Iris photo method and device and storage medium

Abstract

The invention provides an iris photo method and device and a storage medium, and the method comprises the steps: inputting a to-be-processed target original iris image into a first image generation model, and obtaining a target intermediate iris image; the first image generation model is used for converting the target original iris image and outputting a target intermediate iris image which meets a set similarity threshold compared with the target original iris image; the first image generation model is used for carrying out deep learning model learning on image features extracted from an iris image after the size, direction, length, weight and / or color depth adjustment is carried out on plaques based on an original iris image sample; forming and determining an intermediate iris image which can output a similarity threshold meeting the set similarity threshold compared with the corresponding original iris image sample; and inputting the target intermediate iris image into a second image generation model, and obtaining a target iris photo image according to a predetermined training requirement. By adopting the photo method provided by the invention, the safety of iris information can be protected.

Description

Technical Field

[0001] The present invention relates to the technical field of iris image processing, and in particular to an iris photography method, device and storage medium. Background Art

[0002] Recently, a unique new photography format—"iris photography"—has become a trend online, especially among young people. "Iris photography" uses high-definition photography to capture the detailed texture of the human iris, then applies special effects to the image. To take the photo, the client places their chin on a tray. Strong lights illuminate the pupil from both sides, and a macro camera captures the iris inside the eye. The entire process takes only a few minutes. After post-processing, a unique "iris portrait" is created. Many young people have been sharing their "iris portraits" on social media, marveling at the unique effects of their irises.

[0003] In the process of implementing the present invention, the inventors discovered that the prior art has at least the following problems:

[0004] In terms of information security, "iris photos" pose a significant risk. As a key biometric characteristic, the iris, like fingerprints and genes, is unique. Each person's iris pattern is unique and remains virtually unchanged throughout life after being fixed in childhood. This constitutes sensitive personal information as defined in Article 28 of the Personal Information Protection Law of the People's Republic of China.

[0005] Because photo-sharing is a new form of identity verification, its management regulations are still underdeveloped. Many businesses store and transmit customers' iris photos online, posing a serious threat to personal privacy. Unlike passwords, iris information cannot be changed, so once leaked, it remains permanently at risk. The security of iris identity information urgently needs to be addressed.

[0006] Therefore, a method, device and storage medium for iris photography are needed to at least partially solve the above technical problems. Summary of the Invention

[0007] In view of this, embodiments of the present invention provide an iris photography method, apparatus, and storage medium to solve at least one of the problems in the prior art.

[0008] In a first aspect, an embodiment of the present invention provides an iris photography method, the photography method comprising:

[0009] Inputting the target original iris image to be processed into a pre-trained first image generation model to obtain a transformed target intermediate iris image; the first image generation model is used to transform the target original iris image and output a target intermediate iris image that meets a set similarity threshold compared to the target original iris image;

[0010] Among them, the first image generation model is obtained by learning the image features extracted from the iris images after adjusting the size, direction, aspect ratio, and / or color depth of the patches based on multiple original iris image samples through a first deep learning model, and forming an intermediate iris image that can output an intermediate iris image satisfying a set similarity threshold compared to the corresponding original iris image sample.

[0011] Input the target intermediate iris image into a pre-trained second image generation model to obtain a target iris portrait image according to predetermined training requirements; the second image generation model is used to convert the target intermediate iris image into the style of the target image to be fused and output a target iris portrait image according to predetermined training requirements.

[0012] In a second aspect, an embodiment of the present invention further provides an iris portrait device, where the portrait device includes:

[0013] A memory for storing computer-executable instructions;

[0014] A processor for implementing the portrait method of the above technical solution when executing the computer-executable instructions stored in the memory.

[0015] In a third aspect, an embodiment of the present invention further provides a storage medium storing computer instructions, where the computer instructions are used to cause the computer to execute the portrait method of the above technical solution.

[0016] According to the portrait method of the embodiment of the present invention, through a pre-trained first image generation model, when the target original iris image to be processed is input into the first image generation model, a target intermediate iris image can be obtained, which has made certain modifications to the patch information of the target original iris image and can retain the main biological information of the target original iris image by additional conditions that meet the set similarity threshold, realizing iris information security protection and protecting the user's personal privacy; then, through a pre-trained second image generation model, intelligent stylization of the target intermediate iris image is realized, and an iris portrait image is obtained, finally realizing the dual requirements of iris image information security protection and stylization.

[0017] The additional advantages, objectives, and features of the present invention will be partially described below, and will become partially apparent to those of ordinary skill in the art after studying the following text, or can be learned from the practice of the present invention. The objectives and other advantages of the present invention can be achieved and obtained by the structures specifically pointed out in the specification and the drawings.

[0018] Those skilled in the art will understand that the objectives and advantages achievable by the present invention are not limited to those specifically described above, and the above and other objectives achievable by the present invention will be more clearly understood from the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The drawings described herein are provided to further understand the present invention, form a part of this application, and do not limit the present invention. The components in the drawings are not drawn to scale, but only to illustrate the principles of the present invention. For the convenience of showing and describing some parts of the present invention, the corresponding parts in the drawings may be enlarged, that is, they may become larger relative to other components in the exemplary device actually manufactured according to the present invention. In the drawings:

[0020] Figure 1 is a flowchart of a photo-taking method according to an embodiment of the present invention;

[0021] Figure 2 is another flowchart of a photo-taking method according to an embodiment of the present invention

[0022] Figure 3 is a schematic diagram of a generator model in a first image generation model in a photo-taking method according to an embodiment of the present invention;

[0023] Figure 4 is a schematic diagram of a discriminator model in a first image generation model in a photo-taking method according to an embodiment of the present invention;

[0024] Figure 5 is a schematic diagram of an intermediate iris image sample (left image) and a to-be-fused image sample (right image) fused therewith in a photo-taking method according to an embodiment of the present invention;

[0025] Figure 6 is a schematic diagram of a photo-taking device according to an embodiment of the present invention;

[0026] Figure 7 is a schematic diagram of a photo-taking system according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0027] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in conjunction with the embodiments and the drawings. Herein, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but do not limit the present invention.

[0028] Herein, it should also be noted that, in order to avoid obscuring the present invention due to unnecessary details, only the structures and / or processing steps closely related to the solution according to the present invention are shown in the drawings, and other details less related to the present invention are omitted.

[0029] It should be emphasized that when the term "comprising / including" is used herein, it refers to the presence of features, elements, steps or components, but does not exclude the presence or addition of one or more other features, elements, steps or components.

[0030] Here, it should also be noted that if not otherwise specified, the term "connection" in this text can refer not only to direct connection, but also to indirect connection with an intermediate object.

[0031] In the following, embodiments of the present invention will be described with reference to the accompanying drawings. In the drawings, the same reference numerals represent the same or similar components, or the same or similar steps.

[0032] First, reference will be made to Figure 1 Describe an iris photo-taking method 100 according to an embodiment of the present application. As Figure 1 shown, the photo-taking method 100 may include step S120 and step S140.

[0033] In step S120, the target original iris image to be processed is input into a pre-trained first image generation model to obtain a transformed target intermediate iris image; the first image generation model is used to transform the target original iris image and output a target intermediate iris image that meets a set similarity threshold compared to the target original iris image;

[0034] Among them, the first image generation model is determined by learning the image features extracted from the iris images after adjusting the size, direction, length, width, thickness and / or color depth of the patches based on multiple original iris image samples through a first deep learning model, so as to output an intermediate iris image that meets a set similarity threshold compared to the corresponding original iris image sample.

[0035] In step S140, the target intermediate iris image is input into a pre-trained second image generation model to obtain a target iris photo-taking image according to a predetermined training requirement; the second image generation model is used to convert the style of the target intermediate iris image into the style of the target image to be fused and output a target iris photo-taking image according to a predetermined training requirement.

[0036] According to the photo-taking method 100 of the embodiment of the present application, through the first image generation model, the target original iris image to be processed is input into the first image generation model, and a target intermediate iris image is obtained, which has made certain modifications to the patch information of the target original iris image and can retain the main biological information of the target original iris image by additionally meeting the set similarity loss threshold, realizing iris information security protection; then through the second image generation model, the intelligent stylization of the target intermediate iris image is realized, and an iris photo-taking image is obtained, finally realizing the dual requirements of iris image information security protection and stylization.

[0037] Next, in order to describe the photocopying method of the present application in more detail, reference will be made to Figure 2 The iris photography method 200 according to another embodiment of the present application is described. Figure 2 As shown, the photography method 200 may include steps S210 to S290, which are specifically as follows:

[0038] In step S210, a first image generation model is constructed; the first image generation model is used to transform the original iris image and output an intermediate iris image that meets a set similarity threshold compared to the original iris image.

[0039] In step S230, a second image generation model is constructed; the second image generation model is used to fuse the intermediate iris image and the image to be fused with a certain image style, and output an iris portrait image that meets the set training requirements.

[0040] In step S250 , the target original iris image to be processed is input into the first image generation model to obtain a transformed target intermediate iris image.

[0041] In step S270, before inputting the target intermediate iris image obtained by the first image generation model into the second image generation model, it is necessary to determine the iris feature similarity with the corresponding target original iris image again to determine whether it meets the set similarity threshold.

[0042] In step S290, the target intermediate iris image that meets the set similarity threshold is input into the second image generation model to obtain a target iris photograph image according to the predetermined training requirements.

[0043] In an embodiment of the present application, a first image generation model is first constructed; the first image generation model is used to transform an original iris image and output an intermediate iris image that meets a predetermined similarity threshold with the original iris image. Next, a second image generation model is constructed; the second image generation model is used to fuse the intermediate iris image with an image to be fused having a certain image style, outputting an iris portrait image that meets predetermined training requirements. The target original iris image to be processed is then input into the first image generation model to obtain a transformed target intermediate iris image. Before inputting the target intermediate iris image transformed by the first image generation model into the second image generation model, the iris feature similarity of the target intermediate iris image compared to the corresponding target original iris image is again determined to determine whether it meets a predetermined similarity threshold. Finally, the target intermediate iris image that meets the predetermined similarity threshold is input into the second image generation model to obtain a target iris portrait image that meets predetermined training requirements.

[0044] As can be seen from the above description of the process, according to the photo-taking method 200 of the embodiments of the present application, by constructing a first image generation model that meets the requirements, the transformation of the target original iris image can be realized intelligently, and the iris information security protection and the retention of the main biological information of the target original iris image can be achieved. Then, by means of constructing a second image generation model that meets the requirements and determining the iris feature similarity compared to the corresponding target original iris image again before the target intermediate iris image is input into the second image generation model, the dual requirements of iris image information security protection and stylization are finally realized.

[0045] Among them, in Figure 2 Steps S210 to S290 are shown to be arranged in sequence one after another, which is only an example. It can be understood that the order between step S210 and step S230 is not limited.

[0046] Next, in combination with Figure 1 and Figure 2 The content of each of the above steps of the photo-taking method 200 according to the embodiments of the present application will be specifically described.

[0047] In the embodiments of the present application, in step S210, a first image generation model is constructed; the first image generation model is used to transform the original iris image and output an intermediate iris image that meets a set similarity threshold compared to the original iris image.

[0048] Specifically, the steps of constructing the first image generation model may include:

[0049] Step S211, adjust the patches of multiple acquired original iris image samples in terms of size, direction, length, width, thickness, and / or color depth. Among them, the original iris image samples can be obtained by existing technical means such as photographing. For example, the size, direction, length, width, thickness, and color depth of the patches of the original iris image samples can be adjusted, or some of them can be adjusted. The patches may include spots, stripes, and wrinkles, etc. The adjusted iris image looks similar to the original iris image sample, but from the perspective of features, it is different. Therefore, the subsequent generated intermediate iris image cannot be successfully recognized with the original iris image sample.

[0050] Step S212, determine the iris feature similarity of the adjusted iris image compared to the corresponding original iris image sample. In the embodiments of the present invention, the determination (calculation) of the iris feature similarity index adopts existing technologies, and it is not an innovation point of the present invention, so it will not be elaborated here.

[0051] Step S213, perform image feature extraction on the adjusted iris image.

[0052] Among them, image feature extraction mainly extracts the texture information of the iris, including the information of the patches themselves, such as the shape and size of the patches, the depth of color, the position of the patches, and the number of patches; and also includes the relative information of the patches, such as the contrast with the surrounding area, the relative position between patches, the position of the patches relative to the pupil, etc.

[0053] Step S214, perform learning on the image features with different iris feature similarities extracted until an intermediate iris image that can output a similarity loss threshold meeting the set requirements compared with the corresponding original iris image sample is formed.

[0054] Specifically, among them, the first deep learning model can adopt a Generative Adversarial Network (GAN for short) and be trained with a large number of original iris image samples. However, for the convenience of training, reducing the amount of training data, and improving the training efficiency, in this embodiment, the original iris image samples are square iris images with the iris center as the center and a side length of 1.1 - 1.2 times the iris diameter. For example, with a side length of 1.1 times the iris diameter, rather than complete and large-sized original iris images.

[0055] The generative adversarial network consists of two main parts: a generator and a discriminator. The purpose of the generator is to generate as realistic data as possible, while the task of the discriminator is to distinguish between real data and the data generated by the generator.

[0056] The training of the first image generation model is divided into discriminator training and generator training. Both the discriminator and the generator are multi-layer neural networks. The discriminator gives a probability value indicating the likelihood that the sample is real data. The generator receives an iris image, passes through a multi-layer neural network, and outputs a sample similar to the real data distribution.

[0057] Generator training: After the generator receives an adjusted iris image, it extracts image features, passes through multiple downsampling convolutional units, upsampling convolutional units, and feature splicing operations of the corresponding units to generate fake samples, as shown in Figure 3 . The downsampling convolutional unit includes: a pooling layer, a convolutional layer, a BatchNorm layer, and an activation layer. The upsampling convolutional unit includes: an upsampling layer, a convolutional layer, a BatchNorm layer, and an activation layer. Usually, 3 - 5 groups of convolutional unit pairs are used. One downsampling convolutional unit and one upsampling convolutional unit form a group of convolutional unit pairs. The downsampling convolutional unit is used to capture the overall features, and the upsampling convolutional unit is used to restore the detailed information of the iris texture. The goal of the generator is to deceive the discriminator into misjudging the generated fake samples as real samples. Therefore, the loss calculation of the generator is based on the discriminator's judgment result for the fake samples, and the model parameters of the generator are updated by backpropagation through the discriminator's result, so that the samples generated by the generator meet the requirements.

[0058] Discriminator Training: The discriminator model consists of multiple fully connected units. Each fully connected unit includes a fully connected layer and an activation layer. Usually, 3 - 6 fully connected units are used. See Figure 4 . During each round of network training, a batch of samples is randomly drawn from the iris image dataset. At the same time, the generator generates a batch of fake samples of the same quantity. The real samples and the fake samples are respectively input into the discriminator, and the discriminator tries to distinguish between them. The parameters of the discriminator model are updated through backpropagation, enabling the discriminator to better distinguish between real samples and fake samples, and at the same time generating data that is similar to but different from the input data (i.e., the input data and the generated data cannot be successfully recognized).

[0059] Similarity Loss Calculation Method: The score range of the iris feature similarity is 1 - 100. The calculation of this iris feature similarity is an existing algorithm. 60 is used as the standard for determining whether the irises belong to the same person. That is, when the score is greater than 60, it is determined that the irises belong to the same person. The higher the score, the closer the two irises are. In practice, it is hoped that the generated data is similar to the original data, but there will be some changes in the iris texture to protect biological information. Therefore, in the design of the similarity loss calculation, it is required that the iris feature similarity score of the generated data compared to the original data is preferably between 50 - 60. At this time, loss_sim is the smallest, and the other scores should be larger than this score range. First, the iris feature similarity is normalized to the range of 0 - 1 by dividing it by 100. Among them, the similarity loss loss_sim is calculated using the following formula:

[0060]

[0061] Among them, y' represents the normalized score of the iris feature similarity of the intermediate iris image compared to the corresponding original iris image sample, and this score takes values in the range of 0 - 1.

[0062] Standard for Ending Training: When the losses of the discriminator and the generator are basically stable and do not decrease, that is, the set similarity loss threshold is met. Among them, meeting the set similarity loss threshold means that loss_sim ≤ x', where 0 ≤ x' ≤ 0.15. For example, when the similarity loss loss_sim is 0, the training ends.

[0063] In the embodiment of the present application, in step S230, a second image generation model is constructed; the second image generation model is used to fuse the intermediate iris image and the to - be - fused image with a certain image style, and output an iris portrait image that meets the set training requirements.

[0064] Specifically, the steps for constructing the second image generation model include:

[0065] Step S231: Extract image features from the obtained multiple intermediate iris image samples and multiple original iris image samples, and extract image styles from the obtained multiple images to be fused.

[0066] The intermediate iris image samples and the original iris image samples are iris images that have undergone positioning processing and only the iris region remains. Among them, the intermediate iris image samples can be obtained in large quantities using the intermediate iris images generated by the first image generation model in the early stage. The images to be fused are images to be fused with the same shape and size as the intermediate iris image samples. For example, for a painter's picture collection, or landscape photos, space pictures, etc. with a unified style, randomly select a circular part with the same size. See Figure 5 .

[0067] Among them, extracting image features from the intermediate iris image samples and the original iris image samples includes extracting the shape and size of the patches, the depth of color, the position of the patches, the number of patches, the contrast with the surrounding area, the relative position between the patches, and the position information of the patches relative to the pupil.

[0068] Step S232: Use the second deep learning model to learn the extracted different image features and image styles until an iris portrait image that can output and meet the set training requirements compared to the corresponding intermediate iris image samples is formed.

[0069] Specifically, the iris portrait should not only ensure the information security of the iris but also meet the user's pursuit of iris beauty. For example, the second image generation model can be trained using the CycleGAN network.

[0070] The CycleGAN model has two generators and two discriminators. The role of the generator is to generate stylized data as much as possible to deceive the discriminator, and the role of the discriminator is to determine as much as possible whether the image is a generated image. Two sets of data are required for training, one is the intermediate iris image sample dataset (X domain), and the other is the image sample dataset to be fused (Y domain). The two generators are responsible for converting the image into another style of image. For example, generator G1 is responsible for converting the iris image into a style image, and generator G2 is responsible for converting the style image back into an iris image. The two discriminators respectively determine whether it is a real image. For example, D1 is responsible for determining whether the data sent to discriminator D1 is generated by G2 or real iris data, and D2 is responsible for determining whether the data sent to discriminator D2 is generated by G1 or real stylized image.

[0071] The second image generation model not only transforms the image style but also preserves the information of the original image as much as possible. During training, the model requires the images in the X domain to pass through the generator G1 and then through the generator G2, continuously updating the parameters of the generator and the discriminator. Eventually, the generator can learn the mapping relationship between different image domains and achieve the transformation of the image style. The output result should be as close as possible to the X domain, and vice versa.

[0072] Among them, setting the training requirement can be to reach the set number of training rounds. For example, the set number of training rounds is preset to at least 1000 rounds. And after the number of training rounds is reached, the model can be stored every certain number of rounds, so that multiple second image generation models (iris stylization models) can be obtained, providing customers with stylization options, and subsequently different styles of iris portrait images can be obtained.

[0073] In the embodiment of the present application, in step S250, the target original iris image to be processed is input into the trained first image generation model to obtain the transformed target intermediate iris image.

[0074] Specifically, after obtaining the trained first image generation model through step S210, at this time, the target original iris image to be processed can be input into the first image generation model to obtain the transformed target intermediate iris image that meets the set similarity threshold compared to the target original iris image.

[0075] Similarly, there is the following relationship between the set similarity threshold and the set similarity loss threshold during model training: the set similarity loss threshold satisfies loss_sim ≤ x, where 0 ≤ x ≤ 0.15. Among them, the similarity loss loss_sim is calculated using the following formula:

[0076]

[0077] Among them, y represents the score after normalizing the set similarity threshold, and this score takes values from 0 to 1.

[0078] In the embodiment of the present application, before the target intermediate iris image obtained by the transformation of the first image generation model in step S270 is input into the second image generation model, it is also necessary to determine again the iris feature similarity compared to the corresponding target original iris image and judge whether it meets the set similarity threshold.

[0079] Specifically, the trained first image generation model can ensure that the vast majority of the generated target intermediate iris images cannot be successfully matched with the target original iris image, but it is not 100% guaranteed. To ensure 100% that iris information is not leaked and protect the privacy of users, an additional step of calculating iris feature similarity is required during actual use.

[0080] That is, the iris feature similarity between the target intermediate iris image and the corresponding target original iris image is determined again by the existing algorithm, and it is determined whether the set similarity threshold is satisfied.

[0081] If the set similarity threshold is not satisfied, the first image generation model is re-input until it is changed to satisfy the set similarity threshold compared with the corresponding target original iris image.

[0082] In an embodiment of the present application, in step S290, the target intermediate iris image that satisfies the set similarity threshold is input into the trained second image generation model. The second image generation model converts the target intermediate iris image into the style of the target image to be fused, and obtains the target iris portrait image according to the predetermined training requirements. Among them, the target image to be fused is a certain image to be fused with a certain image style during model training.

[0083] For example, the predetermined training requirement is that the second image generation model is trained for 1300 rounds to obtain the target iris portrait image that meets its own requirements.

[0084] Based on the above description, according to the portrait method 200 of the embodiments of the present application, the iris biometric information security can be protected while realizing intelligent iris portrait.

[0085] Reference Figure 6 , the portrait device 300 for implementing the portrait method according to the embodiments of the present application includes a processor 310 and a memory 320. The portrait device 300 may include one or more processors 310 and one or more memories 320. The memory 320 stores an executable program run by the processor 310. When the executable program is run by the processor 310, the processor 310 executes the portrait method 100 or 200 according to the embodiments of the present application described above.

[0086] The processor 310 may be a central processing unit (CPU) or other forms of processing units with data processing capabilities and / or instruction execution capabilities.

[0087] The memory 320 may include one or more computer program products, and the computer program products may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory, etc. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage media, and the processor 310 may run the program instructions to implement the client functions (implemented by the processor) in the embodiments of the present application described herein and / or other desired functions. Various application programs and various data may also be stored in the computer-readable storage media, such as various data used and / or generated by the application programs, etc.

[0088] The photo device 300 may further include an input device and an output device, and these components are interconnected through a bus system and / or other forms of connection mechanisms. It should be noted that Figure 6 The components and structures of the photo device 300 shown are only exemplary and not restrictive. According to needs, the photo device 300 may also have other components and structures.

[0089] The input device may be a device used by a user to input instructions, and may include one or more of a keyboard, a mouse, a microphone, a touch screen, etc. In addition, the input device may also be any interface for receiving information.

[0090] The output device may output various information (such as images or sounds) to the outside (such as a user), and may include one or more of a display, a speaker, etc. In addition, the output device may also be any other device with an output function.

[0091] Exemplarily, the exemplary photo device 300 for implementing the photo method 100 according to the embodiments of the present application may be applied to electronic devices such as terminal devices (such as mobile phones), tablet computers, laptop computers, ultra-mobile personal computers (UMPCs), handheld computers, netbooks, personal digital assistants (PDAs), wearable devices (such as smart watches, smart glasses or smart helmets, etc.), augmented reality (AR), virtual reality (VR) devices, smart home devices, in-vehicle computers, etc. The embodiments of the present application do not make any restrictions on this.

[0092] Those skilled in the art can understand the specific operations of the photo-taking device 300 for implementing the photo-taking method 200 according to the embodiments of the present application in combination with the content described above. For the sake of brevity, the specific details are not described here again, and only some main operations of the processor 310 are described.

[0093] In an embodiment of the present application, when the executable program is run by the processor 310, the processor 310 is caused to perform the following steps: constructing a first image generation model; the first image generation model is used to transform the original iris image and output an intermediate iris image that meets a set similarity threshold compared to the original iris image. Constructing a second image generation model; the second image generation model is used to fuse the intermediate iris image and a to-be-fused image with a certain image style, and output an iris photo-taking image that meets the set training requirements. Inputting the target original iris image to be processed into the trained first image generation model to obtain a transformed target intermediate iris image. Before the target intermediate iris image obtained by the transformation of the first image generation model is input into the second image generation model, it is also necessary to determine again the iris feature similarity compared to the corresponding target original iris image, and determine whether it meets the set similarity threshold. Inputting the target intermediate iris image that meets the set similarity threshold into the trained second image generation model to obtain a target iris photo-taking image according to the predetermined training requirements.

[0094] The above exemplarily shows the photo-taking method 200 according to the embodiments of the present application. The following combines Figure 7 to describe the photo-taking system 400 provided by another aspect of the embodiments of the present application.

[0095] Referring to Figure 7 to describe an exemplary photo-taking system 400 for implementing the photo-taking method of the embodiments of the present application. The photo-taking system 400 may include a first construction module 410, a second construction module 420, a first acquisition module 430, a re-determination module 440, and a second acquisition module 450. Among them:

[0096] The first construction module 410 is used to: construct a first image generation model; the first image generation model is used to transform the original iris image and output an intermediate iris image that meets a set similarity threshold compared to the original iris image.

[0097] The second construction module 420 is used to: construct a second image generation model; the second image generation model is used to fuse the intermediate iris image and a to-be-fused image with a certain image style, and output an iris photo-taking image that meets the set training requirements.

[0098] The first acquisition module 430 is used to: input the target original iris image to be processed into the trained first image generation model to obtain a transformed target intermediate iris image.

[0099] The re-determination module 440 is configured to: before the target intermediate iris image obtained by transformation through the first image generation model is input into the second image generation model, it is also necessary to re-determine the iris feature similarity with respect to the corresponding target original iris image, and determine whether the set similarity threshold is satisfied.

[0100] The second acquisition module 450 is configured to: input the target intermediate iris image that satisfies the set similarity threshold into the trained second image generation model, and acquire a target iris portrait image according to the predetermined training requirements.

[0101] The portrait system 400 proposed in the embodiment of the present invention can protect the iris information security on the basis of completing the stylized portrait of the iris image.

[0102] In addition, according to the embodiment of the present application, the present application also provides a storage medium, on which a computer program is stored, and when the computer program is run by a processor, it is used to execute the corresponding steps of the portrait method 100 or 200 of the embodiment of the present application. The storage medium may include, for example, a memory card of a smart phone, a storage component of a tablet computer, a hard disk of a personal computer, a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a portable compact disc read-only memory (CD-ROM), a USB memory, or any combination of the above storage media. The computer-readable storage medium may be any combination of one or more computer-readable storage media.

[0103] In addition, according to the embodiment of the present application, the present application also provides a computer program product, including computer instructions, and when the computer instructions are executed by a processor, the steps of the portrait method of the embodiment of the present application are implemented.

[0104] Although example embodiments have been described herein with reference to the accompanying drawings, it should be understood that the above example embodiments are merely exemplary and are not intended to limit the scope of the present application thereto. Those of ordinary skill in the art can make various changes and modifications therein without departing from the scope and spirit of the present application. All such changes and modifications are intended to be included within the scope of the present application as claimed in the appended claims.

[0105] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the present application.

[0106] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed.

[0107] In addition, those skilled in the art can understand that although some embodiments described herein include certain features included in other embodiments rather than other features, the combination of the features of different embodiments means that it is within the scope of the present application and forms different embodiments. For example, in the claims, any one of the claimed embodiments can be used in any combination.

[0108] It should be noted that the above embodiments illustrate the present application rather than limit the present application, and those skilled in the art can design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in the claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present application can be implemented by means of hardware including several different elements and by means of a properly programmed computer. In a unit claim listing several devices, several of these devices can be embodied by the same item of hardware. The use of the words first, second, and third, etc. does not denote any order. These words can be interpreted as names.

[0109] As described above, this is only the specific implementation manner of the present application or the description of the specific implementation manner. The protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in the present application, and all should be covered by the protection scope of the present application. The protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. An iris photography method, characterized in that, The described photo-taking method includes: Inputting a target original iris image to be processed into a pre-trained first image generation model to obtain a transformed target intermediate iris image; the first image generation model is used to transform the target original iris image and output a target intermediate iris image that meets a set similarity threshold compared to the target original iris image; Among them, the first image generation model is determined by learning the image features extracted from the iris images after adjusting the size, direction, aspect ratio, and / or color depth of the patches based on multiple original iris image samples through a first deep learning model, so as to form an intermediate iris image that can output a middle iris image that meets the set similarity threshold compared to the corresponding original iris image sample; Inputting the target intermediate iris image into a pre-trained second image generation model to obtain a target iris photo-taking image according to a predetermined training requirement; the second image generation model is used to convert the target intermediate iris image into the style of the target image to be fused and output a target iris photo-taking image according to a predetermined training requirement.

2. The photo-taking method according to claim 1, characterized in that, It also includes constructing a first image generation model; the steps of constructing the first image generation model include: Respectively adjusting the size, direction, aspect ratio, and / or color depth of the patches of multiple obtained original iris image samples, determining the iris feature similarity of the adjusted iris image compared to the corresponding original iris image sample; extracting image features from the adjusted iris image, and learning the image features with different iris feature similarities through a first deep learning model until an intermediate iris image that can output a middle iris image that meets the set similarity loss threshold compared to the corresponding original iris image sample is formed; Among them, extracting image features from the adjusted iris image includes extracting the shape and size of the patches, the depth of color, the position of the patches, the number of patches, the contrast with the surrounding area, the relative position between patches, and the position information of the patches relative to the pupil.

3. The photo-taking method according to claim 2, wherein The following relationship exists between the set similarity threshold and the set similarity loss threshold in model training: The set similarity loss threshold satisfies loss_sim≤x, 0≤x≤0.15; where the similarity loss loss_sim is calculated using the following formula: Among them, y represents the normalized score of the set similarity threshold, and this score takes values from 0 to 1.

4. The photo-taking method according to claim 1, characterized in that, It also includes constructing a second image generation model; the steps of constructing the second image generation model include: Extracting image features from multiple obtained intermediate iris image samples and multiple original iris image samples, and extracting image styles from multiple image samples to be fused with a certain image style obtained, and learning the different image features and image styles through a second deep learning model until an iris photo-taking image that can output a middle iris image that meets the set training requirement compared to the corresponding intermediate iris image sample is formed; Among them, extracting image features from the intermediate iris image samples and the original iris image samples includes extracting the shape and size of the patches, the depth of color, the position of the patches, the number of patches, the contrast with the surrounding area, the relative position between patches, and the position information of the patches relative to the pupil.

5. The photo-taking method according to claim 4, characterized in that, The middle iris image sample is a middle iris image that has undergone positioning processing and only the iris region remains. The image sample to be fused is an image to be fused with the same shape and size as the middle iris image sample.

6. The photo-taking method according to claim 1, characterized in that, Before the target middle iris image obtained by transformation through the first image generation model is input into the second image generation model, it is also necessary to determine again the iris feature similarity compared to the corresponding target original iris image. If the set similarity threshold is not met, it is re-input into the first image generation model until the transformation meets the set similarity threshold compared to the corresponding target original iris image.

7. The photo-taking method according to claim 1, wherein The original iris image sample is a square iris image centered on the iris center with a side length of 1.1 - 1.2 times the iris diameter.

8. The photo-taking method according to claim 4, characterized in that, The set training requirement is to reach the set number of training rounds.

9. An iris photography device, characterized in that, The photo-taking device includes: A memory for storing computer-executable instructions; A processor for implementing the photo-taking method according to any one of claims 1 to 8 when executing the computer-executable instructions stored in the memory.

10. A storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to execute the photo-taking method according to any one of claims 1 to 8.

Images