Data acquisition device of multispectral face living body recognition technology

By adopting lighting mechanisms and image acquisition devices in multispectral face live recognition technology, combining spectral cameras and color SLR cameras, real faces and high imitation two-dimensional fake face data are collected, and the problem of cubic error variables of spectral data is solved, and the problem of spectral data is improved Spectral accuracy and reliability of face recognition.

CN221977405UActive Publication Date: 2024-11-08SUZHOU PUXIANG TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202420220222.3
Authority / Receiving Office
CN · China
Patent Type
Utility models(China)
Current Assignee / Owner
Filing Date
2024-01-30
Publication Date
2024-11-08
Estimated Expiration
2034-01-30

AI Technical Summary

Technical Problem

In the existing multispectral face live body recognition technology, the error variable of the spectral data cube causes the spectral characteristics of the fake face and the real face to be masked, reducing the success rate of live body detection based on the spectral characteristics.

Method used

Using a data acquisition device including a lighting mechanism and an image acquisition mechanism, the same light source spectrum is provided through several groups of emission sources to cover the facial recognition feature band, and combining a spectral camera and a color SLR camera to collect data of real faces and high-imitation two-dimensional fake faces. , reduce interference from error variables and improve the accuracy of spectrum determination.

Benefits of technology

It effectively improves the accuracy of spectrum determination, enhances the reliability and robustness of face recognition, solves the problem of fake face attack, and makes the system more adaptable in complex environments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN221977405U_ABST
    Figure CN221977405U_ABST
Patent Text Reader

Abstract

The utility model discloses a data acquisition device of a multispectral face living body recognition technology. The data acquisition device comprises an illumination mechanism, an image acquisition mechanism arranged on one side of the illumination mechanism, and a to-be-detected target face module used for monitoring a target face, the illumination mechanism comprises a pair of identical emission sources which are the only light source of the image acquisition mechanism and can be used for providing the same light source spectrum to cover a face recognition characteristic wave band; the image acquisition mechanism comprises a spectrum camera and a color single lens reflex which is arranged symmetrically with the spectrum camera; the spectrum camera and the color single lens reflex camera are arranged right opposite to the target face module to be detected, and the focal lengths of the spectrum camera and the color single lens reflex camera are the same; therefore, the accuracy of spectrum activity judgment is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The utility model relates to the field of face recognition, in particular to a data acquisition device of multi-spectral face liveness recognition technology. Background Art

[0002] Multispectral face liveness recognition technology analyzes the spectral data cube of the face, trains the database and builds a deep neural network. Due to the introduction of additional information dimensions, this technology can complete liveness judgment while identifying the identity, effectively improving the anti-counterfeiting ability of face recognition technology.

[0003] In the existing device, 202123422324.9 is a multi-spectral face recognition integrated machine, including an upper light, a fixed plate fixedly connected in the middle of the upper light, two anti-dismantling ports arranged in the middle of the fixed plate, a lower light fixedly connected on the rear side of the upper light, an anti-dismantling mechanism slidably connected on the front side wall of the lower light, a card interface is arranged at the upper and lower ends of the rear side of the lower light, a wall-mounted bracket is connected to the rear side of the lower light, a desktop-mounted bracket is fixedly connected to the rear side of the wall-mounted bracket, the upper end of the desktop-mounted bracket is a vertical structure, and the lower end of the desktop-mounted bracket is an inclined structure, a fill light is arranged at the upper end of the upper light, two infrared lamps are arranged on the lower side of the fill light, an infrared light camera module and a visible light camera module are arranged in the middle of the two infrared lamps, a liquid crystal display is arranged in the middle of the upper light, and a card reading module is arranged at the lower end of the upper light. However, most of the existing devices use traditional two-dimensional sensors with similar structures, strong versatility of image data, and low difficulty in database construction. The spectral data cube required for multispectral face liveness recognition technology is synthesized from the spectral information and spatial information collected by the spectral camera. However, the working principles and performances of spectral cameras are different, and the spectra provided by the experimental light sources are also different. More error variables make it more difficult to obtain training data. It is easy for the spectral features of fake faces and real faces to be masked by other error variables, resulting in a lower success rate of liveness detection based on spectral features.

[0004] Therefore, it is necessary to improve a data acquisition device of a multi-spectral face liveness recognition technology in the prior art to solve the above problems. Summary of the invention

[0005] The utility model overcomes the shortcomings of the prior art, provides a data acquisition device for multi-spectral human face liveness recognition technology, and aims to solve the problems in the prior art.

[0006] To achieve the above-mentioned purpose, the technical solution adopted by the utility model is: a data acquisition device for multi-spectral face liveness recognition technology, comprising: an illumination mechanism, an image acquisition mechanism arranged on one side of the illumination mechanism, and a target face module to be detected for monitoring the target face;

[0007] The lighting mechanism includes: a plurality of groups of emission sources, each group of emission sources is the only light source of the image acquisition mechanism, and can be used to provide the same light source spectrum to cover the characteristic band of face recognition;

[0008] The image acquisition mechanism includes: a spectral camera and a color SLR camera symmetrically arranged with the spectral camera; the spectral camera and the color SLR camera are arranged opposite to the face module of the target to be measured, and the spectral camera and the color SLR camera have the same focal length;

[0009] The target face module to be tested controls the lighting mechanism and the image acquisition mechanism respectively.

[0010] In a preferred embodiment of the utility model, the target face module to be tested includes: a real face data acquisition unit and a high-imitation two-dimensional fake face data acquisition unit; the real face data acquisition unit and the high-imitation two-dimensional fake face data acquisition unit can respectively control the lighting mechanism, the image acquisition mechanism and the target face module to be tested to operate and perform detection.

[0011] In a preferred embodiment of the utility model, the emission source includes but is not limited to white light LED, incandescent lamp, halogen tungsten lamp, etc., and its spectral working band should be able to cover the visible light characteristic reflection spectrum of a real face obtained from the literature, so as to obtain the facial feature spectrum.

[0012] In a preferred embodiment of the present invention, the spectral camera is used to capture and photograph image grayscale data and its corresponding spectral data.

[0013] In a preferred embodiment of the present invention, the spectral camera, the color SLR camera and the emission source are all arranged tilted downward, and the focal points of the three converge at the same position.

[0014] In a preferred embodiment of the utility model, both the spectral camera and the color SLR camera are adjusted and controlled by the target face module to be measured.

[0015] In a preferred embodiment of the utility model, the real face data acquisition unit is electrically connected to the spectral camera and the color SLR camera respectively, and controls the spectral camera and the color SLR camera to shoot the human face to prepare a high-imitation two-dimensional fake face.

[0016] In a preferred embodiment of the utility model, the high-imitation two-dimensional fake face data acquisition unit is electrically connected to the spectral camera and the color SLR camera respectively, and controls the spectral camera and the color SLR camera to shoot the high-imitation two-dimensional fake face.

[0017] In a preferred embodiment of the utility model, the position of the high-imitation two-dimensional fake face during photographing is the same as that of the real face.

[0018] The utility model solves the defects existing in the background technology, and has the following beneficial effects:

[0019] (1) The utility model provides a data acquisition device for multi-spectral face liveness recognition technology. By using a digital camera to generate high-imitation two-dimensional fake face data that is similar to real face data in color, facial shadow, makeup, and face orientation but different in reflectance spectrum, the device greatly avoids the interference of error variables on the classification process based on spectral features, thereby improving the accuracy of spectral liveness judgment.

[0020] (2) The utility model provides a data acquisition device for multi-spectral face liveness recognition technology. Through this data acquisition device for multi-spectral face liveness recognition technology, the problem of fake face attacks in face recognition can be effectively solved while ensuring accuracy and security. At the same time, this method can also improve the reliability and robustness of face recognition, making it better adaptable to various complex environments and changing factors. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The utility model is further described below in conjunction with the accompanying drawings and embodiments;

[0022] Figure 1 This is a structural principle diagram of the real face data collection component provided by the utility model;

[0023] Figure 2 This is a schematic diagram of the component structure of the high-imitation two-dimensional fake face data acquisition device provided by the utility model;

[0024] In the figure: 101, target face module to be tested; 102, high-imitation two-dimensional fake face; 201, lighting mechanism; 301, image acquisition mechanism; 311, spectral camera; 312, color SLR camera. DETAILED DESCRIPTION

[0025] The present invention will now be further described in detail in conjunction with the accompanying drawings and embodiments. These drawings are simplified schematic diagrams that only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0026] like Figure 1 As shown, a data acquisition device for multi-spectral face liveness recognition technology includes: an illumination mechanism 201, an image acquisition mechanism 301 arranged on one side of the illumination mechanism 201, and a target face module 101 to be detected for monitoring a target face;

[0027] The lighting mechanism 201 includes: a plurality of groups of emission sources, each group of emission sources is the only light source of the image acquisition mechanism 301, and can be used to provide the same light source spectrum covering the facial recognition feature band to obtain facial features;

[0028] In a preferred embodiment of the present invention, the emission source includes but is not limited to a white light LED, a halogen tungsten lamp, an incandescent lamp, and the like.

[0029] In a preferred embodiment of the present invention, the spectral camera 311 includes three types: a multi-spectral camera 311, a hyperspectral camera 311 and a computational spectral camera 311. The spectral camera 311 is used to capture and shoot spectral data corresponding to the image grayscale data.

[0030] It should be noted that the same light source spectrum that the emission source can provide covers the visible light face recognition feature band. Visible light is an electromagnetic wave with a wavelength in the range of 380-780nm, and its wavelength range is from blue light to red light.

[0031] The image acquisition mechanism 301 includes: a spectral camera 311, and a color SLR camera 312 symmetrically arranged with the spectral camera 311; the spectral camera 311 and the color SLR camera 312 are arranged opposite to the target face module 101 to be measured, and the spectral camera 311 and the color SLR camera 312 have the same focal length;

[0032] In a preferred embodiment of the present invention, the spectral camera 311 , the color SLR camera 312 and the emission source are all arranged to be tilted downward, and the focal points of the three are gathered at the same position.

[0033] In a preferred embodiment of the present invention, both the spectral camera 311 and the color SLR camera 312 are adjusted and controlled by the target face module to be measured.

[0034] It should be noted that the image acquisition mechanism 301 includes a spectral camera 311 and a color SLR camera 312. The spectral camera 311 includes but is not limited to various multi-spectral cameras 311, hyperspectral cameras 311, and computational spectral cameras 311, which capture the spectral data of each pixel within the shooting range during shooting and construct a spectral data cube, so as to obtain the spectral data of each pixel during shooting recognition and construct spectral data; the color SLR camera 312 shoots high-definition color images without white balance correction.

[0035] The target face module 101 to be detected controls the lighting mechanism 201 , the image acquisition mechanism 301 and the target face module 101 to be detected to operate respectively.

[0036] In a preferred embodiment of the utility model, the target face module 101 to be tested includes: a real face data acquisition unit and a high-imitation two-dimensional fake face 102 data acquisition unit; the real face data acquisition unit and the high-imitation two-dimensional fake face 102 data acquisition unit can respectively control the lighting mechanism 201, the image acquisition mechanism 301 and the target face module 101 to be tested to operate and perform detection.

[0037] It should be noted that the target face module 101 to be measured is set in the middle position, the image acquisition mechanism 301 is set opposite to the target face module 101 to be measured, and the line connecting the image acquisition mechanism 301 and the target is the axis. Two groups of illumination light sources 201 are provided, and as the only light energy source for image acquisition, they are respectively deployed on both sides of the axis, and the direct direction is at the same angle with the axis to ensure that the target is evenly illuminated. The optical axis of the image acquisition mechanism 301, the center of the illumination light source 201 and the center point of the target face module 101 to be measured are in the same horizontal plane. The spectral camera 311 and the color SLR camera 312 are respectively located on both sides of the axis, and the lenses of the two cameras are aimed at the target, and the optical axes of the lenses are in the same horizontal plane and at the same angle with the axis. The angle should be as small as possible under the premise of avoiding obstruction to ensure that the spectral camera 311 and the color SLR camera 312 can capture a closer target image. While the spectral camera 311 captures the image information and spectral information of the target face, the color SLR camera 312 takes the original color image of the target face without white balance correction and produces a high-imitation two-dimensional fake face 102.

[0038] In a preferred embodiment of the present invention, the real face data acquisition unit is electrically connected to the spectral camera 311 and the color SLR camera 312 respectively, and controls the spectral camera 311 and the color SLR camera 312 to shoot the human face to prepare a high-imitation two-dimensional fake face 102.

[0039] It should be noted that the size of the high-imitation two-dimensional fake face 102 is similar to that of the target face. The image information and spectral information of the target face captured by the spectral camera 311, and the synthesized spectral data cube can be used as real face data for neural network training.

[0040] In a preferred embodiment of the present invention, the data acquisition unit of the high-imitation two-dimensional fake face 102 is electrically connected to the spectral camera 311 and the color SLR camera 312 respectively, and controls the spectral camera 311 and the color SLR camera 312 to shoot the high-imitation two-dimensional fake face 102.

[0041] In a preferred embodiment of the present invention, the position of the high-imitation two-dimensional fake face 102 during photographing is the same as that of the real face.

[0042] It should be noted that the position of the high-imitation two-dimensional fake face 102 is the same as the position of the target face module 101 to be tested described in the real face data acquisition component, and the deployment positions of the lighting source 201 and the spectral camera 311 are respectively the same as the lighting source 201 and the spectral camera 311 described in the real face data acquisition component. The spectral camera 311 captures the image information and spectral information of the high-imitation two-dimensional fake face 102 under this device, and the synthesized spectral data cube can be used as high-imitation fake face data for neural network training.

[0043] When the utility model is used, first, a real face image is collected through a real face data collection unit; after collecting the real face data, the face feature data is processed for subsequent fake face production, and the specific processing includes denoising and enhancing the image to improve the image quality and make it more suitable for fake face production; according to the processed real face data, a highly simulated two-dimensional fake face image is generated by using computer graphics and image processing technology; using the prepared high-imitation two-dimensional fake face 102 image, data is collected through the high-imitation two-dimensional fake face 102 data collection unit to obtain the difference in spectral information between the fake face and the real face under various light, angle and environmental conditions; finally, the collected data is deeply analyzed and used. By comparing the real and fake face spectral data, the difference between the two can be found, and the face recognition algorithm can be further optimized to improve the recognition accuracy and robustness of the system.

[0044] Through this multi-spectral face liveness recognition technology data acquisition device, the problem of fake face attacks in face recognition can be effectively solved while ensuring accuracy and security. At the same time, this method can also improve the reliability and robustness of face recognition, making it better adaptable to various complex environments and changing factors.

[0045] The above is based on the ideal embodiment of the utility model. Through the above description, relevant personnel can make various changes and modifications without deviating from the technical concept of the utility model. The technical scope of this utility model is not limited to the content in the specification, and the technical scope must be determined according to the scope of the claims.

Claims

1. A data acquisition device for multi-spectral face liveness recognition technology, comprising:

1. An illumination mechanism (201), an image acquisition mechanism (301) arranged on one side of the illumination mechanism (201), and a target face module (101) for monitoring a target face, characterized in that: The lighting mechanism (201) comprises: a plurality of groups of emission sources, each group of the emission sources being the only light source of the image acquisition mechanism (301) and capable of providing the same light source spectrum covering a characteristic waveband for face recognition; The image acquisition mechanism (301) comprises: a spectral camera (311), and a color SLR camera (312) symmetrically arranged with the spectral camera (311); the spectral camera (311) and the color SLR camera (312) are arranged directly opposite to the target face module (101) to be measured, and the spectral camera (311) and the color SLR camera (312) have the same focal length; The target face module (101) to be detected controls the operation of the lighting mechanism (201) and the image acquisition mechanism (301) respectively.

2. The data acquisition device of the multi-spectral face liveness recognition technology according to claim 1 is characterized in that: The target face module (101) to be tested comprises: a real face data acquisition unit and a high-imitation two-dimensional fake face (102) data acquisition unit; the real face data acquisition unit and the high-imitation two-dimensional fake face (102) data acquisition unit can respectively control the lighting mechanism (201), the image acquisition mechanism (301) and the target face module (101) to operate and perform detection.

3. The data acquisition device of the multi-spectral face liveness recognition technology according to claim 1 is characterized in that: The emission source includes but is not limited to white light LED, incandescent lamp, halogen tungsten lamp, and its spectral working band should be able to cover the visible light characteristic reflection spectrum of real human face obtained from the literature, so as to obtain the facial feature spectrum.

4. The data acquisition device of the multi-spectral face liveness recognition technology according to claim 1, characterized in that: The spectral camera (311) is used to capture and photograph image grayscale data and its corresponding spectral data.

5. The data acquisition device of multi-spectral face liveness recognition technology according to claim 1, characterized in that: The spectral camera (311), the color SLR camera (312) and the emission source are all arranged tilted downward, and the focal points of the three converge at the same position.

6. The data acquisition device of multi-spectral face liveness recognition technology according to claim 1, characterized in that: The spectral camera (311) and the color SLR camera (312) are both adjusted and controlled by the target face module (101).

7. The data acquisition device of multi-spectral face liveness recognition technology according to claim 2, characterized in that: The real face data acquisition unit is electrically connected to the spectral camera (311) and the color SLR camera (312) respectively, and controls the spectral camera (311) and the color SLR camera (312) to shoot a human face, thereby preparing a high-imitation two-dimensional fake face (102).

8. The data acquisition device of multi-spectral face liveness recognition technology according to claim 7, characterized in that: The high-imitation two-dimensional fake face (102) data acquisition unit is electrically connected to the spectral camera (311) and the color SLR camera (312) respectively, and controls the spectral camera (311) and the color SLR camera (312) to shoot the high-imitation two-dimensional fake face (102).

9. The data acquisition device of multi-spectral face liveness recognition technology according to claim 7, characterized in that: The position of the high-imitation two-dimensional fake face (102) during photographing is the same as that of the real face.