Detection device and detection method for human face skin

Two hyperspectral cameras collect hyperspectral images of the face skin from different perspectives and perform three-dimensional spectral image correlation calculations, solving the impact of structured light projection on the skin and insufficient accuracy of traditional technology, and achieving efficient and accurate faces Skin testing, suitable for long-term testing.

CN118787316BActive Publication Date: 2024-11-12TIANJIN UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411273947.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-12
Publication Date
2024-11-12
Estimated Expiration
2044-09-12

AI Technical Summary

Technical Problem

The existing facial skin detection technology has problems such as thermal effects and chronic damage to the skin by structured light projection, low measurement accuracy, long time for traditional hyperspectral imaging, and difficulty in meeting the requirements of micro-expression detection in pure visual solutions.

Method used

Two hyperspectral cameras were used to collect hyperspectral images of human face skin from different perspectives, and the strain of each pixel was determined through three-dimensional spectral image correlation calculations. Only spectral information in the three-band blue, blue and green bands were collected to avoid structured light projection. and direct contact to achieve fully passive measurement.

Benefits of technology

It realizes long-term facial skin detection without any impact, improves the accuracy and efficiency of measurement, reduces equipment costs, and is suitable for people who are sensitive to the test environment, capturing slight changes and micro-strains of facial skin.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118787316B_ABST
    Figure CN118787316B_ABST
Patent Text Reader

Abstract

The present application discloses a detection device for human facial skin, comprising: a first hyperspectral camera, used to obtain an image and a hyperspectral image of human facial skin from a first viewing angle; a second hyperspectral camera, used to obtain an image and a hyperspectral image of human facial skin from a second viewing angle; a calculation module, which determines the spectral characteristics of each pixel point as its identity information through the hyperspectral images obtained from the first viewing angle and the second viewing angle, uses the identity information to perform a matching search, and obtains the strain of the sub-area where each point is located through three-dimensional spectral image correlation calculation of the image images collected from the first viewing angle and the second viewing angle; the first hyperspectral camera and the second hyperspectral camera only collect spectral information of three bands to ensure the acquisition speed of the hyperspectral image; the present application also discloses a detection method for human facial skin.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The embodiments of the present application relate to the field of facial skin detection, and in particular, to a detection device and method for facial skin. Background Art

[0002] In the field of facial skin detection technology, structured light, as a non-contact measurement technology, can be applied to obtain the three-dimensional shape of the face. Structured light technology projects a specific light pattern on the surface of the face and uses a camera to capture the deformation of the pattern, and then calculates the three-dimensional coordinates of the face. However, the existing technology has several problems and challenges in practical applications.

[0003] Structured light projection may have an impact on the performance of facial skin. The encoding, decoding and projector-camera calibration of structured light are key technologies, but the projection intensity and wavelength of structured light may cause certain thermal effects on the skin, leading to dryness, redness and other discomforts, affecting the accuracy of the measurement. In addition, long-term or high-intensity exposure may also cause chronic damage to the skin. Especially in fields such as lie detection that require continuous marathon fatigue tests, projecting visible or invisible structured light onto the face for a long time will not only cause changes in expressions and facial skin performance due to structured light, but this long-term structured light projection will also damage the eyes, resulting in abnormal expressions.

[0004] Although speckles can be directly applied or painted by pen or other means, this method may temporarily change the color and texture of the skin, resulting in inaccurate measurement results. In addition, this direct contact method may cause psychological discomfort to the subject, resulting in abnormal expressions, especially in sensitive areas, leading to distorted test results.

[0005] Traditional hyperspectral imaging takes several seconds or even more than ten seconds to capture each image, resulting in missing out on subtle facial expression changes or transient micro-changes in the skin during hyperspectral imaging.

[0006] The disadvantage of the pure visual solution to track specific points on the human face is that the accuracy is difficult to meet the requirements. Although it can be located through pores or virtual specific points, its accuracy cannot meet the requirements of micro-expression detection or micro-strain detection of the human face skin. Summary of the invention

[0007] In view of this, an embodiment of the present application provides a detection device for facial skin, comprising:

[0008] A first hyperspectral camera, used to obtain a video image and a hyperspectral image of a human face skin from a first perspective;

[0009] A second hyperspectral camera, used for acquiring an image of human face skin and a hyperspectral image from a second viewing angle;

[0010] A calculation module, which determines the spectral characteristics of each pixel point as its identity information through the hyperspectral images obtained from the first viewing angle and the second viewing angle, uses the identity information to perform matching search, and obtains the strain of the sub-area where each point is located through the image images collected from the first viewing angle and the second viewing angle through three-dimensional spectral image correlation calculation;

[0011] The first hyperspectral camera and the second hyperspectral camera only collect spectral information of three bands: a blue light band of 461nm to 491nm, a cyan light band of 492nm to 540nm, and a green light band of 542nm to 580nm;

[0012] The first hyperspectral camera and the second hyperspectral camera have the same resolution, which is 512×512 to 1024×1024, to ensure the acquisition speed of the hyperspectral image;

[0013] The three-dimensional spectral image correlation calculation includes the following steps:

[0014] The spectral identity information is used to perform three-dimensional spectral image correlation calculation on each point of the image captured by the two hyperspectral cameras, and two sets of three-dimensional coordinates of each point of the facial skin under two different stretching states are obtained respectively;

[0015] Differencing the two sets of three-dimensional coordinates to obtain a three-dimensional displacement of each point;

[0016] The maximum principal strain of each point is calculated based on the three-dimensional displacement and is used as the strain of the facial skin at that point.

[0017] According to another aspect of the present application, a method for detecting facial skin is also provided, comprising:

[0018] Acquire hyperspectral images and video images of facial skin in two different stretching states from a first perspective and a second perspective respectively;

[0019] Determine the spectral characteristics of each pixel point as its identity information through the hyperspectral images obtained from the first viewing angle and the second viewing angle, use the identity information to perform matching search, and obtain the strain of the sub-area where each point is located through the image images collected from the first viewing angle and the second viewing angle through three-dimensional spectral image correlation calculation;

[0020] The first hyperspectral camera and the second hyperspectral camera only collect spectral information of three bands: a blue light band of 461nm to 491nm, a cyan light band of 492nm to 540nm, and a green light band of 542nm to 580nm;

[0021] The first hyperspectral camera and the second hyperspectral camera have the same resolution, which is 512×512 to 1024×1024, to ensure the acquisition speed of the hyperspectral image;

[0022] The three-dimensional spectral image correlation calculation includes the following steps:

[0023] The spectral identity information is used to perform three-dimensional spectral image correlation calculation on each point of the image captured by the two hyperspectral cameras, and two sets of three-dimensional coordinates of each point of the facial skin under two different stretching states are obtained respectively;

[0024] Differencing the two sets of three-dimensional coordinates to obtain a three-dimensional displacement of each point;

[0025] The maximum principal strain of each point is calculated based on the three-dimensional displacement and is used as the strain of the facial skin at that point.

[0026] Non-limitingly, the beneficial effects of this application include at least but are not limited to:

[0027] The "fully passive" measurement method does not cause any active influence on the human body, thus avoiding any slight changes in the human body or the test subject's psychology or physiology that may cause changes in facial skin expressions or changes in facial skin characteristics, and avoiding measurement distortion to the greatest extent, minimizing the impact of the measurement factors themselves on the measurement results. Traditional test methods may affect the test results because they use infrared, visible or invisible light to illuminate the human skin. These methods are not suitable for long-term testing because they may make the test subject feel nervous, thus affecting the credibility of the results.

[0028] In addition, structured light and speckle coating are not suitable for people who are sensitive to the test environment, such as infants or polygraph subjects, because too many equipment or heat light sources may cause their emotions to fluctuate, resulting in inaccurate results.

[0029] In the field of medical cosmetology, the projection of structured light may affect the effect of the injection. Therefore, a fully passive detection method that does not damage or affect the facial skin is needed. The speckle applied by a water pen may not have much psychological or physiological impact on the person being tested, but the application process itself may have a psychological or physiological impact on the person being tested.

[0030] This application does not require external personnel or projection devices, thus avoiding external interference. Such a method is more suitable for long-term testing because it allows the testee to relax and obtain more reliable results.

[0031] Furthermore, the grayscale information provided by the speckle used in the traditional digital image correlation method only contains information from one channel, so it is not only necessary to artificially create the speckle, but also requires a larger sub-area to provide sufficient features. The identity information used in this application can contain spectral information of multiple bands, which is richer and more unique. Therefore, the identity information can be used for matching search, and the strain of the sub-area where each point is located can be obtained by calculating the correlation of the three-dimensional spectral image through the image collected from the first and second perspectives.

[0032] Furthermore, since only specific bands are imaged and collected through a hyperspectral camera with relatively low resolution, the efficiency of hyperspectral imaging is much higher than that of traditional hyperspectral camera imaging, which is conducive to capturing instantaneous tiny changes and instantaneous micro-strains of human facial skin.

[0033] Finally, from a hardware perspective, the embodiments of the present application do not require a separate structured light projection device, nor do they require adjustment of the structured light projection device, thereby reducing equipment costs and debugging costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the embodiments of the present application. For those skilled in the art, other drawings can also be obtained based on these drawings.

[0035] Figure 1 is a schematic diagram of a detection device for human facial skin provided in an optional embodiment of the present application;

[0036] Figure 2 is a spectral image of human face skin to a light wave of 480nm wavelength acquired from a hyperspectral camera in an optional embodiment of the present application;

[0037] Figure 3 is a spectral image of human face skin to a light wave of 538 nm wavelength acquired from a hyperspectral camera in an optional embodiment of the present application;

[0038] Figure 4 is a spectral image of human face skin to a light wave of 576 nm wavelength acquired from a hyperspectral camera in an optional embodiment of the present application;

[0039] Figure 5 A flowchart of a method for finding the original reference point based on spectral identity information before and after facial skin deformation is shown;

[0040] Figure 6A method for implementing 3D-DIC by three 2D-DICs to obtain the spatial position of a specific point in an optional embodiment of the present application is shown;

[0041] Figure 7 A flow chart of the steps of a 2D-DIC method in an optional embodiment of the present application is shown. DETAILED DESCRIPTION

[0042] In order to enable those skilled in the art to better understand the technical solutions in the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments in the embodiments of the present application should fall within the scope of protection of the embodiments of the present application.

[0043] In this application, a hyperspectral camera can take photos like an ordinary camera, and can also collect spectral images of target objects, which is called spectral imaging. In this application, photos taken like ordinary cameras are called "image images" and the images taken with light reflectivity information are called "spectral images". Since the spectral field of human facial skin before and after deformation only changes within a specific range, the strain of human facial skin in different states can be detected by tracking points with the same spectral information.

[0044] Non-restrictive, such as Figure 1 As shown, a detection device for facial skin is shown, including two halogen light sources for providing fill light from both sides of a subject 10, namely a first light source 11 and a second light source 12 (also referred to as a "right light source"), a first hyperspectral camera 13 arranged on the left and right sides of the subject 10, a second hyperspectral camera 14 (located on the right side from the subject 10, also referred to as a "right camera"), and a matching computing module 15 for data storage and processing.

[0045] Preferably, Figure 1 The first hyperspectral camera 13 and the second hyperspectral camera 14 are arranged at the same height facing the subject and approximately at the same height as the subject's nose tip, and are evenly and symmetrically distributed on both sides of the subject's nose bridge, and are approximately the same distance from the subject's left ear and right ear, that is, the angle α=β in the figure and the sum of the two is 10° to 120°, preferably greater than or equal to 60° and less than or equal to 90°, and when the angle is 90°, the amount of calculation related to the spectral image is the lowest.

[0046] like Figure 2 , Figure 3 , Figure 4As shown, it shows the facial spectral field of 852 areas (852 acquisition points) of the human face skin under no obvious stretching or loading conditions collected by a 512×512 hyperspectral camera, that is, the spectral reflectance distribution field; because the traditional spectral information acquisition is full spectrum information, it takes more than ten seconds or even dozens of seconds to collect once. In order to improve the acquisition efficiency and quickly capture the changes in the facial spectrum caused by the instantaneous micro-expression changes, the embodiment of the present application Figure 2 , Figure 3 , Figure 4 Only the reflection data of 480nm, 538nm and 576nm wavelengths are collected and analyzed. It can be seen from the figure that the spectral field of each position in the human face skin is different. Therefore, the spectral field of each point or each sub-area can be used to mark each point, so that the specific point can be regarded as a "virtual speckle" to track and search it, and detect its displacement and strain.

[0047] It should be understood that in order to avoid missing the strain caused by slight changes in facial skin, the resolution of the two hyperspectral cameras should not exceed 1024×1024. Preferably, sensitive bands can be selected in a targeted manner according to the type of facial skin for spectral information collection to improve resolution and collection efficiency. The hyperspectral camera only collects spectral information of three bands where the reflectance of facial skin changes more obviously, namely the blue light band, the cyan light band and the green light band; wherein the blue light band is 461nm to 491nm, the cyan light band is 492nm to 540nm, and the green light band is 542nm to 580nm.

[0048] like Figure 2 , Figure 3 and Figure 4 As shown in the figure, the sub-area pointed by the arrow can be abstracted as a point-shaped speckle, whose reflectivity to 480nm light is about 0.145, to 538nm light is about 0.152, and to 576nm light is about 0.249. Therefore, the identity information of this point (point X) is: 480-0.145; 538-0.152; 576-0.249 or abbreviated as (0.145, 0.152, 0.249). When the facial skin is stretched or deformed, the points with the same or closest identity information (when there is no same point) are regarded as "virtual speckles" of the same mark.

[0049] Next, if Figure 5 As shown, it is explained how to define the point "closest" to the identity information. In this embodiment, a specific point (point X) before the deformation of the face skin is regarded as a reference point, and the identity information of each point in the deformed face spectrum image is compared with the reference point. The steps include:

[0050] S501, select a reference point and record its spectral identity information, non-limitingly, for example, if point X is selected, its identity information is recorded as (0.145, 0.152, 0.249);

[0051] S502, compare the spectral identity information of each point of the facial skin after deformation with the above identity information. For example, there are X1, X2...Xn points after deformation. If there is a point whose identity information is completely consistent with the reference point X before deformation, execute step S503 to identify the point as the original reference point; otherwise, execute step S504 to calculate the distance between the reference point and each other point. If there is an extreme value, that is, the minimum value, execute step S505 to use it as the reference point; if it is determined according to step S506 that there are two or more extreme values, it is necessary to further execute step S507 to identify the point with the smallest absolute value among the three coordinates of the distance as the original reference point.

[0052] Next, the above method is illustrated by taking the simplest search of two points (Xn, n=2).

[0053] There are two points X1 (0.149, 0.153, 0.246) and X2 (0.142, 0.151, 0.249). The three reflectivities are calculated by arithmetic difference with point X, and we get X-X1=(-0.004, -0.001, 0.003) and X-X2=(0.003, 0.001, 0.000). The absolute value of the difference of each reflectivity is added, and the point with the smallest sum is considered to be the point closest to point X, that is, point X (here, point X2). It should be understood that if there are two points whose absolute values ​​are equal to point X, the point with the smallest absolute value among the three differences is taken as the closest point.

[0054] Preferably, the identity information of the two points can be regarded as two positions in the three-dimensional space. For example, the identity information of the point X1 is (0.149, 0.153, 0.246). Its identity information can be regarded as the coordinate information in the three-dimensional space. The three-dimensional spatial distance between different points and the reference point can be calculated to determine the degree of approximation to the reference point X (0.145, 0.152, 0.249). The point with the closest "spatial distance" is regarded as the most approximate point. The calculation of the spatial distance can be calculated using the following approximate "Euclidean" formula.

[0055] d=(X a -Y a ) 2 +(X b -Y b ) 2 +(X c -Y c )2

[0056] Among them, X a , X b , X c are the three values ​​of the identity information of the reference point, i.e., the reflectivity corresponding to the three light waves, Y a , Y b , Y c are the three values ​​of the identity information of another point, that is, the reflectivity of the three light waves. According to the above formula, it can be calculated that the "spatial distance" between point X1 and point X is 26×10 -6 , the "spatial distance" between point X2 and point X is 10×10 -6 , so point X2 is considered to be closer to point X. Further, if the approximate "Euclidean distance" of the two points is the same, then the Chebyshev distance, Manhattan distance, Minkowski distance, Mahalanobis distance or Mahalanobis distance between them is calculated to determine the point that is closest to point X.

[0057] Next, an example is given to illustrate how to use virtual speckle to perform three-dimensional spectral image correlation calculation to obtain the strain of the sub-area, including:

[0058] Performing three-dimensional spectral image correlation (3D-DIC) calculation on the image images collected by the first hyperspectral camera 13 and the second hyperspectral camera 14 to respectively obtain the three-dimensional coordinates (two sets of three-dimensional coordinates) of the X point of the facial skin in different states;

[0059] The spectral identity information is used to perform three-dimensional spectral image correlation calculation on each point of the image captured by the two hyperspectral cameras, and two sets of three-dimensional coordinates of each point of the facial skin under two different stretching states are obtained respectively;

[0060] Differencing the two sets of three-dimensional coordinates to obtain a three-dimensional displacement of each point;

[0061] The maximum principal strain of each point is calculated based on the three-dimensional displacement and is used as the strain of the facial skin at that point.

[0062] like Figure 6 As shown, the three-dimensional spectral image correlation calculation can be regarded as the superposition of multiple (for example, three) two-dimensional spectral image correlation calculations (2D-DIC). The two images of the sub-area where the point X is located before deformation obtained from the first hyperspectral camera 13 (left camera) and the second hyperspectral camera 14 (right camera) can be subjected to a two-dimensional spectral image correlation operation (2D-DIC) to obtain the three-dimensional coordinates of the middle point X of the sub-area before deformation ( X 0, Y 0, Z0); Similarly, by performing two 2D-DIC operations on the two images obtained by the two cameras after deformation and the image obtained by any hyperspectral camera before deformation, the three-dimensional coordinates of the center point of the specific sub-area after deformation can be obtained ( X 1, Y 1, Z 1); Non-limitingly, the above-mentioned "sub-area" can be understood as a rectangular area centered on a specific point on the face skin that needs to be calculated.

[0063] In a non-limiting manner, "point X" can be understood as the center point of a specific sub-area of ​​the face area. That is, the three-dimensional coordinates (two sets of three-dimensional coordinates) of the specific point X of the face skin under different stretching states can be obtained through three-dimensional spectral image correlation calculations.

[0064] It should be understood that the spectral information of each area of ​​the human face skin is actually different under different stretching states, but before and after the slight deformation, there is always a point with the same identity information (a very similar point) in the vicinity of a specific point. Therefore, although the spectral identity information of each point under different stretching states is different, similar to the grayscale information provided by speckle for each pixel, when it is impossible to find two points with completely consistent information under different stretching states, the spectral identity information can be used as the only feature information provided by similar speckle to perform three-dimensional spectral image correlation calculation on point X.

[0065] Optionally, in the above process, since the thickness change of the skin during deformation is extremely small, the thickness change has little effect on the strain of the skin in the x and y directions, so it is omitted in the modeling process to improve the calculation efficiency and save calculation resources. Therefore, the strain of the facial skin based on a single point without considering the change in the z direction (skin depth direction) can be obtained as follows: Figure 7 The calculation method shown in the figure, that is, the above two-dimensional spectral image correlation calculation method can be implemented through the following steps.

[0066] S701, calculating the spectral identity information of a specific sub-area in one camera and calculating the corresponding spectral identity information in another camera based on the information; non-limiting implementation may be as follows:

[0067] Determine the initial calculation point in the left camera image, and determine a sub-area with the initial calculation point as the center and record the spectral identity information as , which is different from the traditional digital image correlation method, the digital image is stored in the computer in the form of a matrix, and the matrix elements are the grayscale values ​​required for the matching calculation. The spectral image contains richer spectral information, and the storage form in the computer is a multi-dimensional array, that is, each pixel contains the spectral information of several bands, and the calculation method of the spectral identity information is:

[0068]

[0069] Similarly, the spectral identity information of the sub-area corresponding to the same position in the right camera is recorded as , calculated as:

[0070]

[0071] in and An index representing the pixel position; and Represents the reflectivity of blue light with a wavelength of 461nm to 491nm; and Represents the reflectivity of cyan light with a wavelength of 492nm to 540nm; and It represents the reflectivity of green light with a wavelength of 542nm to 580nm.

[0072] Different from the traditional digital image correlation method, the characteristics of spectral information are stronger than those provided by pure grayscale information. Therefore, the sub-region required for matching is smaller, and it can be assumed that there is no stretching or shearing deformation in the sub-region. Therefore, when performing spectral image correlation calculation, the assumption of zero-order shape function can be directly used to calculate and , further improving the computational efficiency.

[0073]

[0074] in, u To calculate the displacement of a point in the x direction, v is the displacement of the calculation point in the y direction.

[0075] S701 is performed when the facial skin is in a relatively calm state, or when the subject is in a relatively stable emotional state.

[0076] S702, establish a correlation function type of the spectral identity information of the sub-area in the two cameras and optimize it iteratively. When the function obtains an extreme value, the deformation amount contained in the shape function is the deformation amount to be calculated; the non-limiting implementation can be as follows:

[0077] Described by the following correlation function and The correlation degree of the sub-areas before and after deformation is optimized and iterated. When the correlation function reaches the extreme value, the deformation contained in the shape function is the deformation to be calculated.

[0078]

[0079] in, M Sub-area radius , and Represent the average values ​​of the spectral identity information in the sub-region before and after deformation, respectively.

[0080] Among them, the "sub-area radius" is also called the "sub-area half height" or "sub-area half width". The sub-area is generally square or rectangular. The sub-area half height is the distance from the center point of the sub-area to the upper or lower boundary, and the half width is the distance to the left boundary or the middle boundary. When writing the formula, it is generally assumed to be a square for convenience, so the half height is usually equal to the half width, that is, the "radius".

[0081] S703, the position of the selected point in another camera image is obtained by calculating the deformation amount and obtaining the spatial position of the center point by binocular vision technology; the non-limiting implementation is as follows:

[0082] The calculated deformation amount can be used to obtain the position of the calculation point in the right camera image. The position of the calculation point in the left and right cameras can be used to obtain the spatial coordinates of the initial calculation point through binocular stereo vision technology, that is, the "sub-area radius" in the left figure is also called the "sub-area half height" or "sub-area half width". The sub-area is generally square or rectangular. The half height of the sub-area is the distance from the center point of the sub-area to the upper or lower boundary, and the half width is the distance to the left or right boundary. When writing the formula, it is generally defaulted to a square for convenience, so the half height is usually equal to the half width, that is, the "radius".

[0083] S703, the position of the selected point in another camera image is obtained by calculating the deformation amount and obtaining the spatial position of the center point by binocular vision technology; the non-limiting implementation is as follows:

[0084] The calculated deformation can be used to obtain the position of the calculated point in the right camera image. The position of the calculated point in the left and right cameras can be used to obtain the spatial coordinates of the initial calculated point through binocular stereo vision technology, that is, ( X 0, Y 0, Z 0).

[0085] Next, repeat S701 to S703 to obtain the spatial coordinates of the point during the deformation of the facial skin, which can be implemented as follows:

[0086] Using the above steps to determine the position of the initial calculation point in the left and right camera images at a certain moment when speaking, the binocular stereo vision technology can be used again to obtain the spatial coordinates of the initial calculation point when speaking, that is, ( X 1, Y 1, Z 1).

[0087] Next, it is explained how to obtain the maximum principal strain of the above calculation point through three-dimensional displacement calculation.

[0088] Among them, the three-dimensional displacement is calculated and the maximum principal strain of the center point is obtained, which can be implemented as follows:

[0089] The displacements in the horizontal and vertical directions in the three-dimensional displacement are calculated by the following formula (1): u’ and v’ ) is transformed into the local coordinate system and fitted using the least squares method;

[0090] (1)

[0091] in, u’ and v’ is the displacement in the local coordinate system, , , , , and are the fitting parameters to be determined.

[0092] Next, using the above fitting parameters, the strain components, principal strains and directions are obtained through the following formulas (2) and (3).

[0093] (2)

[0094] In the above formula (2), is the horizontal strain, is the vertical strain, is shear strain.

[0095] (3)

[0096] In formula (3), is the maximum principal strain (i.e., “principal strain” or “strain” in this application), Minimum principal strain, The main strain direction.

[0097] It should be pointed out that, according to the needs of implementation, the various components / steps described in the embodiments of the present application can be split into more components / steps, or two or more components / steps or partial operations of components / steps can be combined into new components / steps to achieve the purpose of the embodiments of the present application.

[0098] The above-mentioned method according to the embodiment of the present application can be implemented in hardware, firmware, or implemented as software or computer code that can be stored in a recording medium (such as a CD-ROM, RAM, floppy disk, hard disk or magneto-optical disk), or implemented as a computer code originally stored in a remote recording medium or a non-temporary machine-readable medium downloaded through a network and to be stored in a local recording medium, so that the method described herein can be stored in such software processing on a recording medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware (such as an application-specific integrated circuit (ASIC) or a field programmable gate array (FPGA)). It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component (e.g., a random access memory (RAM), a read-only memory (ROM), a flash memory, etc.) that can store or receive software or computer code, and when the software or computer code is accessed and executed by a computer, a processor, or hardware, the method described herein is implemented. In addition, when a general-purpose computer accesses the code for implementing the method shown herein, the execution of the code converts the general-purpose computer into a dedicated computer for executing the method shown herein.

[0099] Those of ordinary skill in the art will appreciate that the units and method steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the embodiments of the present application.

[0100] The above implementation methods are only used to illustrate the embodiments of the present application, and are not limitations on the embodiments of the present application. Ordinary technicians in the relevant technical field can make various changes and modifications without departing from the spirit and scope of the embodiments of the present application. Therefore, all equivalent technical solutions also belong to the scope of the embodiments of the present application. The scope of patent protection of the embodiments of the present application should be limited by the claims.

Claims

1. A detection device for facial skin, characterized in that: include: A first hyperspectral camera, used to obtain a video image and a hyperspectral image of a human face skin from a first perspective; A second hyperspectral camera, used for acquiring an image of human face skin and a hyperspectral image from a second viewing angle; A calculation module, which determines the spectral characteristics of each pixel point as its identity information through the hyperspectral images obtained from the first viewing angle and the second viewing angle, uses the identity information to perform matching search, and obtains the strain of the sub-area where each point is located through the image images collected from the first viewing angle and the second viewing angle through three-dimensional spectral image correlation calculation; The first hyperspectral camera and the second hyperspectral camera only collect spectral information of three bands: a blue light band of 461nm to 491nm, a cyan light band of 492nm to 540nm, and a green light band of 542nm to 580nm; The first hyperspectral camera and the second hyperspectral camera have the same resolution, both of which are 512×512 to 1024×1024; The three-dimensional spectral image correlation calculation is implemented based on multiple two-dimensional spectral image correlation calculations 2D-DIC, including the following steps: The spectral identity information is used to perform three-dimensional correlation calculation on each point of the image captured by the two hyperspectral cameras, and two sets of three-dimensional coordinates of each point of the facial skin under two different stretching states are obtained respectively; Subtracting the two sets of three-dimensional coordinates to obtain a three-dimensional displacement of each point; The maximum principal strain of each point is calculated based on the three-dimensional displacement and is used as the strain of the facial skin at that point.

2. A method for detecting facial skin, characterized in that: include: Using a first hyperspectral camera and a second hyperspectral camera to obtain a hyperspectral image and a video image of the human face skin in two different stretching states from a first viewing angle and a second viewing angle, respectively; Determine the spectral characteristics of each pixel point as its identity information through the hyperspectral images obtained from the first viewing angle and the second viewing angle, use the identity information to perform matching search, and obtain the strain of the sub-area where each point is located through the image images collected from the first viewing angle and the second viewing angle through three-dimensional spectral image correlation calculation; The first hyperspectral camera and the second hyperspectral camera only collect spectral information of three bands: a blue light band of 461nm to 491nm, a cyan light band of 492nm to 540nm, and a green light band of 542nm to 580nm; The first hyperspectral camera and the second hyperspectral camera have the same resolution, both of which are 512×512 to 1024×1024; The three-dimensional spectral image correlation calculation is implemented based on multiple two-dimensional spectral image correlation calculations 2D-DIC, including the following steps: The spectral identity information is used to perform three-dimensional spectral image correlation calculation on each point of the image captured by the two hyperspectral cameras, and two sets of three-dimensional coordinates of each point of the facial skin under two different stretching states are obtained respectively; The two sets of three-dimensional coordinates are subtracted to obtain the three-dimensional displacement of each point; the maximum principal strain of each point is calculated from the three-dimensional displacement as the strain of the human face skin at that point.