KR-20260062456-A - Apparatus and method for identifying fake face using images of various wavelengths

KR20260062456AKR 20260062456 AKR20260062456 AKR 20260062456AKR-20260062456-A

Abstract

An apparatus and method for detecting a forged face using images of various wavelength ranges are provided. The forged face detection apparatus acquires a first image, a second image, and a third image of different wavelength ranges in which a face is captured, respectively, and acquires a first detection coordinate based on the features of the face from the first image. Then, based on a transformation matrix, it acquires a first matching detection coordinate within the second image and a second matching detection coordinate within the third image that match the first detection coordinate, respectively; acquires a first face region corresponding to the first detection coordinate from the first image; acquires a second face region corresponding to the first matching detection coordinate from the second image; and acquires a third face region corresponding to the second matching detection coordinate from the third image. Subsequently, data corresponding to the first face region, the second face region, and the third face region, respectively, are input into a pre-trained forgery detection model to determine whether the captured face is forged.

Inventors

김현지
전승우
장우진
조현수

Assignees

주식회사 에스원

Dates

Publication Date: 20260507
Application Date: 20241029

Claims (14)

As a method for detecting forged faces, A discrimination device acquires a first image, a second image, and a third image of different wavelength ranges, respectively, of a face; The above-described discrimination device obtains first detection coordinates based on the features of the face from the first image; The above-described discrimination device acquires, respectively, a first matching detection coordinate in the second image and a second matching detection coordinate in the third image that are matched to the first detection coordinate based on a transformation matrix; The above-described discrimination device acquires a first face region corresponding to the first detection coordinates from the first image, acquires a second face region corresponding to the first matching detection coordinates from the second image, and acquires a third face region corresponding to the second matching detection coordinates from the third image; and The above-described discrimination device inputs data corresponding to the first face region, the second face region, and the third face region, respectively, into a pre-trained forgery discrimination model to determine whether the captured face is forged. A method for detecting forged faces including
In paragraph 1, Step of training the above forgery detection model Includes more, The above training step involves applying data corresponding to the first face region, the second face region, and the third face region, respectively, to individual models for detecting forged faces, each corresponding to the different wavelengths of the forgery detection model, thereby training the individual models; and A step of training a final classifier based on an ensemble network of the forgery detection model using the analysis results of individual models corresponding to the respective different wavelengths. A method for detecting a forged face, including
In paragraph 1, The first, second, and third images of different wavelength ranges of the above-mentioned face each include an image of a real person's face and a forged image, and A method for detecting a forged face, wherein the first image is a visible light image captured by a visible light camera, the second image is a thermal image captured by a thermal imaging camera, and the third image is an infrared image captured by an infrared camera.
In paragraph 3, The second image and the third image are images taken while the subject is wearing means to facilitate face detection, and the means include glasses and a nose model, a learning method.
In paragraph 1, The step of the above-described discrimination device acquiring, respectively, a first matching detection coordinate in the second image and a second matching detection coordinate in the third image that are matched to the first detection coordinate based on a transformation matrix, A step of obtaining a first matching detection coordinate within the second image using a first transformation matrix that uses a first matrix corresponding to the first detection coordinate and a second matrix corresponding to the second detection coordinate obtained based on the features of the face from the second image; and A step of obtaining second matching detection coordinates within the third image by using a second transformation matrix that utilizes a first matrix corresponding to the first detection coordinates and a third matrix corresponding to the third detection coordinates obtained based on the features of the face from the third image. A method for detecting a forged face, including
In paragraph 5, The first transformation matrix above is as follows: X=(At*A)^-1*At*B It is based on, where X represents the first transformation matrix, A represents the first matrix, At represents the transpose of the first matrix, and B represents the second matrix, The second transformation matrix above is as follows: X'=(At*A)^-1*At*C A method for detecting a forged face, based on, wherein X' represents the second transformation matrix, A represents the first matrix, At represents the transpose of the first matrix, and C represents the third matrix.
In paragraph 5, The first detection coordinates include coordinates corresponding to the eyes, coordinates corresponding to the nose, coordinates corresponding to the mouth, and the distance between the eyes on the face of the first image. The second detection coordinates above include coordinates corresponding to the eyes, coordinates corresponding to the nose, and coordinates corresponding to the mouth in the face of the second image, and A method for detecting a forged face, wherein the third detection coordinates include coordinates corresponding to the eyes, coordinates corresponding to the nose, and coordinates corresponding to the mouth in the face of the third image.
As a forged face detection device, An image acquisition unit configured to acquire a first image, a second image, and a third image of different wavelength ranges, respectively, of a face; A face feature detection unit configured to obtain a first detection coordinate based on the features of the face from the first image, a second detection coordinate based on the features of the face from the second image, and a third detection coordinate based on the features of the face from the third image; A coordinate matching unit configured to acquire, respectively, a first matching detection coordinate in the second image and a second matching detection coordinate in the third image that are matched to the first detection coordinate based on a transformation matrix; A face detection unit configured to acquire a first face region corresponding to the first detection coordinates from the first image, acquire a second face region corresponding to the first matching detection coordinates from the second image, and acquire a third face region corresponding to the second matching detection coordinates from the third image; and A forgery detection unit configured to determine whether the captured face is forged by inputting data corresponding to the first face region, the second face region, and the third face region, respectively, into a pre-trained forgery detection model. A forged face detection device including
In paragraph 8, A learning processing unit configured to apply data corresponding to the first face region, the second face region, and the third face region, respectively, to individual models for detecting forged faces corresponding to the different wavelengths of the forgery detection model to train the individual models, and to train a final classifier based on an ensemble network of the forgery detection model using the analysis results of the individual models corresponding to the different wavelengths. A forged face detection device further comprising
In Paragraph 9, The forgery detection model includes a first individual model corresponding to the first image, a second individual model corresponding to the second image, a third individual model corresponding to the third image, and an ensemble network-based final classifier that integrates the analysis results of the first individual model, the analysis results of the second individual model, and the analysis results of the third individual model. A forged face detection device comprising the first individual model, the second individual model, and the third individual model, which are formed by a DENSENET structure.
In paragraph 8, The first, second, and third images of different wavelength ranges of the above-mentioned face each include an image of a real person's face and a forged image, and A forged face detection device, wherein the first image is a visible light image captured by a visible light camera, the second image is a thermal image captured by a thermal imaging camera, and the third image is an infrared image captured by an infrared camera.
In paragraph 8, Specifically, the above coordinate matching unit is, A first transformation matrix is used with a first matrix corresponding to the first detection coordinates and a second matrix corresponding to the second detection coordinates obtained from the second image based on the features of the face, to obtain the first matching detection coordinates within the second image, and A forged face detection device configured to obtain second matching detection coordinates within the third image by using a second transformation matrix that uses a first matrix corresponding to the first detection coordinates and a third matrix corresponding to the third detection coordinates obtained based on the features of the face from the third image.
In Paragraph 12, The first transformation matrix above is as follows: X=(At*A)^-1*At*B It is based on, where X represents the first transformation matrix, A represents the first matrix, At represents the transpose of the first matrix, and B represents the second matrix, The second transformation matrix above is as follows: X'=(At*A)^-1*At*C A forged face detection device based on, wherein X' represents the second transformation matrix, A represents the first matrix, At represents the transpose of the first matrix, and C represents the third matrix.
In Paragraph 12, The above-mentioned face feature detection unit is, Obtaining the first detection coordinates including coordinates corresponding to the eyes, coordinates corresponding to the nose, coordinates corresponding to the mouth, and the distance between the eyes in the face of the first image, and The second detection coordinates are obtained, including coordinates corresponding to the eyes, coordinates corresponding to the nose, and coordinates corresponding to the mouth in the face of the second image. A forged face detection device configured to acquire the third detection coordinates, including coordinates corresponding to the eyes, coordinates corresponding to the nose, and coordinates corresponding to the mouth in the face of the third image.

Description

Apparatus and method for identifying fake face using images of various wavelengths The present invention relates to face identification, and more specifically, to a forged face identification device and method utilizing images of various wavelength ranges. Facial recognition access systems utilizing biometric technology are currently being widely commercialized. However, since attempted attacks using forged faces can threaten access security, a method to detect forged faces capable of distinguishing between real and fake faces is required. In the case of forgery detection using thermal imaging, an image is generated by measuring the thermal distribution pattern of an object, enabling forgery detection based on the thermal characteristics of the object. Furthermore, since it is not significantly affected by dark environments or light sources, it can be utilized as a reliable forgery detection tool under various conditions. In the case of forgery detection using infrared light, visualizing the differences in the light reflection, absorption, and transmission characteristics of an object allows for the precise visualization of features invisible to the naked eye, and utilizing the unique characteristics of a face can improve the accuracy of forgery detection. In the case of counterfeit detection using visible light, images are acquired and analyzed by detecting light of the visible light spectrum. Visible light images provide high-resolution images, which facilitate the analysis of visual features and pattern recognition of objects, and can be utilized to improve counterfeit detection performance by analyzing the shapes and features of real faces and counterfeit faces. However, since forgery detection using thermal images, infrared images, or visible light is performed by utilizing individual cameras for each method, there are limitations in performance, and it is not possible to accurately detect the difference between a forged face and a real face. Related technology includes "image processing device, image capture system, image processing method and storage medium" disclosed in U.S. Patent Publication No. US 2019-0347775. FIG. 1 is a diagram showing the structure of a forged face detection device utilizing images of various wavelength ranges according to an embodiment of the present invention. Figure 2 is a figure showing an example of a thermal image according to an embodiment of the present invention. FIG. 3 is a figure showing an example of coordinate matching according to an embodiment of the present invention. FIG. 4 is a conceptual diagram showing the operation function of a learning processing unit according to an embodiment of the present invention. FIG. 5 is a flowchart of a learning method according to an embodiment of the present invention. FIG. 6 is a conceptual diagram of a method for detecting a forged face according to an embodiment of the present invention. FIG. 7 is a flowchart of a method for detecting a forged face according to an embodiment of the present invention. FIG. 8 is a structural diagram illustrating a computing device for implementing a method according to an embodiment of the present invention. Embodiments of the present invention are described below with reference to the attached drawings so that those skilled in the art can easily implement them. However, the present invention may be embodied in various different forms and is not limited to the embodiments described herein. Furthermore, in order to clearly explain the present invention in the drawings, parts unrelated to the explanation have been omitted, and similar parts throughout the specification are denoted by similar reference numerals. Throughout the specification, when a part is described as "including" a certain component, this means that, unless specifically stated otherwise, it does not exclude other components but may include additional components. Expressions described in the singular in this specification may be interpreted as singular or plural unless explicit expressions such as "one" or "single" are used. Additionally, terms including ordinal numbers, such as first, second, etc., used in embodiments of the present invention may be used to describe components, but the components should not be limited by these terms. The terms are used solely for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the first component may be named the second component, and similarly, the second component may be named the first component. Hereinafter, a forged face detection device and method utilizing images of various wavelength ranges according to an embodiment of the present invention will be described. FIG. 1 is a diagram showing the structure of a forged face detection device utilizing images of various wavelength ranges according to an embodiment of the present invention. A forged face detection device (1) according to an embodiment of the pres