断被测试目标的虹膜、眼眶、眼球和面部之间是否存在温度分布差异；
    当热成像图像识别到的图像温度高于警报值时，发出报警信号；
    当虹膜、眼眶、眼球和面部之间温度分布差异低于预设值时，判定所述测试目标为非活体人脸；
    当虹膜、眼眶、眼球和面部之间温度分布差异高于预设值时，判定所测试目标为活体人脸。
  3. 根据权利要求2所述一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法，其特征在于：成像传感器的镜头参数与面部捕捉摄像头的镜头参数设置一致。
  4. 根据权利要求2所述一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法，其特征在于：记录热成像图像识别到的该副图的最高温度，用于流行病筛查。
  5. 根据权利要求4所述一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法，其特征在于：将判定为体温异常的热成像图像进行记录处理，存储为病例热成像图像数据样本库，用于辅助后期待监测的待确诊流行病人的热成像图像特征对比、筛查。
  6. 根据权利要求5所述一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法，其特征在于：将判定为体温异常的人脸图像进行记录处理，存储为病例可见光图像数据样本库，用于后期流行病病人的跟踪。
  7. 根据权利要求2所述一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法，其特征在于：判断温度分布的具体步骤为：(1)对拍摄人脸获得的可见光图像，进行人脸、眼部及脸颊、眼眶、眼球区域的分割，确定前额、脸颊、眼眶及眼球在该图像中所占据的平面空间；(2)将这一平面空间进行映射到红外热成像图像中；(3)利用检测算法在热成像灰度图像中，采集温度最高的前额体温，与体温的预设值进行对比，判定体温是否异常。
  8. 根据权利要求7所述一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法，其特征在于：步骤(1)中进行分割判定的方法为：第一，先识别面部所占有的图像空间，将这部分图像裁剪为一个新的图像成为面部图像；第二，将面部图像送入级联分类器，获得眼眶区域、上半部分、前额区域、下半部分、左部分、右部分、单面脸颊部分。
  9. 根据权利要求8所述一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法，其特征在于：其中，通过分类器识别其中的眼球、鼻子的位置，眼球区域向外扩张眼球区域面积的450％-550％并去除眼球区域作为眼眶区域；通过两眼眼球中间点连线将面部图像分划为上半部分、下两部分；上半部分取中心点，向外扩上半部分10-20％的面积作为前额区域；下半部分在以鼻子中间点做垂直于眼球中间点连线的线段并延长，分割为左部分、右部分；在左部分、右部分中，分别单独地取眼球区域最靠外的位置做向下的垂线，与鼻子区域最靠下部位做水平线相交，取得脸颊中心点位置，并向外扩张下半部分10-20％的面积作为单面脸颊部分。
  10. 根据权利要求7所述一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法，其特征在于：步骤(3)中，检测算法的具体设定为，眼球区域温度最高，作为面部温度最高点；其它部位在热成像图像中，按面积取温度中位值，作为单个区域的温度；其中前额区域与面部温度最高点区域温差范围为-0.5到-3摄氏度；眼眶区域与面部温度最高点区域温差范围为-0.5到-3摄氏度；眼眶区域与面部温度最高点区域温差范围应-2到-5摄氏度；脸颊区域与面部温度最高点区域温差范围为-7到-12摄氏度。
PCT/CN2020/091718 2020-05-13 2020-05-22 一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法 WO2021227124A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010400545.7A CN111797677B (zh) 2020-05-13 2020-05-13 一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法
CN202010400545.7 2020-05-13

Publications (1)

Publication Number Publication Date
WO2021227124A1 true WO2021227124A1 (zh) 2021-11-18

Family

ID=72806106

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/091718 WO2021227124A1 (zh) 2020-05-13 2020-05-22 一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法

Country Status (2)

Country Link
CN (1) CN111797677B (zh)
WO (1) WO2021227124A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117475502A (zh) * 2023-12-27 2024-01-30 成都科瑞特电气自动化有限公司 基于矿用的虹膜和人脸融合识别方法及系统

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB2603467A (en) * 2021-01-28 2022-08-10 Sita Advanced Travel Solutions Ltd A method and system for screening a user's temperature
CN113408377A (zh) * 2021-06-03 2021-09-17 山东交通学院 基于温度信息的人脸活体检测方法
CN113313057A (zh) * 2021-06-16 2021-08-27 山东省科学院激光研究所 一种人脸活体检测与识别系统
CN113553928B (zh) * 2021-07-13 2024-03-22 厦门瑞为信息技术有限公司 一种人脸活体检测方法、系统和计算机设备
CN114894337B (zh) * 2022-07-11 2022-09-27 深圳市大树人工智能科技有限公司 一种用于室外人脸识别测温方法及装置
CN117409488A (zh) * 2023-10-20 2024-01-16 湖南远图网络科技有限公司 一种用户身份识别方法、系统、设备及存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105138996A (zh) * 2015-09-01 2015-12-09 北京上古视觉科技有限公司 一种具有活体检测功能的虹膜识别系统
CN106384019A (zh) * 2015-11-19 2017-02-08 小红象医疗科技有限公司 基于中医临床全息面诊的红外医学自动分析系统和方法
CN106845345A (zh) * 2016-12-15 2017-06-13 重庆凯泽科技股份有限公司 活体检测方法及装置
US9886640B1 (en) * 2016-08-08 2018-02-06 International Business Machines Corporation Method and apparatus to identify a live face image using a thermal radiation sensor and a visual radiation sensor
CN110411570A (zh) * 2019-06-28 2019-11-05 武汉高德智感科技有限公司 基于人体检测和人体跟踪技术的红外人体温度筛查方法

Family Cites Families (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7457441B2 (en) * 2005-02-25 2008-11-25 Aptina Imaging Corporation System and method for detecting thermal anomalies
DE102015104768B4 (de) * 2015-03-27 2023-03-16 Bundesdruckerei Gmbh Zugangssystem mit Personenerkennungsvorrichtung
CN104809457A (zh) * 2015-05-26 2015-07-29 牟永敏 一种基于区域化隐函数特征的三维人脸识别方法及系统
KR101729327B1 (ko) * 2015-11-10 2017-04-21 홍미선 듀얼 카메라를 이용한 체온 추적 모니터링 시스템
CN106650666A (zh) * 2016-12-26 2017-05-10 北京旷视科技有限公司 活体检测的方法及装置
CN108694354A (zh) * 2017-04-10 2018-10-23 上海聚虹光电科技有限公司 一种虹膜采集装置采集人脸图像的应用方法
CN107392165B (zh) * 2017-07-31 2020-02-14 南京中科道置智能科技有限公司 面部虹膜识别系统
CN207867523U (zh) * 2017-12-07 2018-09-14 秦家早 活体人脸检测设备
CN108764058B (zh) * 2018-05-04 2021-05-25 吉林大学 一种基于热成像效应的双摄像头人脸活体检测方法
CN109325462B (zh) * 2018-10-11 2021-03-12 深圳斐视沃德科技有限公司 基于虹膜的人脸识别活体检测方法及装置
CN110472519B (zh) * 2019-07-24 2021-10-29 杭州晟元数据安全技术股份有限公司 一种基于多模型的人脸活体检测方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105138996A (zh) * 2015-09-01 2015-12-09 北京上古视觉科技有限公司 一种具有活体检测功能的虹膜识别系统
CN106384019A (zh) * 2015-11-19 2017-02-08 小红象医疗科技有限公司 基于中医临床全息面诊的红外医学自动分析系统和方法
US9886640B1 (en) * 2016-08-08 2018-02-06 International Business Machines Corporation Method and apparatus to identify a live face image using a thermal radiation sensor and a visual radiation sensor
CN106845345A (zh) * 2016-12-15 2017-06-13 重庆凯泽科技股份有限公司 活体检测方法及装置
CN110411570A (zh) * 2019-06-28 2019-11-05 武汉高德智感科技有限公司 基于人体检测和人体跟踪技术的红外人体温度筛查方法

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117475502A (zh) * 2023-12-27 2024-01-30 成都科瑞特电气自动化有限公司 基于矿用的虹膜和人脸融合识别方法及系统
CN117475502B (zh) * 2023-12-27 2024-03-15 成都科瑞特电气自动化有限公司 基于矿用的虹膜和人脸融合识别方法及系统

Also Published As

Publication number Publication date
CN111797677B (zh) 2024-04-02
CN111797677A (zh) 2020-10-20

Similar Documents

Publication Publication Date Title
WO2021227124A1 (zh) 一种基于面部虹膜识别和热成像技术的人脸识别活体检测方法
CN105138954B (zh) 一种图像自动筛选查询识别系统
CN104951773B (zh) 一种实时人脸识别监视系统
CN102201061B (zh) 基于多阶层过滤人脸识别的智能安全监控系统及方法
CN105956572A (zh) 一种基于卷积神经网络的活体人脸检测方法
CN107967458A (zh) 一种人脸识别方法
CN108009482A (zh) 一种提高人脸识别效率方法
CN104091176A (zh) 人像比对在视频中的应用技术
WO2004070563A2 (en) Three-dimensional ear biometrics system and method
CN109508706B (zh) 一种基于微表情识别和无感人脸识别的静默活体检测方法
CN111126219A (zh) 一种基于人工智能的变电站人员的身份识别系统和方法
Ancheta et al. FEDSecurity: implementation of computer vision thru face and eye detection
US20210134146A1 (en) Tracking and alerting traffic management system using iot for smart city
CN114463828B (zh) 基于人证统一的监考方法及系统、电子设备和存储介质
CN111862413A (zh) 实现抗疫情非接触多维身份快速识别的方法及其系统
Ulleri et al. Development of contactless employee management system with mask detection and body temperature measurement using TensorFlow
CN107862298B (zh) 一种基于红外摄像装置下眨眼的活体检测方法
Eyiokur et al. A survey on computer vision based human analysis in the COVID-19 era
RU2316051C2 (ru) Способ и система автоматической проверки присутствия лица живого человека в биометрических системах безопасности
Abd et al. Human fall down recognition using coordinates key points skeleton
Kasturi et al. Classification of human fall in top Viewed kinect depth images using binary support vector machine
Kadam et al. Review paper on contactless attendance system based on face recognition
Fawwaz et al. Vision application of LabVIEW: IGUI for face and pattern detection in real time
Kennedy et al. Implementation of an Embedded Masked Face Recognition System using Huskylens System-On-Chip Module
Menon Leveraging Facial Recognition Technology in Criminal Identification

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: 20935697

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20935697

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 20935697

Country of ref document: EP

Kind code of ref document: A1

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 28.06.2023)

122 Ep: pct application non-entry in european phase

Ref document number: 20935697

Country of ref document: EP

Kind code of ref document: A1