Image conversion method and device applied to image-code joint acquisition system

By using the generated adversarial network model to convert face images and optimize features in the image collection environment, the problem of the difference in the acquisition environment affecting the recognition accuracy is solved, and the accuracy of face image recognition in the image collection system is improved. and efficiency.

CN112766105BActive Publication Date: 2024-11-05BEIJING MA NIU TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202110020925.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-01-07
Publication Date
2024-11-05
Estimated Expiration
2041-01-07

AI Technical Summary

Technical Problem

Due to factors such as the environment, light and occlusion of the existing image code collection system, the difference between the standard face images and the face images collected by the system is large, which affects the accuracy of image recognition.

Method used

By obtaining the reference images collected by the current standard face images and image code collection system in different real environments, extracting the target recognition area and face features, using the generated adversarial network model to convert and feature optimization, and obtaining corresponding to different real environments standard target area images to improve image quality.

Benefits of technology

The recognition accuracy and recognition efficiency of the image code collection system for the acquisition of face images is improved, and the standard face image samples for image recognition comparison are enriched.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN112766105B_ABST
    Figure CN112766105B_ABST
Patent Text Reader

Abstract

The present invention provides an image conversion method and device applied to a joint image and code acquisition system, the method comprising: obtaining a current standard face image and a reference image acquired by the current joint image and code acquisition system under different real environments; extracting a target recognition area and the face features corresponding to the target recognition area from the current standard face image; performing image conversion on the target recognition area using each reference image to obtain an initial target area image corresponding to different real environments; and optimizing the features of the initial target area image based on the face features to obtain a standard target area image. Thus, by using the reference images acquired by the current joint image and code acquisition system under different real environments to perform image conversion on the standard face image, the image quality of the standard target area image is improved, thereby enriching the standard face image samples used for image recognition comparison, and further improving the recognition accuracy and recognition efficiency of the joint image and code acquisition system for the acquired face images.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to an image conversion method and device applied to an image-code joint acquisition system. Background Art

[0002] The existing image and code joint acquisition system is a system based on the collection of structured and unstructured full information, cross-integration of big data, and multi-dimensional information archives and behaviors of key personnel. It can meet the actual combat needs of multiple police types. The image and code joint acquisition system can collect mobile phone identification codes, facial images, and license plate information data in one, complement each other's advantages, and horizontally connect online and offline trajectory data. The image and code joint acquisition system achieves face recognition by comparing the collected facial images with the standard facial images stored in the standard identity information image database. However, due to the influence of factors such as the collection environment, light, and occlusion of the collection equipment, there is a large difference between the standard facial images and the facial images collected by the system, which affects the accuracy of image recognition, and thus affects the accuracy of the personnel information collected by the entire image and code joint acquisition system. Summary of the invention

[0003] In view of this, an embodiment of the present invention provides an image conversion method and device applied to a joint image and code acquisition system to overcome the problem in the prior art that the difference between the standard face image used by the joint image and code acquisition system and the collected face image is large, which affects the accuracy of image recognition.

[0004] According to a first aspect, an embodiment of the present invention provides an image conversion method applied to an image-code joint acquisition system, comprising:

[0005] Obtain the current standard face image and the reference images collected by the current image-code joint acquisition system under different real environments;

[0006] Extracting a target recognition area and facial features corresponding to the target recognition area from the current standard facial image;

[0007] Performing image conversion on the target recognition area using each of the reference images to obtain initial target area images corresponding to different real environments;

[0008] Based on the facial features, the initial target area image is feature optimized to obtain a standard target area image.

[0009] Optionally, extracting a target recognition area and facial features corresponding to the target recognition area from the current standard facial image includes:

[0010] Acquire attribute information of the position to be identified and attribute annotation information of the current standard face image, wherein the attribute information is the type of facial organs, and the attribute annotation information is the position coordinate information of different facial organs;

[0011] Based on the attribute annotation information, extracting a target recognition area from the current standard face image;

[0012] Feature extraction is performed on the target recognition area to obtain the facial features.

[0013] Optionally, based on the facial features, using each of the reference images to perform image conversion on the target recognition area to obtain standard target area images corresponding to different real environments includes:

[0014] The current reference image and the target recognition area are input into a preset generative adversarial network model to obtain a standard target area image corresponding to the current reference image.

[0015] Optionally, the preset generative adversarial network model is trained in the following manner:

[0016] Acquire historical standard facial images, reference images collected by the current image-code joint acquisition system in different real environments, and real facial images corresponding to the historical standard facial images in different real environments;

[0017] Inputting the reference images and historical standard face images collected by the previous image-code joint acquisition system in different real environments into the initial generative adversarial network model to obtain predicted face images in different real environments;

[0018] Based on real face images and predicted face images in different real environments, the parameters of the initial generative adversarial network model are adjusted to obtain the preset generative adversarial network model.

[0019] Optionally, the optimizing the features of the initial target area image based on the facial features to obtain a standard target area image includes:

[0020] Extracting current face image features from the initial target area image;

[0021] Updating the current facial image features based on the facial features;

[0022] The initial target area image is updated using the updated current face image features to obtain the standard target area image.

[0023] According to a second aspect, an embodiment of the present invention provides an image conversion device applied to a joint image and code acquisition system, comprising:

[0024] An acquisition module is used to acquire the current standard face image and the reference image collected by the current image-code joint acquisition system under different real environments;

[0025] A first processing module, used for extracting a target recognition area and facial features corresponding to the target recognition area from the current standard facial image;

[0026] A second processing module is used to perform image conversion on the target recognition area using each of the reference images to obtain initial target area images corresponding to different real environments;

[0027] The third processing module is used to optimize the features of the initial target area image based on the facial features to obtain a standard target area image.

[0028] Optionally, the first processing module is specifically used to obtain attribute information of the position to be identified and attribute annotation information of the current standard facial image, the attribute information is the type of facial organs, and the attribute annotation information is the position coordinate information of different facial organs; based on the attribute image and attribute annotation information, extract the target recognition area from the current standard facial image; perform feature extraction on the target recognition area to obtain the facial features.

[0029] Optionally, the second processing module is specifically used to input the current reference image and the target recognition area into a preset generative adversarial network model to obtain a standard target area image corresponding to the current reference image.

[0030] According to a third aspect, an embodiment of the present invention provides an electronic device, including:

[0031] A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method described in the first aspect and any one of its optional embodiments by executing the computer instructions.

[0032] According to a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, wherein the computer instructions are used to enable the computer to execute the method described in the first aspect or any optional embodiment thereof.

[0033] The technical solution of the present invention has the following advantages:

[0034] The image conversion method and device applied to the image-code joint acquisition system provided by the embodiment of the present invention include: obtaining the current standard face image and the reference image collected by the current image-code joint acquisition system under different real environments; extracting the target recognition area and the facial features corresponding to the target recognition area from the current standard face image; using each reference image to perform image conversion on the target recognition area to obtain the initial target area image corresponding to the different real environments; based on the facial features, feature optimization is performed on the initial target area image to obtain the standard target area image. Thus, by using the reference image collected by the current image-code joint acquisition system under different real environments, the standard face image is converted to obtain the initial area image corresponding to the different real environments, and by using the facial features in the standard face image, the initial area image is feature optimized to better retain the image identity features and improve the image quality of the standard target area image, thereby enriching the standard face image samples used for image recognition comparison, and further improving the recognition accuracy and recognition efficiency of the image-code joint acquisition system for the collected face images. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0036] Figure 1 It is a flow chart of an image conversion method applied to an image-code joint acquisition system according to an embodiment of the present invention;

[0037] Figure 2 A schematic diagram of the training process of a preset generative adversarial network model according to an embodiment of the present invention;

[0038] Figure 3 It is a structural schematic diagram of an image conversion device applied to an image-code joint acquisition system according to an embodiment of the present invention;

[0039] Figure 4 FIG. 4 is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0041] The technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0042] The existing image and code joint acquisition system is a system based on the collection of structured and unstructured full information, cross-integration of big data, and multi-dimensional information archives and behaviors of key personnel. It can meet the actual combat needs of multiple police types. The image and code joint acquisition system can collect mobile phone identification codes, facial images, and license plate information data in one, complement each other's advantages, and horizontally connect online and offline trajectory data. The image and code joint acquisition system achieves face recognition by comparing the collected facial images with the standard facial images stored in the standard identity information image database. However, due to the influence of factors such as the collection environment, light, and occlusion of the collection equipment, there is a large difference between the standard facial images and the facial images collected by the system, which affects the accuracy of image recognition, and thus affects the accuracy of the personnel information collected by the entire image and code joint acquisition system.

[0043] Based on the above problems, an embodiment of the present invention provides an image conversion method applied to a joint image and code acquisition system, such as Figure 1 As shown, the image conversion method applied to the image-code joint acquisition system includes:

[0044] Step S101: obtaining a current standard face image and reference images collected by the current image-code joint acquisition system under different real environments.

[0045] The standard face image corresponds to the personnel information of the face and can be used for face recognition. Usually, the current standard face image is an image taken under a unified environment standard, such as a photo with face information on an ID card or passport, or an image collected in a preset personnel information system, such as a student image containing a portrait in a student management system. The reference image can be a face image of different people taken under different lighting, weather, hairstyle, expression and other conditions.

[0046] Step S102: extracting the target recognition area and the facial features corresponding to the target recognition area from the current standard facial image.

[0047] Among them, since facial images are composed of eyes, nose, mouth, eyebrows, and face parts, and each part of each person has its own image characteristics, in order to improve the efficiency of image conversion and improve the accuracy of image conversion, the standard facial image is divided into regions according to different parts. For example, when the eyes are used as the basis for facial image recognition, the target recognition area is extracted as the eye area of ​​the facial image; when the mouth is used as the basis for facial image recognition, the target recognition area is extracted as the eye area of ​​the facial image, and the present invention is not limited to this.

[0048] Step S103: performing image conversion on the target recognition area using each reference image to obtain initial target area images corresponding to different real environments.

[0049] Specifically, by inputting the current reference image and the target recognition area into a preset generative adversarial network model, a standard target area image corresponding to the current reference image is obtained.

[0050] Step S104: Based on the facial features, the initial target area image is optimized to obtain a standard target area image.

[0051] By executing the above steps, the image conversion method applied to the image and code joint acquisition system provided by the embodiment of the present invention, by utilizing the reference images acquired by the current image and code joint acquisition system in different real environments, performs image conversion on the standard face image to obtain the initial area images corresponding to the different real environments, and optimizes the features of the initial area image by utilizing the facial features in the standard face image to better retain the image identity features and improve the image quality of the production standard target area image, thereby enriching the standard face image samples used for image recognition comparison, and further improving the recognition accuracy and efficiency of the image and code joint acquisition system for the acquired face images.

[0052] Specifically, in one embodiment, the above step S102 specifically includes the following steps:

[0053] Step S201: Acquire attribute information of the position to be identified and attribute annotation information of the current standard face image.

[0054] Among them, the attribute information is the type of facial organs, and the attribute annotation information is the position coordinate information of different facial organs.

[0055] Step S202: extracting a target recognition area from the current standard face image based on the attribute annotation information.

[0056] Among them, taking the above-mentioned target recognition area as the eye area of ​​the face image as an example, by obtaining the position coordinate information corresponding to the glasses from the attribute standard information, the image of the area where the glasses are located is extracted from the current standard face image as the target recognition area.

[0057] Step S203: extracting features from the target recognition area to obtain facial features.

[0058] Among them, for different target recognition areas, feature extraction can be performed using a neural network, facial features can be extracted manually, or both methods can be used. The facial features are features that characterize a person's identity, such as image features related to freckles and scars located in the eye area. The extracted feature types may include: grayscale features, texture features, shape features, etc., but the present invention is not limited to this.

[0059] Specifically, in one embodiment, the preset generative adversarial network model in the above step S103 is trained in the following manner:

[0060] Obtain historical standard facial images, reference images collected by the current image-code joint acquisition system in different real environments, and real facial images corresponding to historical standard facial images in different real environments.

[0061] Input the reference images and historical standard face images collected by the previous image-code joint acquisition system in different real environments into the initial generative adversarial network model to obtain the predicted face images in different real environments;

[0062] Based on real face images and predicted face images in different real environments, the parameters of the initial generative adversarial network model are adjusted to obtain a preset generative adversarial network model.

[0063] Specifically, in one embodiment, the above step S104 specifically includes the following steps:

[0064] Step S401: extracting current face image features from the initial target area image.

[0065] The method of extracting the features of the current face image is similar to the above step S203 and will not be described in detail here.

[0066] Step S402: Update the features of the current face image based on the face features. The face features are features that characterize the identity of a person, such as freckles and scars in the eye area.

[0067] Step S403: Update the initial target area image using the updated current face image features to obtain a standard target area image. By optimizing the current face image features based on the features that characterize the identity of the person, the image identity features are retained, which can not only improve the quality of the generated image, but also be used to expand the data of the image set.

[0068] In practical applications, since the image information collected by the image-code joint acquisition system is mostly facial information, specific attributes are used to mark facial images. For the complex problem of multi-domain image conversion, one of the specific areas is selected as a sub-image for analysis and conversion, especially areas with specific significance, which usually refer to areas that have a significant impact on personnel recognition, such as noses or eyes. In the image conversion process, the elements and attributes of the image of the area can be extracted first. The extraction process is also a layered process. The purpose of layering is to obtain different attributes or element features and generate a feature vector of the target domain. In this way, the target domain feature vector can be controlled and a controllable result can be obtained. The input source domain image is x, the target domain condition information is v, and the generated image is x′. The conversion process is expressed as: (x, v)→x′.

[0069] For feature conversion, a feature-enhanced multi-attribute face image generation model can be used. The source image and the reference image are used as input. The encoded attribute parts are exchanged and decoded to generate an image with reference attributes. The generated image is feature optimized to retain the image identity features. This can not only improve the quality of the generated image, but also can be used for data expansion of the image set. In the image conversion process, the existing neural network method can be used for conversion. The embodiment of the present invention selects a generative adversarial network model as the conversion model, which will not be repeated here. Within a certain range, as the network depth increases, its ability to extract features gradually increases. The network depth of the generative adversarial network model can be increased from 3 layers to multiple layers, and the quality of the generated image can be improved by increasing the network depth. At the same time, the middle layers of the mutually symmetrical encoder and decoder in the model are connected to each other to realize the combination of the low-dimensional features obtained by the encoder part and the high-dimensional features obtained by the decoder as the input of the next layer. In this way, the original features of the image can be better preserved while the facial attributes are converted, and a more natural and realistic image can be generated. In order to extract the deep features of the image, a deep neural network is also added after the generator. Its main function is to map the data from the image space to the deep feature space, and to optimize the deep features extracted from the training samples and the generated images to assist in finding the optimized object features to ensure the generation of more realistic images.

[0070] For the generative adversarial network model, high-level attributes such as posture and expression can be automatically learned and unsupervised segmented, and the generated image also has random changes (such as freckles and hair). This architecture can intuitively and multi-scale control image synthesis, showing better interpolation properties, and further deletes and retains certain image attributes, such as the effect of superimposing low spatial resolution. The generated image replicates the high-level attributes of the original image such as the face shape, but the blurred background, hairstyle and all colors of the image are deleted. This is particularly important to prevent the imitation of facial images, while selectively retaining certain attributes of the hairstyle. Of course, more elements can be retained for high-resolution images. The present invention is not limited to this.

[0071] The principle of the Generative Adversarial Network model, referred to as GAN, is briefly explained as follows:

[0072] GAN contains two models: Generative and Discriminative. In this embodiment, the generative model is a standard template generated from pre-acquired sample face image information, while the discriminative network is a classifier that determines whether a given image is a real image from a dataset or an artificially created fake image. This is basically a binary classifier in the form of a convolutional neural network (CNN).

[0073] The discriminator network attempts to distinguish whether the received data is real data or data generated by the generator network. It needs to classify it based on the category it implements. Generally speaking, GAN is used in binary classification problems. The discriminator result is a number between 0 and 1, which is used to indicate the possibility that the input is considered to be real data. When the judgment result is 1, it is considered to come from real data, otherwise it is generated data.

[0074] 1. The generative network takes the vector of the given sample face pattern as the reference standard image, and retains the important facial feature values ​​as constants, such as the characteristic parameters of the eyes;

[0075] 2. The main task of the discriminant network is to judge whether a face is real or fake, so its input includes assumed real pictures and simulated pictures. Its judgment result will be used as an output.

[0076] 3. The generative network generates more realistic facial images in various environments through continuous iteration.

[0077] 4. The discriminant network continuously optimizes the criteria for distinguishing true from false, trying to identify each picture created by the generative network in different environments.

[0078] 5. In each round of iteration, they will provide each other with feedback on the successful attempts at adjustment they have made.

[0079] 6. Finally, with the help of the discriminant network, when the generative network has been trained to the point where the discriminant network cannot correctly distinguish between genuine and fake products, the iterative process can be stopped.

[0080] 7. In the process of the generator network and the discriminator network confronting each other, the changes in the parameters of each element vector are normalized and recorded, and a mean square deviation index is generated to serve as a reference for determining whether the information is true or false.

[0081] For a more intuitive understanding, Figure 2 As shown in the figure, it is a schematic diagram of the training process of the generative adversarial network model. The thick dotted line in the figure represents the distribution of real samples, the thin dotted line represents the distribution of the discriminator's discrimination probability, and the solid line represents the distribution of generated samples.

[0082] The training goal of the model is to use the generated sample distribution (solid line) to fit the real sample distribution (thick dashed line) to achieve the purpose of generating samples that are indistinguishable from the real ones. It can be seen that when the state (a) is in the initial state, the distribution generated by the generator is quite different from the real distribution, and the probability of the discriminator distinguishing the sample is not very stable, so the discriminator will be trained first to better distinguish the samples. After training the discriminator many times, the sample state (b) is reached, at which time the samples are distinguished very significantly and well. Then the generator is trained again. After training the generator, the sample state (c) is reached, at which time the generator distribution is closer to the real sample distribution than before. After repeated training iterations, it is hoped that the state (d) can be reached, where the generated sample distribution fits the real sample distribution, and the discriminator cannot distinguish whether the sample is generated or real (that is, the discriminant probability is equal to 0.5). At this point, the training goal is achieved and the model training is completed.

[0083] The method provided by the present invention makes image conversion more convenient and accurate, and at the same time has the ability to generate specified images, so that preliminary results can be obtained quickly, which is convenient for data processing and subsequent recognition, and quickly forms a structured data set, which is conducive to improving face recognition rate.

[0084] By executing the above steps, the image conversion method applied to the image and code joint acquisition system provided by the embodiment of the present invention, by utilizing the reference images acquired by the current image and code joint acquisition system in different real environments, performs image conversion on the standard face image to obtain the initial area images corresponding to the different real environments, and optimizes the features of the initial area image by utilizing the facial features in the standard face image to better retain the image identity features and improve the image quality of the production standard target area image, thereby enriching the standard face image samples used for image recognition comparison, and further improving the recognition accuracy and efficiency of the image and code joint acquisition system for the acquired face images.

[0085] The embodiment of the present invention also provides an image conversion device applied to the image-code joint acquisition system, such as Figure 3 As shown, the image conversion device of the image-code joint acquisition system includes:

[0086] The acquisition module 101 is used to acquire the current standard face image and the reference image acquired by the current image-code joint acquisition system in different real environments. For details, please refer to the relevant description of step S101 in the above method embodiment, which will not be repeated here.

[0087] The first processing module 102 is used to extract the target recognition area and the facial features corresponding to the target recognition area from the current standard face image. For details, please refer to the relevant description of step S102 in the above method embodiment, which will not be repeated here.

[0088] The second processing module 103 is used to perform image conversion on the target recognition area using each reference image to obtain an initial target area image corresponding to different real environments. For details, please refer to the relevant description of step S103 in the above method embodiment, which will not be repeated here.

[0089] The third processing module 104 is used to optimize the features of the initial target region image based on the facial features to obtain a standard target region image. For details, please refer to the relevant description of step S104 in the above method embodiment, which will not be repeated here.

[0090] Optionally, the first processing module 102 is specifically used to obtain attribute information of the position to be identified and attribute annotation information of the current standard face image, the attribute information is the type of facial organs, and the attribute annotation information is the position coordinate information of different facial organs; based on the attribute image attribute annotation information, extract the target recognition area from the current standard face image; perform feature extraction on the target recognition area to obtain facial features. For details, please refer to the relevant description of step S102 in the above method embodiment, which will not be repeated here.

[0091] Optionally, the second processing module 103 is specifically used to input the current reference image and the target recognition area into a preset generative adversarial network model to obtain a standard target area image corresponding to the current reference image. For details, see the relevant description of step S103 in the above method embodiment, which will not be repeated here.

[0092] The image conversion device applied to the image and code joint acquisition system provided in an embodiment of the present invention is used to execute the image conversion method applied to the image and code joint acquisition system provided in the above embodiment. Its implementation method and principle are the same. For details, please refer to the relevant description of the above method embodiment and will not be repeated here.

[0093] Through the coordinated cooperation of the above-mentioned components, the image conversion device applied to the image and code joint acquisition system provided by the embodiment of the present invention optimizes the features of the initial area image by utilizing the facial features in the standard facial image, so as to better retain the image identity features and improve the image quality of the produced standard target area image, thereby enriching the standard facial image samples used for image recognition and comparison, and further improving the recognition accuracy and efficiency of the image and code joint acquisition system for the collected facial images.

[0094] Figure 4 An electronic device according to an embodiment of the present invention is shown. Figure 4 As shown, the electronic device includes: a processor 901 and a memory 902, wherein the processor 901 and the memory 902 can be connected via a bus or other means. Figure 4 The example of connecting through bus is taken in the following.

[0095] The processor 901 may be a central processing unit (CPU). The processor 901 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or a combination of the above chips.

[0096] The memory 902 is a non-transitory computer-readable storage medium that can be used to store non-transitory software programs, non-transitory computer executable programs and modules, such as program instructions / modules corresponding to the methods in the above method embodiments. The processor 901 executes various functional applications and data processing of the processor by running the non-transitory software programs, instructions and modules stored in the memory 902, that is, implementing the methods in the above method embodiments.

[0097] The memory 902 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required by at least one function; the data storage area may store data created by the processor 901, etc. In addition, the memory 902 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 902 may optionally include a memory remotely arranged relative to the processor 901, and these remote memories may be connected to the processor 901 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0098] One or more modules are stored in the memory 902 , and when executed by the processor 901 , the method in the above method embodiment is executed.

[0099] The specific details of the above electronic device can be understood by referring to the corresponding descriptions and effects in the above method embodiments, and will not be repeated here.

[0100] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the implemented program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the storage medium can be a disk, an optical disk, a read-only memory (ROM), a random access memory (RAM), a flash memory, a hard disk drive (HDD) or a solid-state drive (SSD), etc.; the storage medium can also include a combination of the above-mentioned types of memory.

[0101] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. An image conversion method applied to a picture-code joint acquisition system, characterized in that: include: Obtaining a current standard face image and a reference image collected by the current image-code joint acquisition system under different real environments; the current standard face image is an image taken under a unified environmental standard; Extracting a target recognition area and facial features corresponding to the target recognition area from the current standard facial image; Performing image conversion on the target recognition area using each of the reference images to obtain initial target area images corresponding to different real environments; Based on the facial features, feature optimization is performed on the initial target area image to obtain a standard target area image; The step of optimizing the initial target area image based on the facial features to obtain a standard target area image includes: Extracting current face image features from the initial target area image; Updating the current facial image features based on the facial features; The initial target area image is updated using the updated current face image features to obtain the standard target area image.

2. The method according to claim 1, characterized in that The step of extracting the target recognition area and the facial features corresponding to the target recognition area from the current standard facial image includes: Acquire attribute information of the position to be identified and attribute annotation information of the current standard face image, wherein the attribute information is the type of facial organs, and the attribute annotation information is the position coordinate information of different facial organs; Based on the attribute annotation information, extracting a target recognition area from the current standard face image; Feature extraction is performed on the target recognition area to obtain the facial features.

3. The method according to claim 1, characterized in that The step of performing image conversion on the target recognition area based on the facial features using each of the reference images to obtain standard target area images corresponding to different real environments includes: The current reference image and the target recognition area are input into a preset generative adversarial network model to obtain a standard target area image corresponding to the current reference image.

4. The method according to claim 3, characterized in that The preset generative adversarial network model is trained in the following way: Acquire historical standard facial images, reference images collected by the current image-code joint acquisition system in different real environments, and real facial images corresponding to the historical standard facial images in different real environments; Inputting the reference images and historical standard face images collected by the previous image-code joint acquisition system in different real environments into the initial generative adversarial network model to obtain predicted face images in different real environments; Based on real face images and predicted face images in different real environments, the parameters of the initial generative adversarial network model are adjusted to obtain the preset generative adversarial network model.

5. An image conversion device applied to a joint image and code acquisition system, characterized in that: include: An acquisition module is used to acquire the current standard face image and the reference image acquired by the current image-code joint acquisition system in different real environments; the current standard face image is an image taken under a unified environmental standard; A first processing module, used for extracting a target recognition area and facial features corresponding to the target recognition area from the current standard facial image; A second processing module is used to perform image conversion on the target recognition area using each of the reference images to obtain initial target area images corresponding to different real environments; A third processing module is used to optimize the features of the initial target area image based on the facial features to obtain a standard target area image; The third processing module is further used for: Extracting current face image features from the initial target area image; Updating the current facial image features based on the facial features; The initial target area image is updated using the updated current face image features to obtain the standard target area image.

6. The device according to claim 5, characterized in that The first processing module is specifically used to obtain attribute information of the position to be identified and attribute annotation information of the current standard face image, the attribute information is the type of facial organs, and the attribute annotation information is the position coordinate information of different facial organs; based on the attribute annotation information, extract the target recognition area from the current standard face image; perform feature extraction on the target recognition area to obtain the facial features.

7. The device according to claim 5, characterized in that The second processing module is specifically used to input the current reference image and the target recognition area into a preset generative adversarial network model to obtain a standard target area image corresponding to the current reference image.

8. An electronic device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method according to any one of claims 1 to 4 by executing the computer instructions.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the method according to any one of claims 1 to 4.