WO2022057719A1 - 识别对象确定方法、装置、设备及存储介质 - Google Patents

识别对象确定方法、装置、设备及存储介质 Download PDF

Info

Publication number
WO2022057719A1
WO2022057719A1 PCT/CN2021/117383 CN2021117383W WO2022057719A1 WO 2022057719 A1 WO2022057719 A1 WO 2022057719A1 CN 2021117383 W CN2021117383 W CN 2021117383W WO 2022057719 A1 WO2022057719 A1 WO 2022057719A1
Authority
WO
WIPO (PCT)
Prior art keywords
face
size value
preset
face size
recognition
Prior art date
Legal status (The legal status 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 status listed.)
Ceased
Application number
PCT/CN2021/117383
Other languages
English (en)
French (fr)
Inventor
康家梁
吴文川
傅宜生
沈玺
卞凯
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China Unionpay Co Ltd
Original Assignee
China Unionpay Co Ltd
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 China Unionpay Co Ltd filed Critical China Unionpay Co Ltd
Publication of WO2022057719A1 publication Critical patent/WO2022057719A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING OR CALCULATING; 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
    • G06V40/161Detection; Localisation; Normalisation
    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00Scenes; Scene-specific elements
    • G06V20/50Context or environment of the image
    • G06V20/52Surveillance or monitoring of activities, e.g. for recognising suspicious objects

Definitions

  • the present disclosure relates to the technical field of computer vision, and in particular, to a method, apparatus, device, and storage medium for determining a recognized object.
  • face recognition is widely used in production, finance, security, transportation and other fields due to its accuracy, safety, convenience and many other characteristics.
  • attendance machines unmanned retail machines, access control systems, etc.
  • Embodiments of the present disclosure provide a method, apparatus, device, and storage medium for determining an identification object, which can improve the accuracy of determining an identification object.
  • an embodiment of the present disclosure provides a method for determining an identification object, the method comprising:
  • the shooting preview image includes at least two faces
  • the feature information includes a face size value
  • the face size value of each face it is determined that the face whose face size value satisfies the preset recognition condition is the recognition object.
  • an embodiment of the present disclosure provides an apparatus for determining an identification object, the apparatus comprising:
  • an acquisition module configured to acquire a shooting preview image, wherein the shooting preview image includes at least two faces;
  • an extraction module configured to extract feature information of each of the at least two faces in the shot preview image, wherein the feature information includes a face size value
  • the determining module is used for determining, according to the face size value of each face, the face whose face size value satisfies the preset recognition condition as the recognition object.
  • embodiments of the present disclosure provide a device for determining an identified object, the device includes: a processor and a memory storing computer program instructions; the processor implements the first aspect or any of the first aspects can be implemented when the processor executes the computer program instructions The identification object determination method described in the method.
  • an embodiment of the present disclosure provides a computer-readable storage medium, where computer program instructions are stored on the computer-readable storage medium, and when the computer program instructions are executed by a processor, the first aspect or any implementation manner of the first aspect is implemented The identification object determination method described in .
  • the face size value of each face in the shot preview image is extracted, and the face size value is determined according to the face size value of each face.
  • the face that meets the preset recognition conditions is the recognition object.
  • a suitable target face can be selected as the recognition object in the face collection process, so as to realize the accurate determination of the recognition object and improve the accuracy of the recognition object determination.
  • FIG. 1 is a schematic structural diagram of a system for determining an object for identifying an embodiment of the system for determining an object for identifying a first aspect of the present disclosure
  • Fig. 2 is a schematic flow chart of a method for determining an identification object according to an embodiment of the method for determining an identification object in the first aspect of the present disclosure
  • FIG. 3 is a schematic flowchart of another identification object determination method according to an embodiment of the identification object determination method of the first aspect of the present disclosure
  • FIG. 4 is a schematic diagram of a region division according to an embodiment of the method for determining an identification object according to the first aspect of the present disclosure
  • FIG. 5 is a schematic diagram of a relative position of an embodiment of the method for determining an identified object according to the first aspect of the present disclosure
  • FIG. 6 is a schematic structural diagram of an apparatus for determining an identification object provided by an embodiment of the apparatus for determining an identification object in the second aspect of the present disclosure
  • FIG. 7 is a schematic structural diagram of a recognition object determination device provided by an embodiment of the recognition object determination device of the third aspect of the present disclosure.
  • the traditional identification object determination scheme mainly reduces the occurrence of multiple faces on the display screen by optimizing the business process and adjusting the deployment angle of the shooting equipment.
  • the first face that appears on the screen is obtained as a follow-up business execution. Identify objects in the process.
  • face recognition scenarios with high traffic flow such as face-swiping payment, face-swiping entry, and face-swiping through gates, multiple faces often appear on the display screen. Therefore, misrecognition may occur, that is, the "bystander" or "behind” face is used as the recognition object, which affects the user experience.
  • the embodiments of the present disclosure provide a method, apparatus, device, and storage medium for determining an identification object.
  • a method, apparatus, device, and storage medium for determining an identification object.
  • FIG. 1 is a schematic structural diagram of a system for determining an object for identifying an embodiment of the system for determining an object for identifying the first aspect of the present disclosure.
  • the identification object determination system may include a photographing device 110 and an electronic device 120 , where an example of the photographing device 110 may be a camera, a device installed with a camera module, and the like.
  • An example of the electronic device 120 may be a mobile electronic device or a non-mobile electronic device.
  • the mobile electronic device may be a mobile phone, a tablet computer, a notebook computer, a palmtop computer, an Ultra-Mobile Personal Computer (UMPC), a netbook or a personal digital assistant (PDA), etc.
  • UMPC Ultra-Mobile Personal Computer
  • PDA personal digital assistant
  • the device can be a server, a Network Attached Storage (NAS), a Personal Computer (PC), a teller machine or a self-service machine, and so on.
  • NAS Network Attached Storage
  • PC Personal Computer
  • teller machine or a self-service machine, and so on.
  • the photographing device 110 may be integrated into the electronic device 120 as a module.
  • the recognition object determination system can be applied to face recognition scenarios such as face-swiping payment, face-swiping entry, and face-swiping check-in.
  • face recognition scenarios such as face-swiping payment, face-swiping entry, and face-swiping check-in.
  • the electronic device 120 may acquire a shooting preview image of the user by the shooting device 110, wherein the shooting preview image includes at least two users, that is, at least two human faces.
  • the shooting preview image includes at least two users, that is, at least two human faces.
  • feature information of each face in the shot preview image is extracted, wherein the feature information includes a face size value.
  • the face size value of each face the face whose face size value meets the preset recognition condition is determined as the recognition object, so as to execute the subsequent face recognition service.
  • the execution body of the method for determining an identified object may be the electronic device 120 in the identification object determination system shown in FIG. 1 , or a module in the electronic device 120 .
  • FIG. 2 is a schematic flowchart of a method for determining an identification object according to an embodiment of the method for determining an identification object of the first aspect of the present disclosure. As shown in FIG. 2 , the method for determining an identification object may include the following steps:
  • An example of the captured preview image may include at least two human faces, which may be captured on-site by the capturing device.
  • the shooting preview image may be a previewed screen input frame, that is, an image displayed on the display screen when the shooting device shoots.
  • S220 Extract feature information of each of the at least two faces in the shot preview image.
  • Examples of feature information may include face size values.
  • the face size value may include the face-eye distance value or the number of face pixels, etc., wherein the face-eye distance value is used to represent the face size value, which can reduce the noise of the face size difference.
  • a face recognition algorithm may be used to perform preliminary face recognition on the captured preview image, and the human face in it may be recognized. Then perform feature extraction on the recognized faces to obtain feature information of each face.
  • the recognition object will be used for face recognition.
  • the face that is the object of recognition is usually the closest to the photographing device than other photographed faces. Therefore, in some embodiments, the face with the largest face size value may be determined as the recognition object, so as to realize accurate determination of the recognition object.
  • the face as the recognition target has a size advantage over other faces.
  • the ratio of the face size value to the largest face size value of each of the at least two faces except the face with the largest face size value may be calculated separately.
  • the first preset ratio threshold can be flexibly set according to the actual situation, for example, it can be 60%.
  • the at least one ratio is less than or equal to the first preset ratio threshold
  • the duration of displaying the face corresponding to the maximum face size value on the display screen can be obtained.
  • the display duration of the face corresponding to the maximum face size value is greater than or equal to the preset duration threshold
  • the face corresponding to the maximum face size value is determined as the recognition object.
  • face tracking of consecutive frames may be performed on the face corresponding to the maximum face size value.
  • a unique face identification can be assigned to the face corresponding to the maximum face size value.
  • the continuous display duration of the face corresponding to the maximum face size value can be counted.
  • T represents a preset duration threshold, which may be an absolute duration, such as 800ms. It can also be a relative duration, such as the time to complete a period of business processing.
  • the duration of passing through the channel of the designated shooting area that is, the length of time from entering to leaving the specific shooting area (or not leaving the coverage of the shooting equipment); another example is to collect continuous frame images to a certain number (such as 5 Zhang) time; another example is the transaction response time (or the time to obtain a recognition transaction result), that is, when starting to track the face corresponding to the maximum face size value, the recognition is initiated until the duration of the response result is received. During this period, the face corresponding to the maximum face size value is always displayed on the display screen.
  • the first position of the face corresponding to the maximum face size value may be obtained.
  • the face corresponding to the maximum face size value is determined as the recognition object, that is, the position is introduced as a judgment factor based on the face size value to improve the accurate determination of the recognition object.
  • the preset position condition may include: the first position matches the preset position, or the first position is located in a preset area. It can be understood that the preset position and the preset area can be flexibly set according to the actual situation, for example, they can be pre-selected according to the debugging experience when the scene layout is implemented.
  • the face size value of each face in the shot preview image by extracting the face size value of each face in the shot preview image, and according to the face size value of each face, it is determined that the face whose face size value satisfies the preset recognition condition is the recognition object.
  • a suitable target face can be selected as the recognition object during the face collection process, and the accuracy of the recognition object determination can be improved.
  • the method may further include:
  • the first preset ratio threshold and the second preset ratio threshold can be adjusted according to the accuracy rate determined by the identification object after a period of time.
  • completing the business process that is, identifying the object and determining the correct face picture, can be selected as a positive sample.
  • the accuracy rate determined by the recognition object is counted, and the first preset ratio threshold and the second preset ratio threshold are adjusted according to the accuracy, so as to realize dynamic adjustment of the threshold.
  • the face corresponding to the maximum face size value is determined as the recognition object.
  • the second preset ratio threshold and the position are introduced for further judgment, and the conditions for determining the identification object are refined, which can further improve the accuracy of determining the identification object.
  • the feature information may further include face angle, face occlusion information, face eye information, and the like. Further, at least two faces can be screened according to the face angle, face occlusion information, and face and eye information, that is, the face angle, face occlusion information, and face and eye information are used as judgment factors to determine the person in the preview image. Faces are screened, and the faces that meet the corresponding conditions are retained.
  • At least two faces can be screened according to face size value, face angle, face occlusion information, and face eye information. And according to the face size value of the screened face, it is determined that the face whose face size value satisfies the preset recognition condition is the recognition object.
  • face size value of the screened face it is determined that the face whose face size value satisfies the preset recognition condition is the recognition object.
  • the accuracy of determining the recognition object can be improved.
  • the method for determining an identification object provided by the embodiment of the present disclosure will be described in detail by taking the method for determining an identification object applied to a face-swiping payment scenario as an example. As shown in FIG. 3 , the method may include:
  • feature information of each face can be extracted first, and the feature information includes face size value, face angle, face occlusion information, and face eye information.
  • the face size value can be judged for each face, and the faces whose face size value is greater than or equal to the preset size threshold are retained.
  • the face size value is represented by the number of face pixels
  • the preset size threshold may be set to 100*100
  • 100*100 represents the width and height of the face pixels.
  • face angle judgment may be performed, and faces whose face angles are less than or equal to a preset angle threshold are retained.
  • the face angle includes a roll angle (roll), a pitch angle (pitch), and a yaw angle (yaw)
  • the preset angle thresholds corresponding to the three types of angles may be ⁇ 20°
  • the three types of angles are reserved to be less than or equal to A face equal to ⁇ 20°.
  • the face occlusion information can be judged, and the faces whose face occlusion information meets the face occlusion condition are reserved.
  • the face occlusion conditions can be flexibly set according to business requirements, such as no occlusion at all, no occlusion at key points, no occlusion area exceeding a certain percentage, and so on.
  • the identification object determination method further includes S304 , judging whether the screened face is unique.
  • the identification object determination method further includes S305, acquiring the duration of the unique face displayed on the display screen.
  • the identification object determination method further includes S306 , judging whether the display duration of the unique face is greater than or equal to a preset duration threshold.
  • the identification object determination method further includes S307 , determining a unique face as an identification object.
  • the identification object determination method further includes S308 , respectively calculating the ratio of the face size value to the maximum face size value of each face except the face with the largest face size value.
  • the identification object determination method further includes S309 , judging whether the ratios are all less than or equal to a first preset ratio threshold.
  • the method for determining an identification object further includes S310 , acquiring the duration of time that the face corresponding to the maximum face size value is displayed on the display screen.
  • the identification object determination method further includes S311 , judging whether the display duration is greater than or equal to a preset duration threshold.
  • the identification object determination method further includes S312 , determining the face corresponding to the maximum face size value as the identification object.
  • the identification object determination method further includes S313 , judging whether the ratios are all less than or equal to a second preset ratio threshold.
  • the identification object determination method further includes S314 , obtaining a first position of the face corresponding to the maximum face size value.
  • the first position may include an absolute position of the face corresponding to the largest face size value, and a relative position using the face corresponding to the second largest face size value as a reference point.
  • the shot preview image may be divided according to a preset area division rule, a position marker of the face may be determined according to the divided area, and the absolute position of the face may be represented by the position mark.
  • the area division rule may be to divide the shot preview image according to the center point, and divide it into four quadrants: the first quadrant, the second quadrant, the third quadrant, and the fourth quadrant according to the mathematical plane coordinate system.
  • taking the center point as the origin taking half the width of the preview image as the width, and taking half the height of the preview image as the height, as the center area C, wherein the center area C and the quadrant area overlap.
  • a plane coordinate system can be constructed with the center point as the origin (0, 0). For the face in the shot preview image, calculate the coordinates (x, y) of the center point of the face, and determine the position mark of the face according to the coordinates of the center point of the face and the divided area.
  • the position marker can be LG, where L represents the quadrant where the coordinates of the center point of the face are located, and takes a value of 1, 2, 3 or 4, and when L is 1, it means it is located in the first quadrant.
  • G indicates whether the coordinates of the center point of the face are located in the central area C, and takes a value of 0 or 1.
  • the location includes the coordinates of the center point on the boundary.
  • the position marks may be as follows: not located in the central area C: the first quadrant: 10, the second quadrant: 20, the third quadrant: 30, the fourth quadrant: 40; located in the central area C: the first quadrant: 11, the first Second quadrant: 21, third quadrant: 31, fourth quadrant: 41.
  • the position marker 10 indicates that the coordinates of the center point of the face are located in the first quadrant and are not located in the center area.
  • the position marker 11 indicates that the coordinates of the center point of the face are located in the center area and the first quadrant.
  • the center coordinate of the face 1 is located in the third quadrant, and is not located in the center area C, so the position is marked as 30.
  • the center coordinates of the face 2 are located in the first and second quadrants, and are located in the center area C, then the positions are marked as 11 and 21.
  • the center coordinates of the face 3 are located in the first quadrant and in the center area C, and the position is marked as 11. It can be understood that the area division rules can be flexibly adjusted according to actual needs, which is not limited here.
  • the relative position can be calculated using a relative position calculation formula.
  • the relative position calculation formula can be as follows:
  • A(X, Y) O 1 (x 1 , y 1 )-O 2 (x 2 , y 2 ) (1)
  • A(X, Y) represents a vector
  • O 1 (x 1 , y 1 ) represents the face center point coordinates of the second largest face
  • O 2 (x 2 , y 2 ) represents the face center point coordinates of the largest face
  • X>0 means that the largest face is on the horizontal axis, on the left side of the second largest face, and vice versa
  • Y>0 means the largest face is on the vertical axis, on the second largest face below, and vice versa above.
  • the identification object determination method further includes S315 , judging whether the first position satisfies a preset position condition.
  • the preset position condition includes: the first position matches the preset position, or the first position is located in a preset area.
  • the absolute and relative positions of the recognition objects on the display screen often appear to have a certain degree of aggregation. It can be understood that the location of the gathering is different depending on the scene or the angle of the shooting equipment. In the scene application, a certain number (such as 100,000) of face recognition scene images with multiple faces can be selected, data processing is performed on them, the recognition object and the second largest face are marked, and the relative position and absolute position of the recognition object can be counted.
  • Position that is, classify each image, count the absolute position LC value, and the relative position A (X, Y), according to the statistical results, select the LC value or combination of LC values with the largest number in the scene, and the relative position, Confirm the absolute position and relative position of the recognition object that often appear in the scene as the preset position, or determine the preset area according to multiple frequently appearing absolute positions and relative positions.
  • the identification object determination method further includes S316 , determining the face corresponding to the maximum face size value as the identification object.
  • the identification object determination method further includes S317 , prompting that the identification object cannot be determined.
  • a voice prompt similar to "The recognition object cannot be determined, please watch the user back" can be issued.
  • S308 may be to calculate the first product of the maximum face size value and the first preset ratio threshold, and use the first product as the first size threshold.
  • S309 may be to determine whether the face size value of each face except the face with the largest face size value is less than or equal to the first size threshold. If yes, execute S310, otherwise, execute S313.
  • S313 may be to calculate the second product of the maximum face size value and the second preset ratio threshold, use the second product as the second size threshold, and determine the face size of each face except the face with the largest face size value. Whether the size values are all less than or equal to the second size threshold, if yes, execute S314, otherwise, execute S317.
  • an embodiment of the disclosure further provides an apparatus for determining an identification object.
  • FIG. 6 is a structure of the apparatus for determining an identification object provided by an embodiment of the apparatus for determining an identification object in the second aspect of the disclosure. Schematic.
  • the apparatus 600 for identifying an object may include: an acquisition module 610 , an extraction module 620 , and a determination module 630 .
  • the obtaining module 610 is configured to obtain a shooting preview image, wherein the shooting preview image includes at least two human faces.
  • the extraction module 620 is configured to extract feature information of each of the at least two human faces in the shot preview image, where the feature information includes a face size value.
  • the determining module 630 is configured to determine, according to the face size value of each face, a face whose face size value satisfies a preset recognition condition as a recognition object.
  • the determining module includes: a first determining unit, configured to determine the face with the largest face size value as the recognition object.
  • the determining module includes: a calculating unit, configured to separately calculate the difference between the face size value and the largest face size value of each face except the face with the largest face size value among the at least two faces. ratio.
  • the second determining unit is configured to determine the face corresponding to the maximum face size value as the recognition object when at least one ratio is less than or equal to the first preset ratio threshold.
  • the second determining unit includes: an obtaining subunit, configured to obtain the face corresponding to the maximum face size value and display it on the display screen when at least one ratio is less than or equal to the first preset ratio threshold length of time.
  • the determining subunit is configured to determine the face corresponding to the maximum face size value as the recognition object when the display duration of the face corresponding to the maximum face size value is greater than or equal to the preset duration threshold.
  • the obtaining module is further configured to obtain the maximum face size when any one of the at least one ratio is greater than the first preset ratio threshold, and the at least one ratio is less than or equal to the second preset ratio threshold The first position of the face corresponding to the value, wherein the second preset ratio threshold is greater than the first preset ratio threshold.
  • the determining module is further configured to determine the face corresponding to the maximum face size value as the recognition object when the first position satisfies the preset position condition.
  • the determining module includes: an obtaining unit, configured to obtain the first position of the face corresponding to the maximum face size value.
  • the third determining unit is configured to determine the face corresponding to the maximum face size value as the recognition object when the first position satisfies the preset position condition.
  • the preset position condition includes: the first position matches the preset position, or the first position is located in a preset area.
  • the feature information further includes face angle, face occlusion information, and face eye information.
  • the determining module includes: a screening unit for screening at least two faces according to face angle, face occlusion information, and face eye information.
  • the fourth determining unit is configured to determine, according to the face size value of the screened face, a face whose face size value satisfies a preset recognition condition as a recognition object.
  • the face size value includes a face-eye distance value or the number of face pixels.
  • each module/unit in the apparatus 600 for determining an object for identification shown in FIG. 6 has the function of implementing each step in the method for determining an object for identification provided by the embodiment of the present disclosure, and can achieve its corresponding technical effect, for the sake of brevity. , and will not be repeated here.
  • FIG. 7 is a schematic structural diagram of a recognition object determination device provided by an embodiment of the recognition object determination device of the third aspect of the present disclosure.
  • the identification object determination device 700 in this embodiment includes an input device 701 , an input interface 702 , a central processing unit 703 , a memory 704 , an output interface 705 , and an output device 706 .
  • the input interface 702, the central processing unit 703, the memory 704, and the output interface 705 are connected to each other through the bus 710, and the input device 701 and the output device 706 are respectively connected to the bus 710 through the input interface 702 and the output interface 705, and then to the identification object determination device 700. connections to other components.
  • the input device 701 receives input information from the outside, and transmits the input information to the central processing unit 703 through the input interface 702; the central processing unit 703 processes the input information based on the computer-executable instructions stored in the memory 704 to generate output information, temporarily or permanently store the output information in the memory 704, and then transmit the output information to the output device 706 through the output interface 705; the output device 706 outputs the output information to the outside of the identification object determination device 700 for the user to use.
  • the identification object determination device 700 shown in FIG. 7 includes: a memory 704 for storing a program; and a processor 703 for running the program stored in the memory, so as to realize the identification object determination provided by the embodiments of the present disclosure method.
  • Embodiments of the present disclosure further provide a computer-readable storage medium, where computer program instructions are stored thereon; when the computer program instructions are executed by a processor, the method for determining an identification object provided by the embodiments of the present disclosure is implemented.
  • Examples of computer-readable storage media shown include non-transitory computer-readable storage media, such as read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), magnetic disks or CD etc.
  • the functional blocks shown in the above-described structural block diagrams may be implemented as hardware, software, firmware, or a combination thereof.
  • it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), suitable firmware, a plug-in, a function card, and the like.
  • ASIC application specific integrated circuit
  • elements of the present disclosure are programs or code segments used to perform the required tasks.
  • the program or code segments may be stored in a machine-readable medium or transmitted over a transmission medium or communication link by a data signal carried in a carrier wave.
  • a "machine-readable medium” may include any medium that can store or transmit information.
  • machine-readable media examples include electronic circuits, semiconductor memory devices, Read-Only Memory (ROM), flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (Radio Frequency, RF) link, etc.
  • the code segments may be downloaded via a computer network such as the Internet, an intranet, or the like.
  • the exemplary embodiments mentioned in the present disclosure describe some methods or systems based on a series of steps or devices.
  • the present disclosure is not limited to the order of the above steps, that is, the steps may be performed in the order mentioned in the embodiments, or may be different from the order in the embodiments, or several steps may be performed simultaneously.
  • processors may be, but are not limited to, general purpose processors, special purpose processors, application specific processors, or field programmable logic circuits. It will also be understood that each block of the block diagrams and/or flowchart illustrations, and combinations of blocks in the block diagrams and/or flowchart illustrations, can also be implemented by special purpose hardware for performing the specified functions or actions, or by special purpose hardware and/or A combination of computer instructions is implemented.

Landscapes

  • Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • General Health & Medical Sciences (AREA)
  • Human Computer Interaction (AREA)
  • Health & Medical Sciences (AREA)
  • Image Analysis (AREA)
  • Geophysics And Detection Of Objects (AREA)
  • Measuring Or Testing Involving Enzymes Or Micro-Organisms (AREA)
  • Automatic Analysis And Handling Materials Therefor (AREA)

Abstract

一种识别对象确定方法、装置、设备及存储介质。该方法包括:获取拍摄预览图像(S210),其中,拍摄预览图像包括至少两个人脸;提取拍摄预览图像中至少两个人脸中每个人脸的特征信息(S220),其中,特征信息包括人脸大小值;根据每个人脸的人脸大小值,确定人脸大小值满足预设识别条件的人脸为识别对象(S230)。能够在人脸采集过程中,从至少两个人脸中选择合适的目标人脸作为识别对象,提高识别对象确定的准确率。

Description

识别对象确定方法、装置、设备及存储介质
相关申请的交叉引用
本申请要求享有于2020年09月17日提交的名称为“识别对象确定方法、装置、设备及存储介质”的中国专利申请202010983486.0的优先权,该申请的全部内容通过引用并入本文中。
技术领域
本公开涉及计算机视觉技术领域,尤其涉及一种识别对象确定方法、装置、设备及存储介质。
背景技术
目前,人脸识别以其准确、安全、方便等诸多特点,而被广泛应用于生产、金融、安全、交通等领域。例如考勤机、无人零售机、门禁系统等等。
但是,由于人脸识别的使用环境开放,在进行人脸采集的过程中,往往会出现多个人脸,尤其是在排队场景中更为明显,例如刷脸支付、闸机过站、人员签到等场景。因此,容易导致识别对象确定错误,识别对象确定的准确率较差。
发明内容
本公开实施例提供了一种识别对象确定方法、装置、设备及存储介质,能够提高识别对象确定的准确率。
第一方面,本公开实施例提供一种识别对象确定方法,该方法包括:
获取拍摄预览图像,其中,拍摄预览图像包括至少两个人脸;
提取拍摄预览图像中至少两个人脸中每个人脸的特征信息,其中,特征信息包括人脸大小值;
根据每个人脸的人脸大小值,确定人脸大小值满足预设识别条件的人脸为识别对象。
第二方面,本公开实施例提供一种识别对象确定装置,该装置包括:
获取模块,用于获取拍摄预览图像,其中,拍摄预览图像包括至少两个人脸;
提取模块,用于提取拍摄预览图像中至少两个人脸中每个人脸的特征信息,其中,特征信息包括人脸大小值;
确定模块,用于根据每个人脸的人脸大小值,确定人脸大小值满足预设识别条件的人脸为识别对象。
第三方面,本公开实施例提供一种识别对象确定设备,该设备包括:处理器以及存储有计算机程序指令的存储器;处理器执行计算机程序指令时实现第一方面或者第一方面任一些可实现方式中所述的识别对象确定方法。
第四方面,本公开实施例提供一种计算机可读存储介质,计算机可读存储介质上存储有计算机程序指令,计算机程序指令被处理器执行时实现第一方面或者第一方面任一些可实现方式中所述的识别对象确定方法。
本公开实施例提供的一种识别对象确定方法、装置、设备及存储介质,通过提取拍摄预览图像中每个人脸的人脸大小值,根据每个人脸的人脸大小值,确定人脸大小值满足预设识别条件的人脸为识别对象。能够在人脸采集过程中选择合适的目标人脸作为识别对象,实现识别对象的精确确定,提高识别对象确定的准确率。
附图说明
为了更清楚地说明本公开实施例的技术方案,下面将对本公开实施例中所需要使用的附图作简单地介绍,对于本领域普通技术人员来讲,在不付出创造性劳动的前提下,还可以根据这些附图获得其他的附图。
图1是本公开第一方面的识别对象确定系统的实施例的一种识别对象确定系统的架构示意图;
图2是本公开第一方面的识别对象确定方法的实施例的一种识别对象 确定方法的流程示意图;
图3是本公开第一方面的识别对象确定方法的实施例的另一种识别对象确定方法的流程示意图;
图4是本公开第一方面的识别对象确定方法的实施例的一种区域划分示意图;
图5是本公开第一方面的识别对象确定方法的实施例的一种相对位置示意图;
图6是本公开第二方面的识别对象确定装置的实施例提供的识别对象确定装置的结构示意图;
图7是本公开第三方面的识别对象确定设备的实施例提供的识别对象确定设备的结构示意图。
具体实施方式
下面将详细描述本公开的各个方面的特征和示例性实施例,为了使本公开的目的、技术方案及优点更加清楚明白,以下结合附图及实施例,对本公开进行进一步详细描述。应理解,此处所描述的具体实施例仅解释本公开,而不是限定本公开。对于本领域技术人员来说,本公开可以在不需要这些具体细节中的一些细节的情况下实施。下面对实施例的描述仅仅是为了通过示出本公开的示例来提供对本公开更好的理解。
需要说明的是,在本文中,诸如第一和第二等之类的关系术语仅仅用来将一个实体或者操作与另一个实体或操作区分开来,而不一定要求或者暗示这些实体或操作之间存在任何这种实际的关系或者顺序。而且,术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含,从而使得包括一系列要素的过程、方法、物品或者设备不仅包括那些要素,而且还包括没有明确列出的其他要素,或者是还包括为这种过程、方法、物品或者设备所固有的要素。在没有更多限制的情况下,由语句“包括……”限定的要素,并不排除在包括所述要素的过程、方法、物品或者设备中还存在另外的相同要素。
目前,传统识别对象确定方案主要通过优化业务流程、调整拍摄设备 布放角度等策略,减少显示屏幕中出现多个人脸的情况,一般获取第一个出现在屏幕中的人脸,作为后续业务执行过程中的识别对象。但是在人流量较大的人脸识别场景,比如刷脸支付、刷脸进站、刷脸过闸等场景,显示屏幕往往会出现多个人脸。因此,可能会出现误识别现象,即将“旁观”或“身后”的人脸作为识别对象,影响用户使用体验。
因此,为了解决现有技术问题,本公开实施例提供了一种识别对象确定方法、装置、设备及存储介质。通过提取拍摄预览图像中每个人脸的人脸大小值,根据每个人脸的人脸大小值,确定人脸大小值满足预设识别条件的人脸为识别对象。能够在人脸采集过程中选择合适的目标人脸作为识别对象,实现识别对象的精确确定,提高识别对象确定的准确率。
下面结合附图,通过具体的实施例及其应用场景对本公开实施例提供的识别对象确定方法、装置、设备和存储介质进行详细地说明。
图1是本公开第一方面的识别对象确定系统的实施例的一种识别对象确定系统的架构示意图。如图1所示,该识别对象确定系统可以包括拍摄设备110、电子设备120,其中,拍摄设备110的示例可以为摄像头、安装有摄像头模组的设备等等。电子设备120的示例可以为移动电子设备,也可以为非移动电子设备。例如,移动电子设备可以为手机、平板电脑、笔记本电脑、掌上电脑、超级移动个人计算机(Ultra-Mobile Personal Computer,UMPC)、上网本或者个人数字助理(personal digital assistant,PDA)等等,非移动电子设备可以为服务器、网络附属存储器(Network Attached Storage,NAS)、个人计算机(Personal Computer,PC)、柜员机或者自助机等等。拍摄设备110和电子设备120之间存在通信连接。例如通过网络进行通信,其中,网络可以是有线通信网络或无线通信网络。示例性地,拍摄设备110可以作为一个模块集成于电子设备120。
作为一个示例,该识别对象确定系统可以应用于刷脸支付、刷脸进站、刷脸签到等人脸识别场景。在这些场景下,由于人脸的快速采集以及较大的人流量,显示屏幕上容易出现多个人脸聚集的现象。
参见图1,电子设备120可以获取拍摄设备110针对用户的拍摄预览 图像,其中,拍摄预览图像包括至少两个用户,即至少两个人脸。接着提取拍摄预览图像中每个人脸的特征信息,其中,特征信息包括人脸大小值。然后根据每个人脸的人脸大小值,确定人脸大小值满足预设识别条件的人脸为识别对象,以执行后续的人脸识别业务。
下面将介绍本公开实施例提供的识别对象确定方法。在一些实施例中,该识别对象确定方法的执行主体可以是图1所示的识别对象确定系统中的电子设备120,或者电子设备120中的模块。
图2是本公开第一方面的识别对象确定方法的实施例的一种识别对象确定方法的流程示意图,如图2所示,该识别对象确定方法可以包括以下步骤:
S210,获取拍摄预览图像。
拍摄预览图像的示例可以包括至少两个人脸,可以由拍摄设备现场采集。例如,拍摄预览图像可以是预览的画面输入帧,即拍摄设备拍摄时在显示屏幕上显示的图像。
S220,提取拍摄预览图像中至少两个人脸中每个人脸的特征信息。
特征信息的示例可以包括人脸大小值。示例性地,人脸大小值可以包括人脸眼间距值或者人脸像素数量等等,其中,以人脸眼间距值表征人脸大小值,可以降低人脸大小差异的噪声。
在一些实施例中,可以利用人脸识别算法,对拍摄预览图像进行初步的人脸识别,识别出其中的人脸。然后对识别出的人脸进行特征提取,得到每个人脸的特征信息。
S230,根据每个人脸的人脸大小值,确定人脸大小值满足预设识别条件的人脸为识别对象。
识别对象作为后续人脸识别业务执行的主体,将用于人脸识别。
作为识别对象的人脸相比其他被拍摄的人脸,通常是最接近拍摄设备的。于是在一些实施例中,可以确定人脸大小值最大的人脸为识别对象,实现识别对象的精确确定。
此外,作为识别对象的人脸相比其他人脸,要占有大小的优势。在另一些实施例中,可以分别计算至少两个人脸中除人脸大小值最大的人脸之 外,每个人脸的人脸大小值与最大人脸大小值的比值。当至少一个比值均小于或等于第一预设比值阈值时,确定最大人脸大小值对应的人脸为识别对象,提高识别对象确定的准确性。可以理解,第一预设比值阈值可以根据实际情况灵活设置,例如可以是60%。
可知,通常用户在识别使用过程中不会离开当前位置,人脸始终出现在屏幕中。因此在一个示例中,当至少一个比值均小于或等于第一预设比值阈值时,可以获取最大人脸大小值对应的人脸在显示屏幕上显示的时长。当最大人脸大小值对应的人脸的显示时长大于或等于预设时长阈值时,确定最大人脸大小值对应的人脸为识别对象。通过在人脸大小值的基础上引入显示时长作为判断因素,可以进一步提高对识别对象的精确确定。
作为一个具体的示例,可以对最大人脸大小值对应的人脸进行连续帧的人脸跟踪。示例性地,可以向最大人脸大小值对应的人脸分配唯一的人脸标识,在连续帧的人脸跟踪过程中,同一人脸在未离开的情况下,人脸标识保持不变,因此可以统计最大人脸大小值对应的人脸的持续显示时长。进而在持续显示时长大于或等于T时,将最大人脸大小值对应的人脸作为识别对象,其中,T表示预设时长阈值,可以是绝对时长,例如800ms。也可以是相对时长,例如完成一段业务处理的时间。比如刷脸进站场景中,从指定拍摄区域通道通过的时长,即用户从进入到离开特定拍摄区域(或者未离开拍摄设备覆盖范围)的时长;又比如采集连续帧图像到一定数量(如5张)的时间;再比如交易应答时间(或者获取一笔识别交易结果的时间),即在开始跟踪最大人脸大小值对应的人脸时,发起识别,直到接收到应答结果的持续时长,在此期间,最大人脸大小值对应的人脸始终显示在显示屏幕中。
在另一个实施例中,可以获取最大人脸大小值对应的人脸的第一位置。当第一位置满足预设位置条件时,确定最大人脸大小值对应的人脸为识别对象,即在人脸大小值的基础上引入位置作为判断因素,提高对识别对象的精确确定。在一些实施例中,预设位置条件可以包括:第一位置与预设位置匹配,或者第一位置位于预设区域。可以理解,预设位置和预设 区域可以根据实际情况灵活设置,例如可以在场景布置实施时,根据调试经验预先选择。
在本公开实施例中,通过提取拍摄预览图像中每个人脸的人脸大小值,根据每个人脸的人脸大小值,确定人脸大小值满足预设识别条件的人脸为识别对象。能够在人脸采集过程中选择合适的目标人脸作为识别对象,提高识别对象确定的准确率。
在一个实施例中,当至少一个比值中任意一个比值大于第一预设比值阈值时,该方法还可以包括:
首先,当至少一个比值均小于或等于第二预设比值阈值时,获取最大人脸大小值对应的人脸的第一位置,其中,第二预设比值阈值大于第一预设比值阈值,比如第一预设比值阈值为60%,第二预设比值阈值为80%。而且第一预设比值阈值和第二预设比值阈值可以在一段时间后,根据识别对象确定的准确率进行调整。示例性地,可以选择完成业务流程即识别对象确定正确的人脸图片,作为正向样本。选择用户手动取消业务流程即识别对象错误的人脸图片,作为负样本。通过正负样本占总样本的比例,统计识别对象确定的准确率,根据准确率调整第一预设比值阈值和第二预设比值阈值,实现阈值的动态调整。
然后,当第一位置满足预设位置条件时,确定最大人脸大小值对应的人脸为识别对象。在该实施例中,引入第二预设比值阈值与位置作进一步的判断,细化确定识别对象的条件,可以进一步提高识别对象确定的准确性。
可以理解,在人脸识别的场景下,作为识别对象的用户通常会主动靠近拍摄设备,其人脸在显示屏幕上是正对着的,一般具有清晰可见、无遮挡、未闭眼等特征。在一些实施例中,特征信息还可以包括人脸角度、人脸遮挡信息、人脸眼部信息等等。进而可以根据人脸角度、人脸遮挡信息、人脸眼部信息筛选至少两个人脸,即以人脸角度、人脸遮挡信息、人脸眼部信息作为判断因素,对拍摄预览图像中的人脸进行筛选,保留满足相应条件的人脸。
进一步地,可以根据人脸大小值、人脸角度、人脸遮挡信息、人脸眼 部信息筛选至少两个人脸。并根据筛选后的人脸的人脸大小值,确定人脸大小值满足预设识别条件的人脸为识别对象。在该示例中,通过在多个维度上筛选人脸,从筛选后的人脸中选择合适的人脸作为识别对象,能够提高识别对象确定的准确率。
下面以识别对象确定方法应用于刷脸支付场景为例,对本公开实施例提供的识别对象确定方法进行详细说明,如图3所示,该方法可以包括
S301、获取拍摄设备现场拍摄用户的拍摄预览图像。
S302、根据人脸识别算法识别拍摄预览图像中的人脸。
S303、对拍摄预览图像中的人脸进行筛选。
具体地,首先可以提取每个人脸的特征信息,特征信息包括人脸大小值、人脸角度、人脸遮挡信息、人脸眼部信息。
其次,可以对每个人脸进行人脸大小值判断,保留人脸大小值大于或等于预设大小阈值的人脸。作为一个示例,以人脸像素数量表征人脸大小值,预设大小阈值可以设置为100*100,100*100表示人脸像素的宽度和高度。
接着可以进行人脸角度判断,保留人脸角度小于或等于预设角度阈值的人脸。作为一个示例,人脸角度包括翻滚角(roll)、俯仰角(pitch)、偏航角(yaw),三类角度对应的预设角度阈值可以均为±20°,保留三类角度均小于或等于±20°的人脸。
然后可以进行人脸遮挡信息判断,保留人脸遮挡信息满足人脸遮挡条件的人脸。作为一个示例,人脸遮挡条件可以根据业务要求灵活设置,比如完全无遮挡、关键点无遮挡、遮挡面积不超过一定比例等等。
再者可以进行人脸眼部信息判断,判断人脸是否闭眼,进而保留未闭眼的人脸。值得注意的是,上述判断步骤可以根据业务需要灵活调整,例如可以适当增减、变动顺序等等,在此不做限制。
如图3所示,该识别对象确定方法还包括S304、判断筛选后的人脸是否唯一。
若是,则执行S305,否则,则执行S308。
如图3所示,该识别对象确定方法还包括S305、获取唯一人脸在显示 屏幕上显示的时长。
如图3所示,该识别对象确定方法还包括S306、判断唯一人脸的显示时长是否大于或等于预设时长阈值。
若是,则执行S307,否则,则执行S317。
如图3所示,该识别对象确定方法还包括S307、确定唯一人脸为识别对象。
如图3所示,该识别对象确定方法还包括S308、分别计算除人脸大小值最大的人脸之外,每个人脸的人脸大小值与最大人脸大小值的比值。
如图3所示,该识别对象确定方法还包括S309、判断比值是否均小于或等于第一预设比值阈值。
若是,则执行S310,否则,则执行S313。
如图3所示,该识别对象确定方法还包括S310、获取最大人脸大小值对应的人脸在显示屏幕上显示的时长。
如图3所示,该识别对象确定方法还包括S311、判断显示时长是否大于或等于预设时长阈值。
若是,则执行S312,否则,则执行S317。
如图3所示,该识别对象确定方法还包括S312、确定最大人脸大小值对应的人脸为识别对象。
如图3所示,该识别对象确定方法还包括S313、判断比值是否均小于或等于第二预设比值阈值。
若是,则执行S314,否则,则执行S317。
如图3所示,该识别对象确定方法还包括S314、获取最大人脸大小值对应的人脸的第一位置。
第一位置可以包括最大人脸大小值对应的人脸的绝对位置,以及以第二大人脸大小值对应的人脸作为参考点的相对位置。
具体地,可以根据预设的区域划分规则划分拍摄预览图像,根据划分的区域确定人脸的位置标记,以该位置标记表征人脸的绝对位置。参见图4,区域划分规则可以是对拍摄预览图像按照中心点进行分割,并按照数学平面坐标系分为第一象限、第二象限、第三象限、第四象限4个象限。 同时以中心点为原点,拍摄预览图像宽度的一半为宽,拍摄预览图像高度的一半为高,作为中心区域C,其中,中心区域C与象限区域有重合。
作为一个示例,可以以中心点为原点(0,0),构建平面坐标系。对于拍摄预览图像中的人脸,计算人脸中心点坐标(x,y),根据人脸中心点坐标和划分的区域确定人脸的位置标记。位置标记可以为LG,其中,L表示人脸中心点坐标位于的象限,取值1、2、3或4,L为1时表示位于第一象限。G表示人脸中心点坐标是否位于中心区域C,取值0或1,其中,G为0时表示未位于中心区域C,G为1时表示位于中心区域C。这里的位于包括中心点坐标在边界上。具体地,位置标记可以如下:未位于中心区域C:第一象限:10,第二象限:20,第三象限:30,第四象限:40;位于中心区域C:第一象限:11,第二象限:21,第三象限:31,第四象限:41。具体地,以位置标记10为例,表示人脸中心点坐标位于第一象限,未位于中心区域。以位置标记11为例,表示人脸中心点坐标位于中心区域与第一象限。如图4所示,人脸1的中心坐标位于第三象限,且未位于中心区域C,则位置标记为30。人脸2的中心坐标位于第一、二象限,且位于中心区域C,则位置标记为11和21。人脸3的中心坐标位于第一象限,且位于中心区域C,则位置标记为11。可以理解,区域划分规则可以根据实际需要灵活调整,在此不做限制。
在一些实施例中,相对位置可以使用相对位置计算公式计算,结合图5,相对位置计算公式可以如下所示:
A(X,Y)=O 1(x 1,y 1)-O 2(x 2,y 2)     (1)
A(X,Y)表示向量,O 1(x 1,y 1)表示第二大人脸的人脸中心点坐标,O 2(x 2,y 2)表示最大人脸的人脸中心点坐标,其中,X>0则表示最大人脸在横轴方向上,在第二大人脸的左侧,反之则在右侧;Y>0则表示最大人脸在纵轴方向上,在第二大人脸的下方,反之则在上方。
如图3所示,该识别对象确定方法还包括S315、判断第一位置是否满足预设位置条件。
在一些实施例中,预设位置条件包括:第一位置与预设位置匹配,或者第一位置位于预设区域。在某一固定场景中,识别对象在显示屏幕中出 现的绝对位置和相对位置往往出现一定的聚集性。可以理解,聚集的位置因场景不同或拍摄设备角度差异而不同。可以在场景应用中,选取一定数量(例如10万张)的存在多个人脸的人脸识别场景图像,对其进行数据处理,标记识别对象和第二大人脸,统计识别对象的相对位置和绝对位置,即对每一张图像进行分类,统计绝对位置LC值,以及相对位置A(X,Y),根据统计结果,选择该场景下,数量最多的LC值或者LC值组合,以及相对位置,确认该场景下识别对象经常出现的绝对位置和相对位置,以此作为预设位置,或者根据多个经常出现的绝对位置和相对位置确定预设区域。通过实际应用的场景设置预设位置或者预设区域,可以提高识别对象确定的准确率。在此基础判断判断第一位置是否满足预设位置条件,若是,则执行S316,否则,则执行S317。
如图3所示,该识别对象确定方法还包括S316、确定最大人脸大小值对应的人脸为识别对象。
如图3所示,该识别对象确定方法还包括S317、提示无法确定识别对象。
具体地,可以发出类似“识别对象无法确定,请旁观用户后退”的语音提示。
在另一些示例中,S308可以是计算最大人脸大小值与第一预设比值阈值的第一乘积,将第一乘积作为第一大小阈值。
S309可以是判断除人脸大小值最大的人脸之外,每个人脸的人脸大小值是否均小于或等于第一大小阈值。若是,则执行S310,否则,则执行S313。
S313可以是计算最大人脸大小值与第二预设比值阈值的第二乘积,将第二乘积作为第二大小阈值,判断除人脸大小值最大的人脸之外,每个人脸的人脸大小值是否均小于或等于第二大小阈值,若是,则执行S314,否则,则执行S317。
基于本公开实施例提供的识别对象确定方法,本公开实施例还提供了一种识别对象确定装置,图6是本公开第二方面的识别对象确定装置的实施例提供的识别对象确定装置的结构示意图。
如图6所示,识别对象确定装置600可以包括:获取模块610、提取模块620、确定模块630。
获取模块610,用于获取拍摄预览图像,其中,拍摄预览图像包括至少两个人脸。
提取模块620,用于提取拍摄预览图像中至少两个人脸中每个人脸的特征信息,其中,特征信息包括人脸大小值。
确定模块630,用于根据每个人脸的人脸大小值,确定人脸大小值满足预设识别条件的人脸为识别对象。
在一些实施例中,,确定模块包括:第一确定单元,用于确定人脸大小值最大的人脸为识别对象。
在一些实施例中,,确定模块包括:计算单元,用于分别计算至少两个人脸中除人脸大小值最大的人脸之外,每个人脸的人脸大小值与最大人脸大小值的比值。
第二确定单元,用于当至少一个比值均小于或等于第一预设比值阈值时,确定最大人脸大小值对应的人脸为识别对象。
在一些实施例中,,第二确定单元包括:获取子单元,用于当至少一个比值均小于或等于第一预设比值阈值时,获取最大人脸大小值对应的人脸在显示屏幕上显示的时长。
确定子单元,用于当最大人脸大小值对应的人脸的显示时长大于或等于预设时长阈值时,确定最大人脸大小值对应的人脸为识别对象。
在一些实施例中,,获取模块,还用于当至少一个比值中任意一个比值大于第一预设比值阈值,且至少一个比值均小于或等于第二预设比值阈值时,获取最大人脸大小值对应的人脸的第一位置,其中,第二预设比值阈值大于第一预设比值阈值。
确定模块,还用于当第一位置满足预设位置条件时,确定最大人脸大小值对应的人脸为识别对象。
在一些实施例中,,确定模块包括:获取单元,用于获取最大人脸大小值对应的人脸的第一位置。
第三确定单元,用于当第一位置满足预设位置条件时,确定最大人脸 大小值对应的人脸为识别对象。
在一些实施例中,,预设位置条件包括:第一位置与预设位置匹配,或者第一位置位于预设区域。
在一些实施例中,,特征信息还包括人脸角度、人脸遮挡信息、人脸眼部信息。
确定模块包括:筛选单元,用于根据人脸角度、人脸遮挡信息、人脸眼部信息筛选至少两个人脸。
第四确定单元,用于根据筛选后的人脸的人脸大小值,确定人脸大小值满足预设识别条件的人脸为识别对象。
在一些实施例中,,人脸大小值包括人脸眼间距值或者人脸像素数量。
可以理解的是,图6所示识别对象确定装置600中的各个模块/单元具有实现本公开实施例提供的识别对象确定方法中的各个步骤的功能,并能达到其相应的技术效果,为了简洁,在此不再赘述。
图7是本公开第三方面的识别对象确定设备的实施例提供的识别对象确定设备的结构示意图。
如图7所示,本实施例中的识别对象确定设备700包括输入设备701、输入接口702、中央处理器703、存储器704、输出接口705、以及输出设备706。输入接口702、中央处理器703、存储器704、以及输出接口705通过总线710相互连接,输入设备701和输出设备706分别通过输入接口702和输出接口705与总线710连接,进而与识别对象确定设备700的其他组件连接。
具体地,输入设备701接收来自外部的输入信息,并通过输入接口702将输入信息传送到中央处理器703;中央处理器703基于存储器704中存储的计算机可执行指令对输入信息进行处理以生成输出信息,将输出信息临时或者永久地存储在存储器704中,然后通过输出接口705将输出信息传送到输出设备706;输出设备706将输出信息输出到识别对象确定设备700的外部供用户使用。
在一些实施例中,图7所示的识别对象确定设备700包括:存储器 704,用于存储程序;处理器703,用于运行存储器中存储的程序,以实现本公开实施例提供的识别对象确定方法。
本公开实施例还提供一种计算机可读存储介质,该计算机可读存储介质上存储有计算机程序指令;该计算机程序指令被处理器执行时实现本公开实施例提供的识别对象确定方法。所示的计算机可读存储介质的示例包括非暂态计算机可读存储介质,如只读存储器(Read-Only Memory,简称ROM)、随机存取存储器(Random Access Memory,简称RAM)、磁碟或者光盘等。
需要明确的是,本说明书中的各个实施例均采用递进的方式描述,各个实施例之间相同或相似的部分互相参见即可,为了简洁,不再赘述。本公开并不局限于上文所描述并在图中示出的特定配置和处理。为了简明起见,这里省略了对已知方法的详细描述。在上述实施例中,描述和示出了若干具体的步骤作为示例。但是,本公开的方法过程并不限于所描述和示出的具体步骤,本领域的技术人员可以在领会本公开的精神后,做出各种改变、修改和添加,或者改变步骤之间的顺序。
以上所述的结构框图中所示的功能块可以实现为硬件、软件、固件或者它们的组合。当以硬件方式实现时,其可以例如是电子电路、专用集成电路(Application Specific Integrated Circuit,ASIC)、适当的固件、插件、功能卡等等。当以软件方式实现时,本公开的元素是被用于执行所需任务的程序或者代码段。程序或者代码段可以存储在机器可读介质中,或者通过载波中携带的数据信号在传输介质或者通信链路上传送。“机器可读介质”可以包括能够存储或传输信息的任何介质。机器可读介质的例子包括电子电路、半导体存储器设备、只读存储器(Read-Only Memory,ROM)、闪存、可擦除ROM(EROM)、软盘、CD-ROM、光盘、硬盘、光纤介质、射频(Radio Frequency,RF)链路,等等。代码段可以经由诸如因特网、内联网等的计算机网络被下载。
还需要说明的是,本公开中提及的示例性实施例,基于一系列的步骤或者装置描述一些方法或系统。但是,本公开不局限于上述步骤的顺序,也就是说,可以按照实施例中提及的顺序执行步骤,也可以不同于实施例 中的顺序,或者若干步骤同时执行。
上面参考根据本公开的实施例的方法、装置(系统)和计算机程序产品的流程图和/或框图描述了本公开的各方面。应当理解,流程图和/或框图中的每个方框以及流程图和/或框图中各方框的组合可以由计算机程序指令实现。这些计算机程序指令可被提供给通用计算机、专用计算机、或其它可编程数据处理装置的处理器,以产生一种机器,使得经由计算机或其它可编程数据处理装置的处理器执行的这些指令使能对流程图和/或框图的一个或多个方框中指定的功能/动作的实现。这种处理器可以是但不限于是通用处理器、专用处理器、特殊应用处理器或者现场可编程逻辑电路。还可理解,框图和/或流程图中的每个方框以及框图和/或流程图中的方框的组合,也可以由执行指定的功能或动作的专用硬件来实现,或可由专用硬件和计算机指令的组合来实现。
以上所述,仅为本公开的具体实施方式,所属领域的技术人员可以清楚地了解到,为了描述的方便和简洁,上述描述的系统、模块和单元的具体工作过程,可以参考前述方法实施例中的对应过程,在此不再赘述。应理解,本公开的保护范围并不局限于此,任何熟悉本技术领域的技术人员在本公开揭露的技术范围内,可轻易想到各种等效的修改或替换,这些修改或替换都应涵盖在本公开的保护范围之内。

Claims (20)

  1. 一种识别对象确定方法,包括:
    获取拍摄预览图像,其中,所述拍摄预览图像包括至少两个人脸;
    提取所述拍摄预览图像中所述至少两个人脸中每个人脸的特征信息,其中,所述特征信息包括人脸大小值;
    根据所述每个人脸的人脸大小值,确定人脸大小值满足预设识别条件的人脸为识别对象。
  2. 根据权利要求1所述的方法,其中,所述根据所述每个人脸的人脸大小值,确定人脸大小值满足预设识别条件的人脸为识别对象,包括:
    确定人脸大小值最大的人脸为识别对象。
  3. 根据权利要求1所述的方法,其中,所述根据所述每个人脸的人脸大小值,确定人脸大小值满足预设识别条件的人脸为识别对象,包括:
    分别计算所述至少两个人脸中除人脸大小值最大的人脸之外,每个人脸的人脸大小值与最大人脸大小值的比值;
    当至少一个所述比值均小于或等于第一预设比值阈值时,确定所述最大人脸大小值对应的人脸为所述识别对象。
  4. 根据权利要求3所述的方法,其中,所述当至少一个所述比值均小于或等于第一预设比值阈值时,确定所述最大人脸大小值对应的人脸为所述识别对象,包括:
    当至少一个所述比值均小于或等于第一预设比值阈值时,获取所述最大人脸大小值对应的人脸在显示屏幕上显示的时长;
    当所述最大人脸大小值对应的人脸的显示时长大于或等于预设时长阈值时,确定所述最大人脸大小值对应的人脸为所述识别对象。
  5. 根据权利要求3所述的方法,其中,当至少一个所述比值中任意一个比值大于第一预设比值阈值时,所述方法还包括:
    当至少一个所述比值均小于或等于第二预设比值阈值时,获取所述最大人脸大小值对应的人脸的第一位置,其中,所述第二预设比值阈值大于所述第一预设比值阈值;
    当所述第一位置满足预设位置条件时,确定所述最大人脸大小值对应的人脸为所述识别对象。
  6. 根据权利要求1所述的方法,其中,所述根据所述每个人脸的人脸大小值,确定人脸大小值满足预设识别条件的人脸为识别对象,包括:
    获取最大人脸大小值对应的人脸的第一位置;
    当所述第一位置满足预设位置条件时,确定所述最大人脸大小值对应的人脸为所述识别对象。
  7. 根据权利要求5或6所述的方法,其中,所述预设位置条件包括:
    所述第一位置与预设位置匹配,或者所述第一位置位于预设区域。
  8. 根据权利要求1所述的方法,其中,所述特征信息还包括人脸角度、人脸遮挡信息、人脸眼部信息;
    所述根据所述每个人脸的人脸大小值,确定人脸大小值满足预设识别条件的人脸为识别对象,包括:
    根据所述人脸角度、所述人脸遮挡信息、所述人脸眼部信息筛选所述至少两个人脸;
    根据筛选后的人脸的人脸大小值,确定人脸大小值满足预设识别条件的人脸为识别对象。
  9. 根据权利要求1所述的方法,其中,所述人脸大小值包括人脸眼间距值或者人脸像素数量。
  10. 一种识别对象确定装置,包括:
    获取模块,用于获取拍摄预览图像,其中,所述拍摄预览图像包括至少两个人脸;
    提取模块,用于提取所述拍摄预览图像中所述至少两个人脸中每个人脸的特征信息,其中,所述特征信息包括人脸大小值;
    确定模块,用于根据所述每个人脸的人脸大小值,确定人脸大小值满足预设识别条件的人脸为识别对象。
  11. 根据权利要求10所述的装置,其中,所述确定模块包括:
    第一确定单元,用于确定人脸大小值最大的人脸为识别对象。
  12. 根据权利要求10所述的装置,其中,所述确定模块包括:
    计算单元,用于分别计算所述至少两个人脸中除人脸大小值最大的人脸之外,每个人脸的人脸大小值与最大人脸大小值的比值;
    第二确定单元,用于当至少一个所述比值均小于或等于第一预设比值阈值时,确定所述最大人脸大小值对应的人脸为所述识别对象。
  13. 根据权利要求12所述的装置,其中,所述第二确定单元包括:
    获取子单元,用于当至少一个所述比值均小于或等于第一预设比值阈值时,获取所述最大人脸大小值对应的人脸在显示屏幕上显示的时长;
    确定子单元,用于当所述最大人脸大小值对应的人脸的显示时长大于或等于预设时长阈值时,确定所述最大人脸大小值对应的人脸为所述识别对象。
  14. 根据权利要求12所述的装置,其中,
    所述获取模块,还用于当至少一个所述比值中任意一个比值大于第一预设比值阈值,且至少一个所述比值均小于或等于第二预设比值阈值时,获取所述最大人脸大小值对应的人脸的第一位置,其中,所述第二预设比值阈值大于所述第一预设比值阈值;
    所述确定模块,还用于当所述第一位置满足预设位置条件时,确定所述最大人脸大小值对应的人脸为所述识别对象。
  15. 根据权利要求10所述的装置,其中,所述确定模块包括:
    获取单元,用于获取最大人脸大小值对应的人脸的第一位置;
    第三确定单元,用于当所述第一位置满足预设位置条件时,确定所述最大人脸大小值对应的人脸为所述识别对象。
  16. 根据权利要求14或15所述的装置,其中,所述预设位置条件包括:
    所述第一位置与预设位置匹配,或者所述第一位置位于预设区域。
  17. 根据权利要求10所述的装置,其中,所述特征信息还包括人脸角度、人脸遮挡信息、人脸眼部信息;
    所述确定模块包括:筛选单元,用于根据所述人脸角度、所述人脸遮挡信息、所述人脸眼部信息筛选所述至少两个人脸;
    第四确定单元,用于根据筛选后的人脸的人脸大小值,确定人脸大小 值满足预设识别条件的人脸为识别对象。
  18. 根据权利要求10所述的装置,其中,所述人脸大小值包括人脸眼间距值或者人脸像素数量。
  19. 一种识别对象确定设备,包括:处理器以及存储有计算机程序指令的存储器,所述处理器执行所述计算机程序指令时实现如权利要求1-9任意一项所述的识别对象确定方法。
  20. 一种计算机可读存储介质,所述计算机可读存储介质上存储有计算机程序指令,所述计算机程序指令被处理器执行时实现如权利要求1-9任意一项所述的识别对象确定方法。
PCT/CN2021/117383 2020-09-17 2021-09-09 识别对象确定方法、装置、设备及存储介质 Ceased WO2022057719A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010983486.0 2020-09-17
CN202010983486.0A CN112215084B (zh) 2020-09-17 2020-09-17 识别对象确定方法、装置、设备及存储介质

Publications (1)

Publication Number Publication Date
WO2022057719A1 true WO2022057719A1 (zh) 2022-03-24

Family

ID=74050043

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/117383 Ceased WO2022057719A1 (zh) 2020-09-17 2021-09-09 识别对象确定方法、装置、设备及存储介质

Country Status (3)

Country Link
CN (1) CN112215084B (zh)
TW (1) TWI804988B (zh)
WO (1) WO2022057719A1 (zh)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115035578A (zh) * 2022-06-20 2022-09-09 支付宝(杭州)信息技术有限公司 一种支付方法、装置以及设备
CN115496473A (zh) * 2022-09-29 2022-12-20 杭州海康威视数字技术股份有限公司 一种考勤时间确定方法、装置及设备
CN117421729A (zh) * 2023-12-18 2024-01-19 湖南森鹰科技有限公司 一种自动化程序攻击检测方法、装置、系统及介质

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112215084B (zh) * 2020-09-17 2024-09-03 中国银联股份有限公司 识别对象确定方法、装置、设备及存储介质
CN115841503A (zh) * 2022-10-31 2023-03-24 广州市奥威亚电子科技有限公司 跟踪对象的确定方法及目标人物的跟踪方法
CN117079378B (zh) * 2023-10-16 2024-01-09 八维通科技有限公司 站点通行中多人脸过闸处理方法和系统、计算机程序介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109948586A (zh) * 2019-03-29 2019-06-28 北京三快在线科技有限公司 人脸验证的方法、装置、设备及存储介质
CN110166696A (zh) * 2019-06-28 2019-08-23 Oppo广东移动通信有限公司 拍摄方法、装置、终端设备以及计算机可读存储介质
CN110570200A (zh) * 2019-08-16 2019-12-13 阿里巴巴集团控股有限公司 一种支付方法及装置
CN111274965A (zh) * 2020-01-20 2020-06-12 上海眼控科技股份有限公司 人脸识别方法、装置、计算机设备和存储介质
CN112215084A (zh) * 2020-09-17 2021-01-12 中国银联股份有限公司 识别对象确定方法、装置、设备及存储介质

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103824068B (zh) * 2014-03-19 2018-06-01 上海看看智能科技有限公司 人脸支付认证系统及方法
US10532268B2 (en) * 2016-05-02 2020-01-14 Bao Tran Smart device
CN111726521B (zh) * 2017-01-26 2023-09-22 华为技术有限公司 终端的拍照方法、拍照装置和终端
CN108229120B (zh) * 2017-09-07 2020-07-24 北京市商汤科技开发有限公司 人脸解锁及其信息注册方法和装置、设备、程序、介质
TWI760525B (zh) * 2018-07-10 2022-04-11 鴻發國際科技股份有限公司 現金處理系統及現金交易方法
CN111435504A (zh) * 2019-01-13 2020-07-21 潘连香 基于面部识别的支付密码提取方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109948586A (zh) * 2019-03-29 2019-06-28 北京三快在线科技有限公司 人脸验证的方法、装置、设备及存储介质
CN110166696A (zh) * 2019-06-28 2019-08-23 Oppo广东移动通信有限公司 拍摄方法、装置、终端设备以及计算机可读存储介质
CN110570200A (zh) * 2019-08-16 2019-12-13 阿里巴巴集团控股有限公司 一种支付方法及装置
CN111274965A (zh) * 2020-01-20 2020-06-12 上海眼控科技股份有限公司 人脸识别方法、装置、计算机设备和存储介质
CN112215084A (zh) * 2020-09-17 2021-01-12 中国银联股份有限公司 识别对象确定方法、装置、设备及存储介质

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115035578A (zh) * 2022-06-20 2022-09-09 支付宝(杭州)信息技术有限公司 一种支付方法、装置以及设备
CN115496473A (zh) * 2022-09-29 2022-12-20 杭州海康威视数字技术股份有限公司 一种考勤时间确定方法、装置及设备
CN117421729A (zh) * 2023-12-18 2024-01-19 湖南森鹰科技有限公司 一种自动化程序攻击检测方法、装置、系统及介质
CN117421729B (zh) * 2023-12-18 2024-04-26 湖南森鹰科技有限公司 一种自动化程序攻击检测方法、装置、系统及介质

Also Published As

Publication number Publication date
CN112215084B (zh) 2024-09-03
CN112215084A (zh) 2021-01-12
TWI804988B (zh) 2023-06-11
TW202213169A (zh) 2022-04-01

Similar Documents

Publication Publication Date Title
TWI804988B (zh) 識別對象確定方法、裝置、設備及存儲介質
TWI789128B (zh) 人臉識別方法、裝置、設備及存儲介質
CN110826519B (zh) 人脸遮挡检测方法、装置、计算机设备及存储介质
US10922529B2 (en) Human face authentication method and apparatus, and storage medium
CN107292287B (zh) 人脸识别方法、装置、电子设备及存储介质
US20250342718A1 (en) Face authentication apparatus
CN108038176B (zh) 一种路人库的建立方法、装置、电子设备及介质
US11093737B2 (en) Gesture recognition method and apparatus, electronic device, and computer-readable storage medium
CN108038937B (zh) 一种迎宾信息的展示方法、装置、终端设备和存储介质
EP3163500A1 (en) Method and device for identifying region
CN108734067A (zh) 一种人证比对的认证方法、系统及相机
US9898648B2 (en) Face recognition method
EP3282387A1 (en) Fire detection method, fire detection apparatus and electronic equipment
WO2021022698A1 (zh) 尾随检测方法、装置、电子设备及存储介质
CN112149601B (zh) 兼容遮挡的面部属性识别方法、装置和电子设备
WO2021259033A1 (zh) 人脸识别方法、电子设备以及存储介质
US20200020110A1 (en) Image-based object tracking systems and methods
CN111259763A (zh) 目标检测方法、装置、电子设备及可读存储介质
CN108108711A (zh) 人脸布控方法、电子设备及存储介质
CN105868693A (zh) 身份认证方法及系统
CN114037934A (zh) 一种工服穿戴行为的识别方法、终端设备和存储介质
JP6347155B2 (ja) 画像処理装置、画像判断方法、及びプログラム
CN111476808A (zh) 虹膜图像的清晰度评价方法及装置
CN115082326A (zh) 视频去模糊的处理方法、边缘计算设备及中心处理器
CN110795998A (zh) 人流量的检测方法、装置、电子设备及可读存储介质

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: 21868544

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: 21868544

Country of ref document: EP

Kind code of ref document: A1