A household registration image judgment method, system and readable storage medium based on image recognition

By de-illuminating and turning posture into positive processing of face images collected by the self-service certificate application machine, problems such as posture differences and lighting imbalance are solved, the accuracy of identity recognition is improved, and user identity authentication is realized in diverse scenarios.

CN113378779BActive Publication Date: 2024-12-06GUANGZHOU PRESTIGE TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202110736867.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-06-30
Publication Date
2024-12-06
Estimated Expiration
2041-06-30

AI Technical Summary

Technical Problem

When capturing face images, existing self-service certificate application machines have low identity recognition accuracy due to different postures, unbalanced lighting, blur, noise and low resolution problems, especially in diverse scenarios, which are difficult to effectively verify user identity.

Method used

The collected face images are deilluminated and posture-oriented through the self-service certificate application machine. The neural network model is used to extract and adjust the image features to achieve the clarity of the image and restore the real face data, and feature comparison is performed to improve the recognition accuracy sex.

Benefits of technology

It effectively improves the accuracy of identity recognition and judgment in diverse scenarios, so that pre-stored face images can be matched more accurately after image processing, and enhances the reliability of user identity verification.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113378779B_ABST
    Figure CN113378779B_ABST
Patent Text Reader

Abstract

The present invention discloses a household registration image judgment method, system and readable storage medium based on image recognition, wherein the method includes: obtaining facial image information; performing feature extraction on the facial image information to obtain a feature image; processing the feature image to obtain a first facial image; obtaining pre-stored user facial image information; performing feature comparison between the user facial image information and the first facial image to obtain a feature difference; if the feature difference is less than a preset difference threshold, the recognition is passed. The image of the user is collected by a self-service card processing machine, and the image is de-illuminated and posture-corrected to make the image clearer, and the user's real facial data can be restored. After the image is processed, feature comparison and recognition can be performed based on the pre-stored facial image, which improves the accuracy of user identity recognition and judgment in the diversity of actual scenes.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of image recognition and processing, and more specifically, relates to a household registration image judgment method, system and readable storage medium based on image recognition. Background Art

[0002] With the continuous development and progress of society, artificial intelligence technology has become an important driving force for the new round of economic development and the core competitiveness of social innovation. As human society develops towards informatization and intelligence, the use of human biological characteristics such as face, iris, palm print, fingerprint, etc. for intelligent identity recognition has become an important way of personal identity authentication. The identity authentication technology based on biometrics mainly explores identity features that are identity-invariant for a certain individual and identity-distinguishing for other individuals, and uses the identity features to efficiently and quickly identify and authenticate the identities of different individuals. The currently widely used biometric recognition methods include: iris recognition, fingerprint recognition, palm print recognition, voice recognition, face recognition, and gait recognition.

[0003] Current self-service ID machines often need to collect user facial images, and then make judgments and recognitions in order to issue certificates. However, in current self-service ID machines, due to the diversity of actual scenes, under unrestricted conditions, the captured facial images often face problems such as posture differences, uneven lighting, blur, noise, low resolution, etc. There is also the problem of capturing side faces. Therefore, the existing technology has defects and is in urgent need of improvement. Summary of the invention

[0004] The purpose of the embodiments of the present application is to provide a household registration image judgment method, system and readable storage medium based on image recognition. The user's image is collected by a self-service card processing machine, and the image is de-illuminated and posture-corrected to make the image clearer. The user's real facial data can also be restored, thereby improving the accuracy of user identity recognition and judgment.

[0005] A first aspect of the present invention provides a household registration image determination method based on image recognition, comprising:

[0006] Obtaining facial image information;

[0007] Performing feature extraction on the facial image information to obtain a feature image;

[0008] Processing the feature image to obtain a first face image;

[0009] Obtain pre-stored user face image information;

[0010] Performing feature comparison between the user face image information and the first face image to obtain a feature difference;

[0011] If the feature difference is less than the preset difference threshold, the recognition is successful.

[0012] In this solution, the feature image is processed to obtain the first face image, specifically:

[0013] De-illumination the feature image to obtain a second face image;

[0014] Performing posture correction processing on the second facial image to obtain a third facial image;

[0015] The third facial image is processed according to a preset rule to obtain a first facial image.

[0016] In this solution, the feature image is de-illuminated to obtain the second face image, specifically:

[0017] Classifying the characteristic images to obtain illumination low-frequency image information and high-frequency illumination image information;

[0018] Performing de-illumination processing on the illuminated low-frequency image information to obtain fourth face image information;

[0019] The fourth facial image information is fused with the high-frequency illumination image information to obtain a second facial image.

[0020] In this solution, the illumination low-frequency image information is de-illuminated to obtain the fourth face image information, specifically:

[0021] Obtain lighting information and noise information;

[0022] The fourth face image information is obtained by calculating the illumination information and the noise information according to the following formula:

[0023]

[0024] Among them, S is the fourth face image information after de-illumination, Y is the inherent illumination low-frequency image information of the face, G is the illumination information, n is the noise information, and x is the coordinate value of each point in the image.

[0025] In this solution, the second face image is subjected to posture correction processing to obtain a third face image, specifically:

[0026] Get posture information;

[0027] The second facial image is input into the first neural network model according to the posture information, and a third facial image is output.

[0028] In this solution, the acquisition of posture information is specifically as follows:

[0029] The second face image is sent to a second neural network model and the posture information is output.

[0030] A second aspect of the present invention provides a household registration image determination system based on image recognition, the system comprising: a memory and a processor, the memory comprising a program of a household registration image determination method based on image recognition, the program of the household registration image determination method based on image recognition being executed by the processor to implement the following steps:

[0031] Obtaining facial image information;

[0032] Performing feature extraction on the facial image information to obtain a feature image;

[0033] Processing the feature image to obtain a first face image;

[0034] Obtain pre-stored user face image information;

[0035] Performing feature comparison between the user face image information and the first face image to obtain a feature difference;

[0036] If the feature difference is less than the preset difference threshold, the recognition is successful.

[0037] In this solution, the feature image is processed to obtain the first face image, specifically:

[0038] De-illumination the feature image to obtain a second face image;

[0039] Performing posture correction processing on the second facial image to obtain a third facial image;

[0040] The third facial image is processed according to a preset rule to obtain a first facial image.

[0041] In this solution, the feature image is de-illuminated to obtain the second face image, specifically:

[0042] Classifying the characteristic images to obtain illumination low-frequency image information and high-frequency illumination image information;

[0043] Performing de-illumination processing on the illuminated low-frequency image information to obtain fourth face image information;

[0044] The fourth facial image information is fused with the high-frequency illumination image information to obtain a second facial image.

[0045] In this solution, the illumination low-frequency image information is de-illuminated to obtain the fourth face image information, specifically:

[0046] Obtain lighting information and noise information;

[0047] The fourth face image information is obtained by calculating the illumination information and the noise information according to the following formula:

[0048]

[0049] Among them, S is the fourth face image information after de-illumination, Y is the inherent illumination low-frequency image information of the face, G is the illumination information, n is the noise information, and x is the coordinate value of each point in the image.

[0050] In this solution, the second face image is subjected to posture correction processing to obtain a third face image, specifically:

[0051] Get posture information;

[0052] The second facial image is input into the first neural network model according to the posture information, and a third facial image is output.

[0053] In this solution, the acquisition of posture information is specifically as follows:

[0054] The second face image is sent to a second neural network model and the posture information is output.

[0055] A third aspect of the present invention provides a storage medium, which includes a household registration image judgment method program based on image recognition. When the household registration image judgment method program based on image recognition is executed by a processor, the steps of a household registration image judgment method based on image recognition as described in any one of the above items are implemented.

[0056] The present invention discloses a household registration image judgment method, system and readable storage medium based on image recognition. The method collects the user's image through a self-service card issuing machine, performs de-illumination and posture correction processing to make the image clearer and restore the user's real face data. After the image processing, the method can also perform feature comparison and recognition based on pre-stored face images, thereby improving the accuracy of user identity recognition and judgment in the diversity of actual scenes. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 A flowchart of a household registration image determination method based on image recognition according to the present invention is shown;

[0058] Figure 2 A block diagram of a household registration image determination system based on image recognition according to the present invention is shown. DETAILED DESCRIPTION

[0059] 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 a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0060] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of this application, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0061] Figure 1 A flow chart of a household registration image determination method based on image recognition according to the present invention is shown.

[0062] like Figure 1 As shown, the present invention provides a household registration image determination method based on image recognition, comprising:

[0063] S101, obtaining facial image information;

[0064] S102, performing feature extraction on the facial image information to obtain a feature image;

[0065] S103, processing the feature image to obtain a first face image;

[0066] S104, obtaining pre-stored user face image information;

[0067] S105, performing feature comparison between the user face image information and the first face image to obtain a feature difference; if the feature difference is less than a preset difference threshold, the recognition is successful.

[0068] It should be noted that the self-service card issuing machine may cause the collected facial image to be unclear, or there may be light or high noise due to environmental factors or various hardware factors. In this case, since the acquired image is not clear or complete, it is not conducive to verifying the identity of the user. First, the camera device of the self-service card issuing machine can collect images of the user's face in operation and obtain the user's facial image information. Then, the acquired facial image information is feature-extracted to obtain a feature image. The feature extraction method in the prior art can be used to extract features from the image, and this application will not repeat the feature extraction method. Then, the feature image is processed to obtain a first facial image. At this time, the first facial image is a clear facial image after processing. The pre-stored user facial image information is feature-compared with the first facial image to obtain a feature difference, wherein the pre-stored user facial image can be an image already stored in the background server, or an image pre-stored locally. The feature difference is compared with the preset difference threshold to see whether it is less than the preset difference threshold. If it is less than, it means that the current user has a high degree of match with the pre-stored user face image, and the current user is considered to be the real person and the recognition is passed. If it is greater than the preset difference threshold, it means that the current user has a low degree of match with the pre-stored user face image, and a warning message is sent to remind that the current user may not be the real person. The difference threshold is set by those skilled in the art according to actual needs.

[0069] According to an embodiment of the present invention, the processing of the feature image to obtain the first face image is specifically as follows:

[0070] De-illumination the feature image to obtain a second face image;

[0071] Performing posture correction processing on the second facial image to obtain a third facial image;

[0072] The third facial image is processed according to a preset rule to obtain a first facial image.

[0073] It should be noted that due to the uncertainty of the environment in which the self-service card issuing machine is located, the collected user face image may be illuminated, and the presence of light will affect the result of face recognition, and the user may not be able to aim at the camera device, resulting in the collected image not facing the user's face, and face recognition cannot be performed well. The second face image is the image after de-illumination. The third face image is the face image after face straightening, that is, the posture is straightened. After the de-illumination processing and the posture straightening processing, the third face image can be obtained. The third face image is a face image without illumination facing the face, and image recognition can be better performed at this time.

[0074] According to an embodiment of the present invention, the processing of the third face image according to a preset rule is specifically:

[0075] Obtain multiple face image information;

[0076] Processing the multiple facial image information to obtain multiple third facial images;

[0077] The plurality of third facial images are mean-fused to obtain a first facial image.

[0078] It should be noted that after the de-illumination process and the posture correction process, the image information obtained may have a large deviation from the actual person, that is, distortion. In order to avoid this situation, the third face image obtained by multiple processing can be fused. The present invention adopts mean fusion, that is, the weight of each image is the same, and multiple images are fused together to form a first face image. For example, if the face image information is obtained 5 times, the weight of each image information is 0.2, and then fusion is performed to obtain the first face image.

[0079] According to an embodiment of the present invention, performing de-illumination processing on the feature image to obtain the second face image is specifically:

[0080] Classifying the characteristic images to obtain illumination low-frequency image information and high-frequency illumination image information;

[0081] Performing de-illumination processing on the illuminated low-frequency image information to obtain fourth face image information;

[0082] The fourth facial image information is fused with the high-frequency illumination image information to obtain a second facial image.

[0083] It should be noted that the illumination low-frequency image information is the face area information that is sensitive to illumination, and the high-frequency illumination image information is the face area information that is insensitive to illumination, such as the edge and deflection angle of the face. First, the feature image is classified to obtain the illumination low-frequency image information and the high-frequency illumination image information. The image of the illuminated part and the image of the insensitive part can be separated by classification. Then, the illumination low-frequency image information is de-illuminated to obtain the fourth face image information, and finally, the fourth face image information is fused with the high-frequency illumination image information to obtain the second face image.

[0084] According to an embodiment of the present invention, the de-illumination process is performed on the illuminated low-frequency image information to obtain the fourth face image information, specifically:

[0085] Obtain lighting information and noise information;

[0086] The fourth face image information is obtained by calculating the illumination information and the noise information according to the following formula:

[0087]

[0088] Among them, S is the fourth face image information after de-illumination, Y is the inherent illumination low-frequency image information of the face, G is the illumination information, n is the noise information, and x is the coordinate value of each point in the image.

[0089] It should be noted that Y is the inherent illumination low-frequency image information of the face, which may be the characteristic information of the inherent illumination low-frequency image of the face. The inherent illumination low-frequency image information of the face is obtained by analyzing and calculating the current image or after a large number of image training. It is worth mentioning that the image after the inherent illumination low-frequency image information of the face is multiplied by the illumination information may still have noise, which can be determined by collecting and analyzing the current image. The noise here may make the image unclear, so it is necessary to remove the noise. The analysis and determination of noise can be obtained using the noise analysis method commonly used in the field, which will not be repeated in this application.

[0090] According to an embodiment of the present invention, performing posture correction processing on the second face image to obtain the third face image is specifically:

[0091] Get posture information;

[0092] The second facial image is input into the first neural network model according to the posture information, and a third facial image is output.

[0093] It should be noted that, since the user's face in the acquired user face image information may not be frontal, but may be at a certain angle, it is necessary to perform face straightening processing for this situation, but the straightening processing requires the correct acquisition of the face deflection angle, that is, the posture information. Then, the second face image is input into the first neural network model according to the posture information, and the third face image is output. The third face image is the face image after the face straightening.

[0094] According to an embodiment of the present invention, the acquiring of posture information is specifically:

[0095] The second face image is sent to a second neural network model and the posture information is output.

[0096] It should be noted that a neural network model can be used to obtain posture information. Through the trained neural network model, the second face image can be sent to the second neural network model to output posture information, that is, information such as the deflection angle of the face.

[0097] According to an embodiment of the present invention, the training method of the first neural network model is specifically as follows:

[0098] Obtain a large amount of face image training data;

[0099] Preprocessing the face image training data to obtain a training image set;

[0100] Inputting the training image set into the initialized neural network model to obtain a trained neural network model;

[0101] Calculating the deviation rate of the trained neural network model;

[0102] If the deviation rate is less than a preset deviation rate threshold, the training is stopped to obtain a first neural network model.

[0103] It should be noted that the training method of the second neural network is the same, and the large amount of face image training data can be obtained through the cloud network or historical face image data.

[0104] Figure 2 A block diagram of a household registration image determination system based on image recognition according to the present invention is shown.

[0105] like Figure 2 As shown, the second aspect of the present invention provides a household registration image determination system 2 based on image recognition, the system comprising: a memory 21 and a processor 22, the memory comprising a program of a household registration image determination method based on image recognition, and the program of the household registration image determination method based on image recognition is executed by the processor to implement the following steps:

[0106] Obtaining facial image information;

[0107] Performing feature extraction on the facial image information to obtain a feature image;

[0108] Processing the feature image to obtain a first face image;

[0109] Obtain pre-stored user face image information;

[0110] Performing feature comparison between the user face image information and the first face image to obtain a feature difference;

[0111] If the feature difference is less than the preset difference threshold, the recognition is successful.

[0112] It should be noted that the self-service card issuing machine may cause the collected facial image to be unclear, or there may be light or high noise due to environmental factors or various hardware factors. In this case, since the acquired image is not clear or complete, it is not conducive to verifying the identity of the user. First, the camera device of the self-service card issuing machine can collect images of the user's face in operation and obtain the user's facial image information. Then, the acquired facial image information is feature-extracted to obtain a feature image. The feature extraction method in the prior art can be used to extract features from the image, and this application will not repeat the feature extraction method. Then, the feature image is processed to obtain a first facial image. At this time, the first facial image is a clear facial image after processing. The pre-stored user facial image information is feature-compared with the first facial image to obtain a feature difference, wherein the pre-stored user facial image can be an image already stored in the background server, or an image pre-stored locally. The feature difference is compared with the preset difference threshold to see whether it is less than the preset difference threshold. If it is less than, it means that the current user has a high degree of match with the pre-stored user face image, and the current user is considered to be the real person and the recognition is passed. If it is greater than the preset difference threshold, it means that the current user has a low degree of match with the pre-stored user face image, and a warning message is sent to remind that the current user may not be the real person. The difference threshold is set by those skilled in the art according to actual needs.

[0113] According to an embodiment of the present invention, the processing of the feature image to obtain the first face image is specifically as follows:

[0114] De-illumination the feature image to obtain a second face image;

[0115] Performing posture correction processing on the second facial image to obtain a third facial image;

[0116] The third facial image is processed according to a preset rule to obtain a first facial image.

[0117] It should be noted that due to the uncertainty of the environment in which the self-service card issuing machine is located, the collected user face image may be illuminated, and the presence of light will affect the result of face recognition, and the user may not be able to aim at the camera device, resulting in the collected image not facing the user's face, and face recognition cannot be performed well. The second face image is the image after de-illumination. The third face image is the face image after face straightening, that is, the posture is straightened. After the de-illumination processing and the posture straightening processing, the third face image can be obtained. The third face image is a face image without illumination facing the face, and image recognition can be better performed at this time.

[0118] According to an embodiment of the present invention, the processing of the third face image according to a preset rule is specifically:

[0119] Obtain multiple face image information;

[0120] Processing the multiple facial image information to obtain multiple third facial images;

[0121] The plurality of third facial images are mean-fused to obtain a first facial image.

[0122] It should be noted that after the de-illumination process and the posture correction process, the image information obtained may have a large deviation from the actual person, that is, distortion. In order to avoid this situation, the third face image obtained by multiple processing can be fused. The present invention adopts mean fusion, that is, the weight of each image is the same, and multiple images are fused together to form a first face image. For example, if the face image information is obtained 5 times, the weight of each image information is 0.2, and then fusion is performed to obtain the first face image.

[0123] According to an embodiment of the present invention, performing de-illumination processing on the feature image to obtain the second face image is specifically:

[0124] Classifying the characteristic images to obtain illumination low-frequency image information and high-frequency illumination image information;

[0125] Performing de-illumination processing on the illuminated low-frequency image information to obtain fourth face image information;

[0126] The fourth facial image information is fused with the high-frequency illumination image information to obtain a second facial image.

[0127] It should be noted that the illumination low-frequency image information is the face area information that is sensitive to illumination, and the high-frequency illumination image information is the face area information that is insensitive to illumination, such as the edge and deflection angle of the face. First, the feature image is classified to obtain the illumination low-frequency image information and the high-frequency illumination image information. The image of the illuminated part and the image of the insensitive part can be separated by classification. Then, the illumination low-frequency image information is de-illuminated to obtain the fourth face image information, and finally, the fourth face image information is fused with the high-frequency illumination image information to obtain the second face image.

[0128] According to an embodiment of the present invention, the de-illumination process is performed on the illuminated low-frequency image information to obtain the fourth face image information, specifically:

[0129] Obtain lighting information and noise information;

[0130] The fourth face image information is obtained by calculating the illumination information and the noise information according to the following formula:

[0131]

[0132] Among them, S is the fourth face image information after de-illumination, Y is the inherent illumination low-frequency image information of the face, G is the illumination information, n is the noise information, and x is the coordinate value of each point in the image.

[0133] It should be noted that Y is the inherent illumination low-frequency image information of the face, which may be the characteristic information of the inherent illumination low-frequency image of the face. The inherent illumination low-frequency image information of the face is obtained by analyzing and calculating the current image or after a large number of image training. It is worth mentioning that the image after the inherent illumination low-frequency image information of the face is multiplied by the illumination information may still have noise, which can be determined by collecting and analyzing the current image. The noise here may make the image unclear, so it is necessary to remove the noise. The analysis and determination of noise can be obtained using the noise analysis method commonly used in the field, which will not be repeated in this application.

[0134] According to an embodiment of the present invention, performing posture correction processing on the second face image to obtain the third face image is specifically:

[0135] Get posture information;

[0136] The second facial image is input into the first neural network model according to the posture information, and a third facial image is output.

[0137] It should be noted that, since the user's face in the acquired user face image information may not be frontal, but may be at a certain angle, it is necessary to perform face straightening processing for this situation, but the straightening processing requires the correct acquisition of the face deflection angle, that is, the posture information. Then, the second face image is input into the first neural network model according to the posture information, and the third face image is output. The third face image is the face image after the face straightening.

[0138] According to an embodiment of the present invention, the acquiring of posture information is specifically:

[0139] The second face image is sent to a second neural network model and the posture information is output.

[0140] It should be noted that a neural network model can be used to obtain posture information. Through the trained neural network model, the second face image can be sent to the second neural network model to output posture information, that is, information such as the deflection angle of the face.

[0141] According to an embodiment of the present invention, the training method of the first neural network model is specifically as follows:

[0142] Obtain a large amount of face image training data;

[0143] Preprocessing the face image training data to obtain a training image set;

[0144] Inputting the training image set into the initialized neural network model to obtain a trained neural network model;

[0145] Calculating the deviation rate of the trained neural network model;

[0146] If the deviation rate is less than a preset deviation rate threshold, the training is stopped to obtain a first neural network model.

[0147] It should be noted that the training method of the second neural network is the same, and the large amount of face image training data can be obtained through the cloud network or historical face image data.

[0148] A third aspect of the present invention provides a storage medium, which includes a household registration image judgment method program based on image recognition. When the household registration image judgment method program based on image recognition is executed by a processor, the steps of a household registration image judgment method based on image recognition as described in any one of the above items are implemented.

[0149] The present invention discloses a household registration image judgment method, system and readable storage medium based on image recognition. The method collects the user's image through a self-service card issuing machine, performs de-illumination and posture correction processing to make the image clearer and restore the user's real face data. After the image processing, the method can also perform feature comparison and recognition based on pre-stored face images, thereby improving the accuracy of user identity recognition and judgment in the diversity of actual scenes.

[0150] Among them, the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable red-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0151] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0152] In addition, the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0153] Furthermore, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0154] In this document, relational terms such as first and second, etc. are used merely to distinguish one entity or operation from another entity or operation, but do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0155] The above description is only an embodiment of the present application and is not intended to limit the protection scope of the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A household registration image judgment method based on image recognition, It is characterized in that include: Obtaining facial image information; Performing feature extraction on the facial image information to obtain a feature image; Processing the feature image to obtain a first face image; Obtain pre-stored user face image information; Performing feature comparison between the user face image information and the first face image to obtain a feature difference; If the feature difference is less than the preset difference threshold, the recognition is successful; The processing of the feature image to obtain the first face image is specifically as follows: De-illumination the feature image to obtain a second face image; Performing posture correction processing on the second facial image to obtain a third facial image; Processing the third facial image according to a preset rule to obtain a first facial image; The de-illumination process is performed on the feature image to obtain a second face image, specifically: Classifying the characteristic images to obtain illumination low-frequency image information and high-frequency illumination image information; Performing de-illumination processing on the illuminated low-frequency image information to obtain fourth face image information; fusing the fourth facial image information with the high-frequency illumination image information to obtain a second facial image; The second face image is subjected to posture correction processing to obtain a third face image, specifically: Sending the second face image to a second neural network model and outputting posture information; The second facial image is input into the first neural network model according to the posture information, and a third facial image is output.

2. According to the household registration image determination method based on image recognition according to claim 1, It is characterized in that The de-illumination process is performed on the illuminated low-frequency image information to obtain the fourth face image information, specifically: Obtain lighting information and noise information; The fourth face image information is obtained by calculating the illumination information and the noise information according to the following formula: S(x)=Y(x)G(x)-n(x) Among them, S is the fourth face image information after de-illumination, Y is the inherent illumination low-frequency image information of the face, G is the illumination information, n is the noise information, and x is the coordinate value of each point in the image.

3. A household registration image judgment system based on image recognition, It is characterized in that The system includes: a memory and a processor, wherein the memory includes a program of a household registration image determination method based on image recognition, and when the program of the household registration image determination method based on image recognition is executed by the processor, the following steps are implemented: Obtaining facial image information; Performing feature extraction on the facial image information to obtain a feature image; Processing the feature image to obtain a first face image; Obtain pre-stored user face image information; Performing feature comparison between the user face image information and the first face image to obtain a feature difference; If the feature difference is less than the preset difference threshold, the recognition is successful; The processing of the feature image to obtain the first face image is specifically as follows: De-illumination the feature image to obtain a second face image; Performing posture correction processing on the second facial image to obtain a third facial image; Processing the third facial image according to a preset rule to obtain a first facial image; The de-illumination process is performed on the feature image to obtain the second face image, specifically: Classifying the characteristic images to obtain illumination low-frequency image information and high-frequency illumination image information; Performing de-illumination processing on the illuminated low-frequency image information to obtain fourth face image information; fusing the fourth facial image information with the high-frequency illumination image information to obtain a second facial image; The second face image is subjected to posture correction processing to obtain a third face image, specifically: Sending the second face image to a second neural network model and outputting posture information; The second facial image is input into the first neural network model according to the posture information, and a third facial image is output.

4. A storage medium, It is characterized in that The storage medium includes a household registration image determination method program based on image recognition. When the household registration image determination method program based on image recognition is executed by a processor, the steps of a household registration image determination method based on image recognition as described in any one of claims 1 to 2 are implemented.