Face recognition analysis method based on loan review

Abstract

The invention discloses a face recognition analysis method based on loan review, and relates to the field of face recognition technology. The invention comprises a database, an image module, an analysis unit, a horizontal grating module, a vertical grating module, a display module and a backlight module, a port of the database establishes data communication with a port of the analysis unit, an output end of the image module is connected to an input end of the analysis unit, an output end of the analysis unit is respectively connected to the input ends of the horizontal grating module, the vertical grating module, the display module and the backlight module, and the horizontal grating module and the vertical grating module are both covered in the display area of ​​the display module, the invention makes the display direction of the display module always face the face of the lender through the joint action of the horizontal grating module and the vertical grating module, thereby avoiding leakage of the lender's sensitive data, and analyzing the lip shape similarity and the reading continuity can avoid pre-recording videos instead of real people for verification, thereby enhancing the strength of face verification.

Description

Technical Field

[0001] The present invention relates to the field of face recognition technology, and in particular to a face recognition analysis method based on loan review. Background Art

[0002] Loan applicant facial recognition is an identity verification technology used by lending institutions in the process of handling loan business. By comparing the borrower's facial features collected on site with information such as the ID card photo, it is verified whether the borrower is applying for the loan in person to prevent others from impersonating their identity.

[0003] The current facial recognition analysis method cannot effectively protect the lender's sensitive data, such as identity information and income. When the lender performs facial recognition operations, there is a risk of privacy leakage when checking information on the display screen. At the same time, the existing facial recognition technology only verifies whether the lender is a real person through steps such as turning the head and blinking. The verification method is too simple, and the verification is passed by playing the corresponding video in advance. Therefore, how to enhance the strength of facial verification while protecting the lender's sensitive data is a technical problem that technical personnel in this field need to solve. Summary of the invention

[0004] In view of the deficiencies of the prior art, the present invention provides a face recognition analysis method based on loan review, which solves the problems raised in the above-mentioned background technology.

[0005] To achieve the above purpose, the present invention is implemented through the following technical solutions: a face recognition analysis method based on loan review, including a database, an image module, an analysis unit, a horizontal grating module, a vertical grating module, a display module and a backlight module, the port of the database establishes data communication with the port of the analysis unit, the output end of the image module is connected to the input end of the analysis unit, the output end of the analysis unit is respectively connected to the input ends of the horizontal grating module, the vertical grating module, the display module and the backlight module, and the horizontal grating module and the vertical grating module are both covered in the display area of ​​the display module;

[0006] The database is used to store the word library and the lip shape pictures corresponding to the word library for reading by the analysis unit. The word library is THUOCL open Chinese word library. The database is also used to obtain the identity information and portrait of the loan personnel through the network. The image module obtains the image information of the area facing the display module and transmits it to the analysis unit. The analysis unit sends a backlight instruction to the backlight module. The backlight module emits light of different intensities through the display module to adjust the backlight brightness of the display module. The analysis unit sends loan review information to the display module, and the display module displays the loan review information for the loan personnel to view;

[0007] The horizontal grating module and the vertical grating module are both composed of a controller, a frame, a micro grating, a connecting rod, a first contact, a second contact and an SMA wire. The frame of the horizontal grating module is fixed to the front of the display module, and the frame of the vertical grating module is fixed to the front of the horizontal grating module. The outer dimensions of the frame are consistent with those of the display module. The first contact of the horizontal grating module and the first contact of the vertical grating module are both fixed at the corners of the frame.

[0008] The inner surface of the frame is rotatably connected to the 200 micro-gratings via a rotating shaft, one side of the 200 micro-gratings is rotatably connected to a connecting rod via a rotating shaft, the top of the outermost micro-grating is fixedly connected to the second contact, and the first contact is connected to the second contact via an SMA wire;

[0009] The output end of the analysis unit is respectively connected to the controller of the vertical grating module and the controller of the horizontal grating module, and the output end of the controller is respectively connected to the first contact and the second contact at the outermost side of the micro-grating. The controller short-circuits the SMA wire by energizing the first contact and the second contact. The SMA wire is made of memory alloy. The SMA wire will contract and pull the second contact to move. The movement of the second contact will drive the outermost micro-grating to rotate. The outermost micro-grating will drive all the micro-gratings to rotate in the direction close to the first contact through the connecting rod. Similarly, the contraction of the SMA wire on the other side will also pull all the micro-gratings to rotate in the opposite direction, so as to achieve angle adjustment of all the micro-gratings.

[0010] The following steps are involved:

[0011] Step 1: Face positioning, the image module obtains image information and transmits it to the analysis unit, and the analysis unit executes the face positioning program to obtain the face coordinates and the origin of the plane coordinate system using the pre-selected box;

[0012] Step 2: Adjust the light emission direction. The analysis unit executes a direction analysis program according to the face coordinates to obtain an adjustment direction θfa. The analysis unit executes a direction adjustment program. The analysis unit decomposes the adjustment direction θfa into a horizontal component θx and a vertical component θy. The horizontal grating module and the vertical grating module are adjusted together so that the display direction of the display module is always toward the face coordinates.

[0013] Step 3: Adjust the backlight. The analysis unit executes the backlight adjustment program to adjust the backlight brightness of the display module. The analysis unit obtains the identity information of the loan officer from the database and transmits it to the display module. The display module displays the identity information for the loan officer to check. After checking, the loan officer manually confirms that the identity information is correct and proceeds to the next step.

[0014] Step 4: Lip shape comparison. The analysis unit executes the lip shape comparison program to output the review result. The loan officer can re-enter the application request into the analysis unit, and the analysis unit jumps to step 1 and re-executes it.

[0015] Furthermore, the face positioning procedure specifically includes the following steps:

[0016] Step 101: The analysis unit establishes a plane coordinate system based on the image information. The analysis unit marks the intersection of the diagonals of the image information as the origin of the plane coordinate system. The analysis unit uses each pixel as a coordinate unit of the plane coordinate system.

[0017] Step 102: the analysis unit counts the number of occurrences of the RGB value of each pixel in the image information, and marks the RGB value with the largest number of occurrences as the background color. Removing the background color can prevent the non-face background of the image information from affecting the subsequent similarity calculation;

[0018] Step 103: the analysis unit sets a vertical reference line, the length of the vertical reference line is consistent with the vertical height of the image information, and the vertical reference line scans the image information pixel by pixel from left to right;

[0019] Step 104: The analysis unit presets a trigger threshold, moves the vertical reference line one pixel to the right, removes the background color in the pixels on both sides of the vertical reference line, and performs an axisymmetric comparison of the RGB values ​​of the pixels on the left and right sides of the vertical reference line to calculate the similarity. When the axisymmetric similarity of the pixels on the left and right sides of the vertical reference line is greater than or equal to 70% of the trigger threshold, the analysis unit marks the similar pixel areas on both sides of the vertical reference line as preliminary screening targets;

[0020] Step 105: The analysis unit obtains the portrait of the loan officer from the database, and the analysis unit expands and aligns the initial screening target according to the size of the portrait. The analysis unit uses the formula Calculate the similarity Fa between the primary screening target and the portrait, where m is the pixel length of the portrait, n is the pixel width of the portrait, I is the pixel selected from the primary screening target, K is the pixel selected from the portrait, i is the horizontal coordinate of the pixels selected from the primary screening target and the portrait, and j is the vertical coordinate of the pixels selected from the primary screening target and the portrait;

[0021] Step 106: The analysis unit presets a similarity threshold, and judges based on the similarity Fa. If the similarity Fa is less than or equal to the similarity threshold of 90%, it means that the initial screening target and the portrait do not match. The analysis unit repeats steps 104 and 105. If the similarity is greater than the similarity threshold of 90%, the analysis unit executes the next step 107.

[0022] Step 107: The analysis unit records the initial screening target as the target face, and sets a pre-selection box to select the target face. The four borders of the pre-selection box are tangent to the top, bottom, leftmost and rightmost sides of the target face respectively. The analysis unit marks the intersection of the diagonals of the pre-selection box as the face coordinates.

[0023] Furthermore, the analysis unit repeatedly executes the direction analysis program to obtain the adjustment direction θfa of each frame of image information, and the analysis unit repeatedly executes the direction adjustment program according to the continuously changing adjustment direction θfa to match the continuously changing adjustment direction θfa. The direction analysis program specifically includes the following steps:

[0024] Step 201: the analysis unit calculates the distance d1 between the pre-selected box in the image information and the display module according to the formula d1=W×f / w, where f is the focal length of the image module, W is the average height of a face, w is the pixel height of the pre-selected box in the image information, and the focal length f is a constant;

[0025] Step 202: The analysis unit calculates the distance d2 between the face coordinates in the image information and the origin. The analysis unit calculates the distance d2 between the face coordinates in the image information and the origin according to the formula Decompose d2 into a first component d2x and a second component d2y;

[0026] Step 203: Analyze the unit according to the formula Calculate the adjustment direction θfa of the face coordinates.

[0027] Furthermore, the direction adjustment procedure specifically includes the following steps:

[0028] Step 204: the analysis unit decomposes the adjustment direction θfa into a horizontal component θx and a vertical component θy, the analysis unit transmits the horizontal component θx to the controller of the horizontal grating module, and the analysis unit transmits the vertical component θy to the controller of the vertical grating module;

[0029] Step 205: The controller of the horizontal grating module energizes the first contact and the second contact on the side close to the horizontal component θx, and the corresponding SMA wire is short-circuited, heated, and contracted to pull all the micro-gratings in the horizontal grating module to rotate in the direction close to the horizontal component θx until the micro-gratings are parallel to the horizontal component θx, and the controller of the horizontal grating module deenergizes the first contact and the second contact;

[0030] Step 206: The controller of the vertical grating module energizes the first contact and the second contact close to the vertical component θy, and the corresponding SMA wire is short-circuited, heated, and contracted to pull all the micro-gratings in the vertical grating module to rotate in the direction close to the vertical component θy until the micro-gratings are parallel to the vertical component θy, and the controller of the vertical grating module deenergizes the first contact and the second contact;

[0031] Step 207: The micro-gratings in the horizontal grating module and the micro-gratings in the vertical grating module act together on the display module to constrain the display direction of the display module to point to the face coordinates.

[0032] Furthermore, the backlight adjustment procedure specifically includes the following steps:

[0033] Step 301: the analysis unit obtains image information from the image module, and calculates the total brightness value e1 of all pixels in the pre-selected frame;

[0034] Step 302: the analysis unit sends a backlight instruction to the backlight module, and the backlight module outputs the maximum power to increase the backlight brightness of the display module to the highest level;

[0035] Step 303: the analysis unit obtains image information from the image module, and calculates the total brightness value e2 of all pixels in the pre-selected frame;

[0036] Step 304: The analysis unit counts the number of pixels in the pre-selected box as Q, and the analysis unit calculates the number of pixels in the pre-selected box according to the formula The adjustment power PL of the backlight module is calculated, and the analysis unit sends the adjustment power PL to the backlight module. The backlight module adjusts the output power according to the received adjustment power PL, wherein T is a proportional coefficient, and the specific value of the proportional coefficient T is determined according to the actual brightness of the ambient light.

[0037] Furthermore, the lip shape comparison procedure specifically comprises the following steps:

[0038] Step 401: the analysis unit randomly reads 10 phrases and the lip-shaped pictures corresponding to the phrases from the word library of the database, and transmits the phrases to the display module for the loan officer to view;

[0039] Step 402: The loan officer reads aloud the phrase displayed by the display module facing the image module. The image module transmits the picture information read aloud by the loan officer to the analysis unit. The analysis unit executes the face positioning program to obtain a pre-selection box. The analysis unit cuts out and extracts the bottom one-third area of ​​the pre-selection box of all image information and marks it as a reading image. Cutting can reduce the amount of calculation for image comparison and improve the processing speed of the analysis unit.

[0040] Step 403: Lip shape similarity analysis. The analysis unit presets a matching threshold. The display module displays a phrase each time. When the loan officer reads the phrase, the analysis unit compares the reading image with the lip shape image. The analysis unit uses the formula Calculate the similarity Kx between the reading image and the lip picture, where u is the pixel length of the lip picture, g is the pixel width of the lip picture, G is the pixel selected from the reading image, H is the pixel selected from the lip picture, a is the horizontal coordinate of the pixel selected from the lip picture and the reading image, and b is the vertical coordinate of the pixel selected from the lip picture and the reading image;

[0041] When the similarity Kx is less than or equal to the matching threshold of 90%, it means that the reading image and the lip shape picture do not match, and the process goes to step 405. When the similarity Kx is greater than the matching threshold of 90%, it means that the reading image and the lip shape picture match, and the process goes to step 404.

[0042] Step 404: Analyze the continuity of reading aloud. The analysis unit presets an interruption threshold. The analysis unit compares all the reading aloud images pixel by pixel. The time interval between the previous and next frames is Δt, and the number of frames is Q. According to the formula Calculate the continuous index Sld of the read image, where V pg The speed change V is obtained by subtracting the pixels of the previous and next frames and dividing the pixel displacement by the time interval Δt. pg When the continuity index Sld is greater than the interruption threshold 0.9, jump to step 406. On the contrary, when the continuity index Sld is less than or equal to the interruption threshold 0.9, go to the next step 405. The closer the continuity index Sld is to 1, the smoother the previous and next frames of the read image are and the stronger the continuity is. The lower the continuity index Sld is, the worse the continuity is and the stronger the tearing feeling of the read image is, indicating that it is not a real person reading aloud.

[0043] Step 405: the analysis unit outputs a verification result indicating that the verification has failed, and the lip shape comparison procedure is terminated;

[0044] Step 406: The audit result output by the analysis unit is verification success.

[0045] The present invention has the following beneficial effects:

[0046] 1. Through the joint action of the horizontal grating module and the vertical grating module, the display direction of the display module is always toward the face of the lender, thus avoiding the leakage of the lender's sensitive data.

[0047] 2. By asking the lender to read randomly read phrases, the randomness of the verification process can be enhanced. Analyzing the similarity of lip movements and the continuity of reading can avoid pre-recording videos to replace real people for verification, thereby enhancing the strength of facial verification.

[0048] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0050] Figure 1 It is a system block diagram of the face recognition analysis method based on loan review of the present invention;

[0051] Figure 2 It is a schematic structural diagram of the horizontal grating module and the vertical grating module of the present invention.

[0052] In the accompanying drawings, the components represented by the reference numerals are as follows:

[0053] In the figure: 1-frame, 2-micro grating, 3-connecting rod, 4-first contact, 5-second contact, 6-SMA wire. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0055] See also Figure 1 - Figure 2 , the present invention provides a technical solution: a face recognition analysis method based on loan review, such as Figure 1 As shown, it includes a database, an image module, an analysis unit, a horizontal grating module, a vertical grating module, a display module and a backlight module. The port of the database establishes data communication with the port of the analysis unit. The output end of the image module is connected to the input end of the analysis unit. The output end of the analysis unit is respectively connected to the input ends of the horizontal grating module, the vertical grating module, the display module and the backlight module. The horizontal grating module and the vertical grating module are both covered in the display area of ​​the display module.

[0056] The database is used to store the vocabulary and the lip-shaped images corresponding to the vocabulary for reading by the analysis unit. The vocabulary is the THUOCL open Chinese vocabulary. The database is also used to obtain the identity information and portrait of the loan personnel through the Internet. The image module obtains the image information of the area facing the display module and transmits it to the analysis unit. The analysis unit sends a backlight instruction to the backlight module. The backlight module emits light of different intensities through the display module to adjust the backlight brightness of the display module. The analysis unit sends the loan review information to the display module, and the display module displays the loan review information for the loan personnel to view.

[0057] like Figure 2 As shown, the horizontal grating module and the vertical grating module are both composed of a controller, a frame 1, a micro grating 2, a connecting rod 3, a first contact 4, a second contact 5 and an SMA wire 6. The frame 1 of the horizontal grating module is fixed to the front of the display module, and the frame 1 of the vertical grating module is fixed to the front of the horizontal grating module. The outer dimensions of the frame 1 are consistent with those of the display module. The first contact 4 of the horizontal grating module and the first contact 4 of the vertical grating module are both fixed at the corners of the frame 1.

[0058] The inner surface of the frame 1 is rotatably connected to 200 micro-gratings 2 via a rotating shaft, one side of the 200 micro-gratings 2 is rotatably connected to a connecting rod 3 via a rotating shaft, the top of the outermost micro-grating 2 is fixedly connected to the second contact 5, and the first contact 4 and the second contact 5 are connected via an SMA wire 6;

[0059] The output end of the analysis unit is connected to the controller of the vertical grating module and the controller of the horizontal grating module respectively. The output end of the controller is connected to the first contact 4 and the second contact 5 at the outermost side of the micro-grating 2 respectively. The controller makes the SMA wire 6 short-circuited by energizing the first contact 4 and the second contact 5. The material of the SMA wire is a memory alloy. Based on the physical properties of the SMA wire 6, it shrinks when heated and extends when cooled. When the SMA wire 6 is short-circuited, the temperature rises and shrinks. The first contact 4 is fixed, and the SMA wire 6 will pull the second contact 5 to move through the force generated by the contraction. The movement of the second contact 5 will drive the outermost micro-grating 2 to rotate. The outermost micro-grating 2 will drive all the micro-gratings 2 to rotate in the direction close to the first contact 4 through the connecting rod 3. Similarly, the contraction of the SMA wire 6 on the other side will also pull all the micro-gratings 2 to rotate in the opposite direction, so as to achieve the angle adjustment of all the micro-gratings 2. The adjustment angle of the micro-grating 2 is ±7.5°.

[0060] The method specifically comprises the following steps:

[0061] Step 1: Face positioning: the image module obtains image information and transmits it to the analysis unit. The analysis unit executes the face positioning program and uses the pre-selected box to obtain the face coordinates and the origin of the plane coordinate system.

[0062] Step 2: Adjust the light emission direction. The analysis unit executes a direction analysis program according to the face coordinates to obtain an adjustment direction θfa. The analysis unit executes a direction adjustment program. The analysis unit decomposes the adjustment direction θfa into a horizontal component θx and a vertical component θy. The horizontal grating module and the vertical grating module are adjusted together so that the display direction of the display module is always toward the face coordinates.

[0063] Step 3: Adjust the backlight. The analysis unit executes the backlight adjustment program to adjust the backlight brightness of the display module. The analysis unit obtains the identity information of the loan officer from the database and transmits it to the display module. The display module displays the identity information for the loan officer to check. After checking, the loan officer manually confirms that the identity information is correct and proceeds to the next step.

[0064] Step 4: Lip shape comparison. The analysis unit executes the lip shape comparison program to output the review result. The loan officer can re-enter the application request into the analysis unit, and the analysis unit jumps to step 1 and re-executes it.

[0065] The face positioning procedure specifically includes the following steps:

[0066] Step 101: The analysis unit establishes a plane coordinate system based on the image information. The analysis unit marks the intersection of the diagonals of the image information as the origin of the plane coordinate system. The analysis unit uses each pixel as a coordinate unit of the plane coordinate system.

[0067] Step 102: the analysis unit counts the number of occurrences of the RGB value of each pixel in the image information, and marks the RGB value with the largest number of occurrences as the background color. Removing the background color can prevent the non-face background of the image information from affecting the subsequent similarity calculation;

[0068] Step 103: the analysis unit sets a vertical reference line, the length of the vertical reference line is consistent with the vertical height of the image information, and the vertical reference line scans the image information pixel by pixel from left to right;

[0069] Step 104: The analysis unit presets a trigger threshold, moves the vertical reference line one pixel to the right, removes the background color in the pixels on both sides of the vertical reference line, and performs an axisymmetric comparison of the RGB values ​​of the pixels on the left and right sides of the vertical reference line to calculate the similarity. When the axisymmetric similarity of the pixels on the left and right sides of the vertical reference line is greater than or equal to 70% of the trigger threshold, the analysis unit marks the similar pixel areas on both sides of the vertical reference line as preliminary screening targets;

[0070] Most people's left and right face symmetry similarity is between 70% and 90%, so the analysis unit uses 70% as the trigger threshold for the symmetry similarity of the left and right pixel axes of the vertical reference line. When the similarity is greater than or equal to the trigger threshold, it is marked as a preliminary screening target for further verification;

[0071] Step 105: The analysis unit obtains the portrait of the loan officer from the database, expands and aligns the initial screening target according to the size of the portrait, and calculates the loan officer's portrait according to the formula Calculate the similarity Fa between the primary screening target and the portrait, where m is the pixel length of the portrait, n is the pixel width of the portrait, I is the pixel selected from the primary screening target, K is the pixel selected from the portrait, i is the horizontal coordinate of the pixels selected from the primary screening target and the portrait, and j is the vertical coordinate of the pixels selected from the primary screening target and the portrait;

[0072] Step 106: The analysis unit presets a similarity threshold, and judges based on the similarity Fa. If the similarity Fa is less than or equal to the similarity threshold of 90%, it means that the initial screening target and the portrait do not match. The analysis unit repeats steps 104 and 105. If the similarity is greater than the similarity threshold of 90%, the analysis unit executes the next step 107.

[0073] Step 107: The analysis unit records the initial screening target as the target face, and sets a pre-selection box to select the target face. The four borders of the pre-selection box are tangent to the top, bottom, leftmost and rightmost sides of the target face respectively. The analysis unit marks the intersection of the diagonals of the pre-selection box as the face coordinates.

[0074] The analysis unit repeatedly executes the direction analysis program to obtain the adjustment direction θfa of each frame of image information. The analysis unit repeatedly executes the direction adjustment program according to the continuously changing adjustment direction θfa to match the continuously changing adjustment direction θfa. The direction analysis program specifically includes the following steps:

[0075] Step 201: The analysis unit calculates the distance d1 between the pre-selected box in the image information and the display module according to the formula d1=W×f / w, where f is the focal length of the image module, W is the average height of a human face, which is 19.5 cm, and w is the pixel height of the pre-selected box in the image information. The focal length f is a constant value, and the image information transmitted from the image module to the analysis unit includes a record of the focal length f;

[0076] Step 202: The analysis unit calculates the distance d2 between the face coordinates in the image information and the origin. The analysis unit calculates the distance d2 between the face coordinates in the image information and the origin according to the formula Decompose d2 into a first component d2x and a second component d2y;

[0077] Step 203: Analyze the unit according to the formula Calculate the adjustment direction θfa of the face coordinates.

[0078] The direction adjustment procedure specifically includes the following steps:

[0079] Step 204: the analysis unit decomposes the adjustment direction θfa into a horizontal component θx and a vertical component θy, the analysis unit transmits the horizontal component θx to the controller of the horizontal grating module, and the analysis unit transmits the vertical component θy to the controller of the vertical grating module;

[0080] Step 205: the controller of the horizontal grating module energizes the first contact 4 and the second contact 5 close to the horizontal component θx, and the corresponding SMA wire 6 is short-circuited, heated, and contracted to pull all the micro-gratings 2 in the horizontal grating module to rotate in the direction close to the horizontal component θx until the micro-gratings 2 are parallel to the horizontal component θx, and the controller of the horizontal grating module deenergizes the first contact 4 and the second contact 5;

[0081] Step 206: the controller of the vertical grating module energizes the first contact 4 and the second contact 5 close to the vertical component θy, and the corresponding SMA wire 6 is short-circuited, heated up and contracted, pulling all the micro-gratings 2 in the vertical grating module to rotate in the direction close to the vertical component θy until the micro-gratings 2 are parallel to the vertical component θy, and the controller of the vertical grating module deenergizes the first contact 4 and the second contact 5;

[0082] Step 207: The micro-grating 2 in the horizontal grating module and the micro-grating 2 in the vertical grating module work together on the display module to constrain the display direction of the display module to point to the face coordinates. Except for the loan personnel located in the direction of the face coordinates, the display content of the display module cannot be viewed in other directions, thereby realizing sensitive data protection of identity information.

[0083] The backlight adjustment procedure specifically includes the following steps:

[0084] Step 301: the analysis unit obtains image information from the image module, and calculates the total brightness value e1 of all pixels in the pre-selected frame;

[0085] Step 302: the analysis unit sends a backlight instruction to the backlight module, and the backlight module outputs the maximum power to increase the backlight brightness of the display module to the highest level;

[0086] Step 303: the analysis unit obtains image information from the image module, and calculates the total brightness value e2 of all pixels in the pre-selected frame;

[0087] Step 304: The analysis unit counts the number of pixels in the pre-selected box as Q, and the analysis unit calculates the number of pixels in the pre-selected box according to the formula The adjustment power PL of the backlight module is calculated, and the analysis unit sends the adjustment power PL to the backlight module. The backlight module adjusts the output power according to the received adjustment power PL, wherein T is the proportional coefficient. The proportional coefficient T determines the specific value according to the actual brightness of the ambient light, so that the backlight brightness of the display module is adjusted to an appropriate value. The micro-grating 2 will block part of the light emitted by the backlight module through the display module. Therefore, the backlight module is required to synchronously adjust the output power of the backlight module after the direction adjustment program is executed to avoid the display brightness of the display module being affected by the micro-grating 2 and the display content being unable to be seen clearly, and also to avoid the excessive output power of the backlight module causing irritation to the eyes of the loan personnel.

[0088] The lip shape comparison procedure specifically includes the following steps:

[0089] Step 401: the analysis unit randomly reads 10 phrases and the lip-shaped pictures corresponding to the phrases from the word library of the database, and transmits the phrases to the display module for the loan officer to view;

[0090] Step 402: The loan officer faces the image module and reads aloud the phrase displayed by the display module. The image module transmits the picture information read aloud by the loan officer to the analysis unit. The analysis unit executes the face positioning program to obtain a pre-selected box. The analysis unit cuts out and extracts the bottom one-third area of ​​the pre-selected box of all image information and marks it as a read-aloud image. Cutting can reduce the amount of calculation for image comparison and improve the processing speed of the analysis unit.

[0091] Step 403: Lip shape similarity analysis. The analysis unit presets a matching threshold. The display module displays a phrase each time. When the loan officer reads the phrase, the analysis unit compares the reading image with the lip shape image. The analysis unit uses the formula Calculate the similarity Kx between the reading image and the lip picture, where u is the pixel length of the lip picture, g is the pixel width of the lip picture, G is the pixel selected from the reading image, H is the pixel selected from the lip picture, a is the horizontal coordinate of the pixel selected from the lip picture and the reading image, and b is the vertical coordinate of the pixel selected from the lip picture and the reading image;

[0092] When the similarity Kx is less than or equal to the matching threshold of 90%, it means that the reading image and the lip shape picture do not match, and the process goes to step 405. When the similarity Kx is greater than the matching threshold of 90%, it means that the reading image and the lip shape picture match, and the process goes to step 404.

[0093] Step 404: Analyze the continuity of reading aloud. The analysis unit presets an interruption threshold. The analysis unit compares all the reading aloud images pixel by pixel. The time interval between the previous and next frames is Δt, and the number of frames is Q. According to the formula Calculate the continuous index Sld of the read image, where V pg The speed change V is obtained by subtracting the pixels of the previous and next frames and dividing the pixel displacement by the time interval Δt. pg When the continuity index Sld is greater than the interruption threshold 0.9, jump to step 406. On the contrary, when the continuity index Sld is less than or equal to the interruption threshold 0.9, go to the next step 405. The closer the continuity index Sld is to 1, the smoother the previous and next frames of the read image are and the stronger the continuity is. The lower the continuity index Sld is, the worse the continuity is and the stronger the tearing feeling of the read image is, indicating that it is not a real person reading aloud.

[0094] Step 405: the analysis unit outputs a verification result indicating that the verification has failed, and the lip shape comparison procedure is terminated;

[0095] Step 406: The audit result output by the analysis unit is verification success.

[0096] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A face recognition analysis method based on loan review is implemented based on a database, an image module, an analysis unit, a horizontal grating module, a vertical grating module, a display module and a backlight module, and is characterized in that: The following steps are involved: Step 1: The image module acquires image information and transmits it to the analysis unit, and the analysis unit executes the face positioning program to obtain the face coordinates and the origin of the plane coordinate system using the pre-selected box; Step 2: the analysis unit executes a direction analysis program according to the face coordinates to obtain an adjustment direction θfa, and the analysis unit executes a direction adjustment program so that the display direction of the display module is always toward the face coordinates; Step 3: The analysis unit executes the backlight adjustment program to adjust the backlight brightness of the display module. The analysis unit obtains the identity information of the loan officer from the database and transmits it to the display module. The display module displays the identity information for the loan officer to check and verify. After checking, the loan officer manually confirms that the identity information is correct and proceeds to the next step. Step 4: The analysis unit executes the lip shape comparison program to output the audit result; The port of the database establishes data communication with the port of the analysis unit, the output end of the image module is connected to the input end of the analysis unit, the output end of the analysis unit is respectively connected to the input ends of the horizontal grating module, the vertical grating module, the display module and the backlight module, and the horizontal grating module and the vertical grating module are both covered in the display area of ​​the display module; The database is used to store the word library and the lip shape pictures corresponding to the word library for reading by the analysis unit. The database is also used to obtain the identity information and portrait of the loan personnel through the network. The image module obtains the image information of the area facing the display module and transmits it to the analysis unit. The analysis unit sends a backlight instruction to the backlight module. The backlight module emits light of different intensities through the display module to adjust the backlight brightness of the display module. The horizontal grating module and the vertical grating module are both composed of a controller, a frame (1), a micro grating (2), a connecting rod (3), a first contact (4), a second contact (5) and an SMA wire (6); the frame (1) of the horizontal grating module is fixed to the front of the display module, and the frame (1) of the vertical grating module is fixed to the front of the horizontal grating module; the frame (1) has an outer size consistent with that of the display module; and the first contact (4) of the horizontal grating module and the first contact (4) of the vertical grating module are both fixed to the corners of the frame (1); The inner surface of the frame (1) is rotatably connected to a plurality of micro-gratings (2), one side of each of the plurality of micro-gratings (2) is rotatably connected to a connecting rod (3), the top end of the outermost micro-grating (2) is fixedly connected to a second contact (5), and the first contact (4) and the second contact (5) are connected via an SMA wire (6); The output end of the analysis unit is connected to the controller of the vertical grating module and the controller of the horizontal grating module respectively, and the output end of the controller is connected to the first contact (4) and the second contact (5) respectively; The face positioning procedure specifically includes the following steps: Step 101: The analysis unit establishes a plane coordinate system based on the image information. The analysis unit marks the intersection of the diagonals of the image information as the origin of the plane coordinate system. The analysis unit uses each pixel as a coordinate unit of the plane coordinate system. Step 102: the analysis unit counts the number of occurrences of the RGB value of each pixel in the image information, and marks the RGB value with the largest number of occurrences as the background color; Step 103: the analysis unit sets a vertical reference line, the length of the vertical reference line is consistent with the vertical height of the image information, and the vertical reference line scans the image information pixel by pixel from left to right; Step 104: The analysis unit presets a trigger threshold, moves the vertical reference line one pixel to the right, removes the background color in the pixels on both sides of the vertical reference line, and performs an axisymmetric comparison of the RGB values ​​of the pixels on the left and right sides of the vertical reference line to calculate the similarity. When the axisymmetric similarity of the pixels on the left and right sides of the vertical reference line is greater than or equal to the trigger threshold, the analysis unit marks the similar pixel areas on both sides of the vertical reference line as preliminary screening targets; Step 105: The analysis unit obtains the portrait of the loan officer from the database, expands and aligns the initial screening target according to the size of the portrait, and uses the formula Calculate the similarity Fa between the primary screening target and the portrait, where m is the pixel length of the portrait, n is the pixel width of the portrait, I is the pixel selected from the primary screening target, K is the pixel selected from the portrait, i is the horizontal coordinate of the pixels selected from the primary screening target and the portrait, and j is the vertical coordinate of the pixels selected from the primary screening target and the portrait; Step 106: The analysis unit presets a similarity threshold, and judges based on the similarity Fa. If the similarity Fa is less than or equal to the similarity threshold, it means that the initial screening target and the portrait do not match. The analysis unit repeats steps 104 and 105. If the similarity is greater than the similarity threshold, the analysis unit executes the next step 107. Step 107: the analysis unit records the initial screening target as the target face, and sets a pre-selection box to select the target face. The four frames of the pre-selection box are tangent to the top, bottom, left and right of the target face respectively, and the analysis unit marks the intersection of the diagonal lines of the pre-selection box as the face coordinates; The analysis unit repeatedly executes the direction analysis procedure to obtain the adjustment direction θfa of each frame of image information. The direction analysis procedure specifically includes the following steps: Step 201: Calculate the distance d1 between the pre-selected box in the image information and the display module according to the formula d1=W×f / w, where f is the focal length of the image module, W is the average height of a human face, and w is the pixel height of the pre-selected box in the image information; Step 202: The analysis unit calculates the distance d2 between the face coordinates in the image information and the origin. The analysis unit calculates the distance d2 between the face coordinates in the image information and the origin according to the formula Decompose d2 into a first component d2x and a second component d2y; Step 203: Analyze the unit according to the formula Calculate the adjustment direction θfa of the face coordinates; The direction adjustment procedure specifically includes the following steps: Step 204: decompose the adjustment direction θfa into a horizontal component θx and a vertical component θy, the analysis unit transmits the horizontal component θx to the controller of the horizontal grating module, and the analysis unit transmits the vertical component θy to the controller of the vertical grating module; Step 205: the controller of the horizontal grating module energizes the first contact (4) and the second contact (5) on the side close to the horizontal component θx, and the corresponding SMA wire (6) is short-circuited, heated up, and contracted, pulling all the micro-gratings (2) in the horizontal grating module to rotate in a direction close to the horizontal component θx until the micro-gratings (2) are parallel to the horizontal component θx, and the controller of the horizontal grating module deenergizes the first contact (4) and the second contact (5); Step 206: the controller of the vertical grating module energizes the first contact (4) and the second contact (5) close to the vertical component θy, and the corresponding SMA wire (6) is short-circuited, heated up, and contracted, pulling all the micro-gratings (2) in the vertical grating module to rotate in a direction close to the vertical component θy until the micro-gratings (2) are parallel to the vertical component θy, and the controller of the vertical grating module deenergizes the first contact (4) and the second contact (5); Step 207: the micro-grating (2) in the horizontal grating module and the micro-grating (2) in the vertical grating module act together on the display module to constrain the display direction of the display module to point to the face coordinates; The backlight adjustment procedure specifically includes the following steps: Step 301: the analysis unit obtains image information from the image module, and calculates the total brightness value e1 of all pixels in the pre-selected frame; Step 302: the analysis unit sends a backlight instruction to the backlight module, and the backlight module outputs the maximum power to increase the backlight brightness of the display module to the highest level; Step 303: the analysis unit obtains image information from the image module, and calculates the total brightness value e2 of all pixels in the pre-selected frame; Step 304: The analysis unit counts the number of pixels in the pre-selected box as Q, and the analysis unit calculates the number of pixels in the pre-selected box according to the formula Calculate the adjustment power PL of the backlight module, the analysis unit sends the adjustment power PL to the backlight module, and the backlight module adjusts the output power according to the received adjustment power PL, where T is the proportionality coefficient; The lip shape comparison procedure specifically includes the following steps: Step 401: the analysis unit randomly reads a number of phrases and lip-images corresponding to the phrases from the word library of the database, and transmits the phrases to the display module for the loan officer to view; Step 402: The loan officer faces the image module and reads aloud the phrase displayed by the display module. The image module transmits the picture information read aloud by the loan officer to the analysis unit, executes the face positioning program to obtain the pre-selection box, and the analysis unit cuts out and extracts the bottom one-third area of ​​the pre-selection box of all image information and marks it as the reading image; Step 403: Lip shape similarity analysis. The analysis unit presets a matching threshold. The display module displays a phrase each time. When the loan officer reads the phrase, the analysis unit compares the reading image with the lip shape image. The analysis unit uses the formula Calculate the similarity Kx between the reading image and the lip picture, where u is the pixel length of the lip picture, g is the pixel width of the lip picture, G is the pixel selected from the reading image, H is the pixel selected from the lip picture, a is the horizontal coordinate of the pixel selected from the lip picture and the reading image, and b is the vertical coordinate of the pixel selected from the lip picture and the reading image; When the similarity Kx is less than or equal to the matching threshold, it means that the reading image and the lip shape picture do not match, and the process goes to step 405. When the similarity Kx is greater than the matching threshold, it means that the reading image and the lip shape picture match, and the process goes to step 404. Step 404: Analyze the continuity of reading aloud. The analysis unit presets an interruption threshold. The analysis unit compares all the reading aloud images pixel by pixel. The time interval between the previous and next frames is Δt, and the number of frames is Q. According to the formula Calculate the continuous index Sld of the read image, where V pg The speed change V is obtained by subtracting the pixels of the previous and next frames and dividing the pixel displacement by the time interval Δt. pg , when the continuity index Sld is greater than the interruption threshold, jump to step 406, otherwise, when the continuity index Sld is less than or equal to the interruption threshold, proceed to the next step 405; Step 405: the analysis unit outputs a verification result indicating that the verification has failed, and the lip shape comparison procedure is terminated; Step 406: The audit result output by the analysis unit is verification success.

Images