Two-dimensional code image super-resolution reconstruction and identification method and system, electronic device and medium

Abstract

The application provides a two-dimensional code image super-resolution reconstruction and recognition method and system, an electronic device and a medium. The two-dimensional code image super-resolution reconstruction method comprises: respectively performing feature encoding on multiple frames of two-dimensional code images to be super-resolution reconstructed to obtain corresponding feature information of each frame of two-dimensional code image; fusing the feature information of the two-dimensional code images to obtain fused feature information; performing feature reconstruction based on the fused feature information to obtain a super-resolution reconstructed feature map; and performing feature decoding based on the super-resolution reconstructed feature map to obtain a two-dimensional code image after super-resolution reconstruction. The image quality of the reconstructed two-dimensional code image is greatly improved, and the recognition rate of two-dimensional code recognition is improved.

Description

Technical Field

[0001] This application relates to the field of image processing technology, and in particular to a method and system for super-resolution reconstruction of QR code images, a method and system for QR code image recognition, electronic devices, and computer-readable storage media. Background Technology

[0002] QR code technology is increasingly widely used in industrial and consumer sectors. By scanning a QR code with a handheld device, users can obtain detailed product information. QR codes improve the transparency of product information, enabling more efficient circulation, exchange, and verification of product information, thereby enhancing the efficiency of society as a whole.

[0003] QR codes are affected by many physical factors during scanning, such as ambient light, scanning distance, scanning angle, incomplete information, and QR code image quality, which can lead to a decrease in recognition rate. Among these factors, QR code image quality has the greatest impact on the recognition rate.

[0004] Image interpolation is the process of generating a high-resolution image from a low-resolution image within a model-based framework, in order to recover lost information in the image. Commonly used image interpolation algorithms include nearest neighbor interpolation, bilinear interpolation, bilinear cubic interpolation, and Lanczos interpolation. However, image interpolation algorithms often result in stepped jagged edges after image processing, which clearly fails to significantly improve image quality. Therefore, image interpolation cannot meet the quality requirements of QR code images. Summary of the Invention

[0005] This application provides a method and system for super-resolution reconstruction of QR code images, a method and system for QR code image recognition, an electronic device, and a computer-readable storage medium, which can solve the above-mentioned shortcomings in the prior art.

[0006] Firstly, this application provides a method for super-resolution reconstruction of QR code images. The method includes: performing feature encoding on the QR code images to be super-resolution reconstructed to obtain feature information corresponding to the QR code images; fusing the feature information to obtain fused feature information; performing feature reconstruction based on the fused feature information to obtain a super-resolution reconstructed feature map; and performing feature decoding based on the super-resolution reconstructed feature map to obtain a super-resolution reconstructed QR code image. In this application, texture features (i.e., feature information) of the QR code image are extracted, and the texture features of the QR code are fused into a more reasonable fused feature for generating the super-resolution image. While maintaining the feature code information of the QR code unchanged, a clear reconstructed QR code image is obtained, greatly improving the image quality of the QR code image.

[0007] In one implementation of the first aspect, feature encoding of the QR code image to be super-reconstructed to obtain feature information corresponding to the QR code image includes: extracting feature information corresponding to the QR code image through a feature encoder network, wherein the feature information includes QR code image features, self-similarity features, and edge features. In this implementation, extracting feature information across multiple dimensions can provide more texture features, which helps improve the quality of the reconstructed QR code image.

[0008] In one implementation of the first aspect, extracting feature information corresponding to the QR code image through a feature encoder network includes: extracting QR code image features from the QR code image, the QR code image features including multiple levels of features, including primary features, intermediate features and high-level features; extracting self-similarity features from the QR code image, the self-similarity features being used to reconstruct QR code image details; and extracting edge features from the QR code image.

[0009] In one implementation of the first aspect, feature encoding of the QR code image to be super-reconstructed to obtain feature information corresponding to the QR code image includes: performing feature encoding on the multiple frames of QR code images to be super-reconstructed to obtain feature information corresponding to each frame of QR code image.

[0010] In one implementation of the first aspect, fusing the feature information to obtain fused feature information includes: calculating the feature information through a feature fusion network to fuse the feature information corresponding to the multidimensional features into pixel-level weights as the fused feature information using convolution operations.

[0011] In one implementation of the first aspect, fusing the feature information of the QR code image to obtain fused feature information includes: aligning multiple frames of the QR code image; and fusing the feature information corresponding to the aligned QR code image into a single feature to obtain the fused feature information.

[0012] In one implementation of the first aspect, aligning multiple frames of the QR code images includes: aligning the multiple frames of the QR code images based on eye-shaped features and / or dot-shaped features. In this implementation, using eye-shaped features and / or dot-shaped features for image alignment helps reduce algorithm complexity and improve computation speed.

[0013] In one implementation of the first aspect, performing feature reconstruction based on the fused feature information to obtain a super-resolution reconstruction feature map includes: using convolution operations through a feature reconstruction network to calculate the super-resolution reconstruction feature map based on the feature information.

[0014] In one implementation of the first aspect, obtaining a super-reconstructed QR code image by performing feature decoding based on the super-reconstruction feature map includes: obtaining super-reconstruction feature decoding information based on the super-reconstruction feature map through a feature decoder network; and calculating the super-reconstructed QR code image based on the super-reconstruction feature decoding information and the QR code image.

[0015] Secondly, this application provides a QR code image recognition method. The QR code image recognition method includes: acquiring a collected QR code image, and using the QR code image super-resolution reconstruction method according to the first aspect of this application to obtain a super-resolution reconstructed QR code image; and recognizing the feature code of the super-resolution reconstructed QR code image. In this application, the image preprocessing for super-resolution reconstruction of the QR code image transforms a low-resolution QR code image (blurry QR code image) into a high-resolution QR code image (clear QR code image) while maintaining the feature code information of the QR code unchanged, thereby improving the QR code recognition rate.

[0016] Thirdly, this application provides a QR code image super-resolution reconstruction system. The QR code image super-resolution reconstruction system includes: a feature encoder network configured to encode features of the QR code image to be super-resolution reconstructed to obtain feature information corresponding to the QR code image; a feature fusion network configured to fuse the feature information to obtain fused feature information; a feature reconstruction network configured to perform feature reconstruction based on the fused feature information to obtain a super-resolution reconstructed feature map; and a feature decoder network configured to perform feature decoding based on the super-resolution reconstructed feature map to obtain a super-resolution reconstructed QR code image. In this application, the feature encoder network extracts the texture features (i.e., feature information) of the QR code image, the feature fusion network fuses the texture features of the QR code into a more reasonable fused feature for generating the super-resolution image, and the feature reconstruction network and feature decoder network perform feature reconstruction and decoding. While maintaining the feature code information of the QR code unchanged, a clear reconstructed QR code image is obtained, greatly improving the image quality of the QR code image.

[0017] In one implementation of the third aspect, the feature encoder network, feature fusion network, feature reconstruction network, and feature decoder network are sequentially cascaded and integrated into a deep learning model.

[0018] Fourthly, this application provides a QR code image recognition system. The QR code image recognition system includes: a super-resolution reconstruction module configured to: acquire a collected QR code image; perform feature encoding on the QR code image to obtain feature information corresponding to the QR code image; fuse the feature information to obtain fused feature information; perform feature reconstruction based on the fused feature information to obtain a super-resolution reconstructed feature map; and perform feature decoding based on the super-resolution reconstructed feature map to obtain a super-resolution reconstructed QR code image; and a recognition module configured to recognize the feature code of the super-resolution reconstructed QR code image. In this application, the super-resolution reconstruction module first performs image preprocessing on the QR code image for super-resolution reconstruction, processing the low-resolution QR code image (blurred QR code image) into a high-resolution QR code image (clear QR code image) while maintaining the feature code information of the QR code unchanged. Then, a recognition model is used to recognize the reconstructed QR code image, improving the QR code recognition rate.

[0019] Fifthly, this application provides an electronic device. The electronic device includes: a memory configured to store a computer program; and a processor configured to invoke the computer program to perform the QR code image super-resolution reconstruction method according to the first aspect of this application and / or the QR code image recognition method according to the second aspect of this application.

[0020] Sixthly, this application provides a computer-readable storage medium having a computer program stored thereon. The computer program is executed to implement the QR code image super-resolution reconstruction method according to the first aspect of this application and / or the QR code image recognition method according to the second aspect of this application.

[0021] In the QR code image super-resolution reconstruction method and system, QR code image recognition method and system, electronic device and computer-readable storage medium according to this application, the QR code image is first super-resolution reconstructed to process the low-resolution QR code image (blurred QR code image) into a high-resolution QR code image (clear QR code image) while maintaining the feature code information of the QR code unchanged. Then, the reconstructed QR code image is recognized by the recognition model, thereby improving the recognition rate of the QR code. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of a scenario where the QR code image super-resolution reconstruction and recognition method according to an embodiment of this application is applied to a barcode scanner.

[0023] Figure 2 This is a flowchart of a QR code image super-resolution reconstruction method according to an embodiment of this application.

[0024] Figure 3This is a flowchart of a QR code image super-resolution reconstruction method according to another embodiment of this application.

[0025] Figure 4 This is a structural block diagram of the QR code image super-resolution reconstruction system according to an embodiment of this application.

[0026] Figure 5 This is a schematic diagram showing QR code images with different eyelet features.

[0027] Figure 6 This is a schematic diagram showing QR code images with different code dot features.

[0028] Figure 7 This is a flowchart of a QR code image recognition method according to an embodiment of this application.

[0029] Figure 8 This is a structural block diagram of a QR code image recognition system according to an embodiment of this application.

[0030] Figure 9 This is a structural block diagram of an electronic device according to an embodiment of this application. Detailed Implementation

[0031] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, unless otherwise specified, the following embodiments and features in the embodiments can be combined with each other.

[0032] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of this application. The illustrations only show the components related to this application and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0033] Existing technical solutions employ image interpolation algorithms to generate high-resolution images from low-resolution images. However, image interpolation algorithms often result in jagged, stepped edges, failing to adequately improve image quality. Therefore, image interpolation algorithms cannot meet the quality requirements of QR code images.

[0034] To address at least the aforementioned problems, this invention provides a super-resolution reconstruction scheme for QR code images. According to embodiments of the invention, a feature encoder network extracts multi-dimensional feature information from a QR code image; a feature fusion network fuses this feature information into pixel-level weighted fused feature information; a feature reconstruction network calculates a super-resolution reconstructed feature map of the QR code based on this fused feature information; and a feature decoder network generates super-resolution reconstructed feature decoding information based on this super-resolution reconstructed feature map, thereby calculating the super-resolution reconstructed QR code image. In this way, while maintaining the QR code's feature code information unchanged, a low-resolution (blurred) QR code image is processed into a high-resolution (clear) QR code image, improving the QR code recognition rate.

[0035] The following description, with reference to exemplary embodiments and accompanying drawings, describes the QR code image super-resolution reconstruction method and system, QR code image recognition method and system, electronic device, and computer-readable storage medium according to the present invention. The application environments of the present invention include, but are not limited to, electronic devices with image acquisition and communication functions such as barcode scanners, smartphones, and tablets. The following description will use a QR code scanner as an example.

[0036] Figure 1 This is a schematic diagram illustrating a scenario where the QR code image super-resolution reconstruction and recognition method according to an embodiment of this application is applied to a barcode scanner. Figure 1 As shown, the QR code recognition system includes a scanner 11, a terminal device 12, and a QR code medium 13. A QR code 14 is displayed on the QR code medium 13. The scanner 11 is communicatively connected to the terminal device 12.

[0037] Terminal device 12 includes PCs, laptops, ultra-mobile personal computers (UMPCs), netbooks, personal digital assistants (PDAs), and other terminal devices. It can also be a server and a service response system based on terminal artificial intelligence. This application embodiment does not limit the specific type of terminal device 12. The communication connection between the barcode scanner 11 and terminal device 12 includes wired and wireless communication connections. This application also does not limit the communication method between the barcode scanner 11 and terminal device 12. Figure 1 In the example, terminal device 12 is shown as a PC, and barcode scanner 11 communicates with terminal device 12 via wired communication.

[0038] The scanner 1 includes a light source emitter, an image sensor, a memory, and a processor. The memory stores a computer program for super-resolution reconstruction and recognition of QR code images, and the processor is configured to call the computer program to perform super-resolution reconstruction and recognition of QR code images. The QR code medium 13 includes a screen medium, a metal engraving medium, a thin-film reflective medium, etc., and then the QR code 14 is presented on the QR code medium 13 to form a screen QR code, a metal engraving QR code, or a thin-film reflective QR code. In this embodiment, the QR code medium 13 is a screen medium. In some embodiments, the screen medium includes, but is not limited to, the liquid crystal screen of an electronic device. This embodiment takes the liquid crystal screen of a smartphone as an example.

[0039] When performing QR code image recognition, a QR code image is displayed on the QR code medium 13, and the terminal device 12 controls the scanner 11 to scan the QR code image. The scanner 11 continuously acquires multiple frames of images during scanning. However, issues such as ambient light, scanning distance, scanning angle, or screen jitter when holding a smartphone can result in poor quality of the scanned QR code image, leading to poor image quality across multiple scanned frames. In this application, after acquiring multiple frames of QR code images, the scanner 11 first performs image super-resolution reconstruction, transforming the low-resolution QR code image (blurry QR code image) into a high-resolution QR code image (clear QR code image) while maintaining the QR code's feature code information. Then, a recognition model is used to recognize the reconstructed QR code image, improving the QR code recognition rate. In the process of super-resolution reconstruction of QR code images, the texture features (i.e. feature information) of multiple frames of QR code images are extracted, and the texture features of the QR codes are fused into more reasonable fusion features to generate super-resolution images. Then, feature reconstruction is performed based on the fusion features to obtain the feature map of super-resolution reconstruction. Finally, feature decoding is performed based on the feature map of super-resolution reconstruction to obtain the QR code image after super-resolution reconstruction.

[0040] The QR code image super-resolution reconstruction and recognition method provided in this application preferably employs an artificial intelligence algorithm based on deep learning. Deep learning is a type of machine learning, and machine learning is an essential path to achieving artificial intelligence. The concept of deep learning originates from the research of artificial neural networks; a multilayer perceptron with multiple hidden layers is a type of deep learning structure. Deep learning discovers distributed feature representations of data by combining low-level features to form more abstract high-level representations of attribute categories or features. The motivation for researching deep learning is to build neural networks that simulate the human brain's analytical learning process. It mimics the mechanisms of the human brain to interpret data, such as images, sounds, and text. Deep learning establishes a functional relationship from input to output by designing and building an appropriate number of neuron computing nodes and a multi-layered computational hierarchy, selecting suitable input and output layers, and through network learning and optimization. Although it cannot find a 100% accurate functional relationship between input and output, it can approximate the real-world correlation as closely as possible. Using a successfully trained network model, the automation requirements for complex task processing can be achieved.

[0041] The technical solutions in the embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0042] Figure 2 This is a flowchart of a QR code image super-resolution reconstruction method according to an embodiment of this application. Figure 2 As shown, the QR code image super-resolution reconstruction method in this embodiment includes the following steps S11 to S14.

[0043] In step S11, feature encoding is performed on the QR code images to be super-reconstructed to obtain feature information corresponding to the QR code images. In some embodiments, the QR code images are acquired by an image sensor.

[0044] In some embodiments, feature encoding of the QR code image to be super-reconstructed to obtain feature information corresponding to the QR code image includes: extracting feature information corresponding to the QR code image through a feature encoder network, wherein the feature information includes QR code image features, self-similarity features, and edge features. In some embodiments, extracting feature information corresponding to the QR code image through a feature encoder network includes: extracting the QR code image features from the QR code image, wherein the QR code image features include multiple levels of features, including primary features, intermediate features, and high-level features; extracting the self-similarity features from the QR code image, wherein the self-similarity features are used to reconstruct QR code image details; and extracting the edge features from the QR code image.

[0045] In step S12, the feature information of the QR code image is fused to obtain fused feature information.

[0046] In some embodiments, fusing the feature information to obtain fused feature information includes: calculating the feature information through a feature fusion network to fuse the feature information corresponding to the multidimensional features into pixel-level weights as the fused feature information using convolution operations.

[0047] In step S13, feature reconstruction is performed based on the fused feature information to obtain a super-resolution reconstructed feature map.

[0048] Step S13 employs a feature reconstruction network for feature fusion. The feature reconstruction network performs feature reconstruction based on the fused feature information to obtain a super-reconstructed feature map.

[0049] In some embodiments, performing feature reconstruction based on the fused feature information to obtain a super-resolution reconstruction feature map includes: using convolution operations through a feature reconstruction network to calculate the super-resolution reconstruction feature map based on the feature information.

[0050] In step S14, feature decoding is performed based on the super-reconstruction feature map to obtain the super-reconstructed QR code super-reconstruction image.

[0051] Step S14 uses a feature decoder network to perform feature decoding on the fully reconstructed feature map to obtain the super-reconstructed QR code image.

[0052] In some embodiments, obtaining a super-reconstructed QR code image by performing feature decoding based on the super-reconstruction feature map includes: obtaining super-reconstruction feature decoding information based on the super-reconstruction feature map through a feature decoder network; and calculating the super-reconstructed QR code image based on the super-reconstruction feature decoding information and the QR code image.

[0053] The above implementation can fully reconstruct a single frame of QR code image, process a low-resolution QR code image (blurry QR code image) into a high-resolution QR code image (clear QR code image), and maintain the QR code feature information unchanged.

[0054] Figure 3 This is a flowchart of a QR code image super-resolution reconstruction method according to an embodiment of this application. Figure 3 As shown, the QR code image super-resolution reconstruction method in this embodiment includes the following steps S11 to S14.

[0055] In step S11, feature encoding is performed on the QR code image to be super-reconstructed to obtain feature information corresponding to the QR code image.

[0056] In step S12, the feature information is fused to obtain fused feature information.

[0057] In step S13, feature reconstruction is performed based on the fused feature information to obtain a super-resolution reconstructed feature map.

[0058] In step S14, feature decoding is performed based on the super-reconstruction feature map to obtain the super-reconstructed QR code super-reconstruction image.

[0059] In this embodiment, steps S11 to S14 are the same as steps S11 to S14 in the QR code image super-resolution reconstruction method provided in the previous embodiment. However, unlike the above embodiment, the QR code image super-resolution reconstruction method in this embodiment can perform super-resolution reconstruction on multiple frames of QR code images.

[0060] Therefore, in step S11 of this embodiment, feature encoding of the QR code images to be super-reconstructed to obtain feature information corresponding to the QR code images includes feature encoding of multiple frames of QR code images to be super-reconstructed to obtain feature information corresponding to each frame of QR code images. The specific implementation method for feature encoding of each frame of QR code images to obtain feature information corresponding to the QR code images is the same as in the above embodiment, and will not be repeated here.

[0061] Furthermore, in step S12 of this embodiment, fusing the feature information to obtain fused feature information includes steps S121 and S122.

[0062] In step S121, multiple frames of the QR code images are aligned. In some embodiments, the multiple frames of the QR code images are aligned based on eye-shaped features and / or dot-shaped features.

[0063] In step S122, the feature information corresponding to the aligned QR code image is fused into a single feature to obtain the fused feature information. In some embodiments, obtaining the fused feature information includes: calculating the feature information through a feature fusion network to fuse the feature information corresponding to the multidimensional features into pixel-level weights as the fused feature information using convolution operations.

[0064] Step S12 of this implementation addresses feature fusion during super-resolution reconstruction of multi-frame QR code images. First, the multi-frame QR code images are aligned. Then, the corresponding feature information from the aligned QR code images is fused into a single feature. This involves fusing multiple dimensions of features from the multi-frame QR code images, such as QR code image features, self-similarity, and edge features, into a single feature. A feature fusion network is then used to calculate and obtain the fused feature information. Therefore, this application fuses the feature information from multiple frames of QR code images to form a more reasonable fusion feature for generating the super-resolution image. This maintains the QR code feature information while obtaining a clear reconstructed QR code image, significantly improving the image quality of the QR code image.

[0065] The specific implementation methods for obtaining a super-reconstructed feature map by performing feature reconstruction based on the fused feature information in step S13 and for obtaining a super-reconstructed QR code super-reconstructed image by performing feature decoding based on the super-reconstructed feature map in step S14 are the same as those in the previous embodiment, and will not be repeated here.

[0066] The feature encoder network, feature fusion network, feature reconstruction network, and feature decoder network that perform feature encoding, feature fusion, feature reconstruction, and feature decoding in the above exemplary embodiments will be specifically described in the following embodiments, and will not be elaborated here.

[0067] In the QR code image super-resolution reconstruction method provided in this application, the texture features (i.e. feature information) of multiple frames of QR code images are extracted, and the texture features of the QR code are fused into more reasonable fusion features to generate a super-resolution image. While maintaining the feature code information of the QR code, a clear reconstructed QR code image is obtained, which greatly improves the image quality of the QR code image.

[0068] Figure 4 This is a structural block diagram of a QR code image super-resolution reconstruction system according to an embodiment of this application. The QR code image super-resolution reconstruction system includes a deep learning model for QR code image super-resolution reconstruction. The deep learning model includes a feature encoder network 41, a feature fusion network 42, a feature reconstruction network 43, and a feature decoder network 44, which are cascaded in sequence. The feature encoder network 41 is connected to an image input module, and the feature decoder network 44 is connected to an image output module. The image input module is used to input the acquired QR code image and input it to the feature encoder network 41. The image output module is used to output the super-reconstructed QR code image obtained by the feature decoder network 44.

[0069] A feature encoder network 41 is configured to encode features in the QR code image to be super-reconstructed to obtain feature information corresponding to the QR code image. A feature fusion network 42 is configured to fuse the feature information to obtain fused feature information. A feature reconstruction network 43 is configured to perform feature reconstruction based on the fused feature information to obtain a super-reconstructed feature map. Furthermore, a feature decoder network 44 is configured to perform feature decoding based on the super-reconstructed feature map to obtain a super-reconstructed QR code image.

[0070] In some embodiments, the feature encoder network is used to extract and decompose the feature information of the QR code image. The feature information includes QR code image features, self-similarity features, and edge features. The QR code image features include multiple levels of features, including primary features, intermediate features, and high-level features.

[0071] In one embodiment, the feature encoder network extracts QR code image features (primary, intermediate, and advanced features) through multiple convolutional layers. The feature encoder network extracts self-similarity features through a transfer convolution module, as QR code images contain a large amount of self-similarity, which helps in reconstructing the details of the QR code image. Furthermore, the feature encoder network extracts edge information of the QR code image through an edge detector, assigning more loss weights to edge regions to force the network to pay more attention to these areas. The edge detector can employ a Laplacian filter, etc.

[0072] In this embodiment, the QR code image features include code eye features and code dot features, and the code eye features and code dot features each include multiple levels.

[0073] Figure 5 This diagram illustrates QR code images with different "eye" features. The "eye" features include low-level, intermediate, and high-level features. Low-level features include the dot / line features of the "eye," intermediate features include the structural features, and high-level features include the configuration features. The dot / line features are the dots / lines that make up the "eye," and the structural features are the pattern features, such as... Figure 5 In China, the code eye patterns include rounded rectangles, inner circle and outer square, squares, rounded rectangles, circles, concentric circles, etc.

[0074] Figure 6 This diagram illustrates QR code images with different code point features. The code point features include low-level, mid-level, and high-level features. Low-level features include the dot / line features of the code point; mid-level features include the structural features of the code point; and high-level features include the configuration features of the code point. The dot / line features are the dots / lines that make up the code point, and the structural features are the pattern features of the code point, such as... Figure 5 In the code, the dot patterns include rhombus, square, circle, small dot, rounded rectangle, rounded rectangle, midpoint, etc.

[0075] In a preferred embodiment, the feature fusion network includes a fusion convolutional learning network. The fusion convolutional learning network calculates the feature information to fuse the feature information corresponding to multi-dimensional features into pixel-level weights using convolution operations, which are then used as the fused feature information. In this embodiment, feature fusion can be achieved in the super-resolution reconstruction process of a single-frame QR code image. In another preferred embodiment, the feature fusion network consists of an image alignment convolutional learning network and a fusion convolutional learning network. The image alignment convolutional learning network is used for alignment operations of multiple frames of QR code images, which is beneficial for fusing the temporal correlation features of QR code images. The fusion convolutional learning network is used to fuse the feature information corresponding to the aligned QR code image into a single feature to obtain the fused feature information. That is, it fuses multiple dimensions of features such as QR code image features, self-similarity, and edge features from multiple frames of QR code images into a single feature, and calculates the single feature to fuse the feature information into pixel-level weights using convolution operations, which are then used as the fused feature information. Preferably, the image alignment convolutional learning network is based on the eye-code feature and / or dot-code feature, and then aligns the QR code image based on the eye-code feature and / or dot-code feature. Aligning QR code images using eye and / or dot features helps reduce the complexity of the network model. In this embodiment, feature fusion of multiple QR code images can be achieved during the super-resolution reconstruction process of multiple frames of QR code images.

[0076] In a preferred embodiment, the feature reconstruction network is composed of a convolutional learning network, which performs feature reconstruction based on the fused feature information to obtain a super-reconstructed feature map.

[0077] In a preferred embodiment, the feature decoder network is composed of a convolutional learning network, which performs feature decoding based on the super-reconstructed feature map to obtain the super-reconstructed QR code image.

[0078] In this application, the QR code image super-resolution reconstruction system extracts the texture features (i.e., feature information) of multiple frames of QR code images through a feature encoder network. The feature fusion network fuses the texture features of the QR code into more reasonable fusion features for generating the super-resolution image. The feature reconstruction network and the feature decoder network perform feature reconstruction and decoding. While maintaining the feature code information of the QR code unchanged, a clear reconstructed QR code image is obtained, which greatly improves the image quality of the QR code image.

[0079] It should be noted that the feature encoder network, feature fusion network, feature reconstruction network, and feature decoder network that constitute the deep learning model are sub-network structures obtained by functionally dividing the end-to-end deep learning model network. The feature encoder network, feature fusion network, feature reconstruction network, and feature decoder network are combined into a whole to form a deep learning model network and are used for model training. The trained deep learning model network is used for QR code image super-resolution reconstruction in this application.

[0080] Deep learning model networks have various basic network structures: convolutional layers, residual networks (ResNet), VGG units, fully connected layers, nonlinear ReLU layers, pooling layers, etc. In some preferred embodiments, the deep learning model network structure used for QR code image super-resolution reconstruction generally employs convolutional neural networks (CNN), generative adversarial networks (GAN), or recurrent neural networks (RNN).

[0081] Convolutional Neural Networks (CNNs) are a class of feedforward neural networks that incorporate convolutional computations and have a deep structure. They are one of the representative algorithms of deep learning. CNNs possess representation learning capabilities, enabling them to perform shift-invariant classification of input information according to their hierarchical structure; therefore, they are also known as Shift-Invariant Artificial Neural Networks (SIANNs).

[0082] Generative Adversarial Networks (GANs) are deep learning models and one of the most promising unsupervised learning methods on complex distributions in recent years. The model produces relatively good outputs through a game-like learning process between (at least) two modules: a generative model and a discriminative model. In the original GAN ​​theory, G and D do not need to be neural networks; they only need to fit the corresponding generation and discrimination functions. During training, the goal of the generative network G is to generate realistic images to deceive the discriminative network D. The goal of D is to distinguish the images generated by G from real images as much as possible. Thus, G and D constitute a dynamic "game process."

[0083] A recurrent neural network (RNN) is a type of recursive neural network that takes sequence data as input, recurses in the direction of sequence evolution, and connects all nodes (recurrent units) in a chain-like manner.

[0084] Figure 7 This is a flowchart of a QR code image recognition method according to an embodiment of this application. Figure 7 As shown, the QR code image recognition method in this embodiment includes steps S21 and S22.

[0085] In step S21, the acquired QR code image is obtained, and the super-resolution reconstructed QR code image is obtained using the QR code image super-resolution reconstruction method in the above embodiments. The QR code super-resolution reconstruction method has been described in detail in the above embodiments and will not be repeated here.

[0086] In step S22, the feature code of the super-resolution reconstructed image of the QR code is identified.

[0087] In the QR code image recognition method provided in this application, the QR code image is first preprocessed by super-resolution reconstruction to transform the low-resolution QR code image (blurry QR code image) into a high-resolution QR code image (clear QR code image) while maintaining the feature code information of the QR code unchanged, thereby improving the recognition rate of the QR code.

[0088] Figure 8 This is a structural block diagram of a QR code image recognition system according to an embodiment of this application. Figure 8 As shown, the QR code image recognition system 8 in this embodiment includes a super-resolution reconstruction module 81 and a recognition module 82.

[0089] The super-resolution reconstruction module 81 is configured to acquire a collected QR code image, perform feature encoding on the QR code image to obtain feature information corresponding to the QR code image, fuse the feature information to obtain fused feature information, perform feature reconstruction based on the fused feature information to obtain a super-resolution reconstruction feature map, and perform feature decoding based on the super-resolution reconstruction feature map to obtain a super-resolution reconstructed QR code image.

[0090] The recognition module 82 is configured to recognize the feature code of the super-resolution reconstructed image of the QR code.

[0091] In this embodiment, the super-resolution reconstruction module 81 uses the deep learning model for super-resolution reconstruction of QR code images in the above-mentioned QR code image super-resolution reconstruction system to perform super-resolution reconstruction of the QR code image, obtain the reconstructed QR code image, and then use the recognition module 82 to perform feature code recognition.

[0092] In the QR code image recognition system provided in this application, the super-resolution reconstruction module first performs super-resolution reconstruction on the QR code image to process the low-resolution QR code image (blurry QR code image) into a high-resolution QR code image (clear QR code image) while maintaining the feature code information of the QR code unchanged. Then, the recognition model is used to recognize the reconstructed QR code image, thereby improving the QR code recognition rate.

[0093] The scope of protection for the QR code image super-resolution reconstruction method and QR code image recognition method described in this application is not limited to the execution order of the steps listed in this embodiment. Any solution implemented by adding, subtracting, or replacing steps in the prior art based on the principles of this application is included within the scope of protection of this application.

[0094] The QR code image super-resolution reconstruction system / QR code image recognition system provided in this application can implement the QR code image super-resolution reconstruction method / QR code image recognition method described in this application. However, the implementation device of the QR code image super-resolution reconstruction method / QR code image recognition method described in this application includes, but is not limited to, the structure of the QR code super-resolution reconstruction system / QR code image recognition system listed in this embodiment. All structural modifications and substitutions of the prior art made based on the principles of this application are included within the protection scope of this application.

[0095] This application also provides an electronic device. Figure 9 This is a structural block diagram of an electronic device according to an embodiment of the present invention. (Reference) Figure 9 In one embodiment of the present invention, the electronic device 9 includes a memory 91 and a processor 92. The memory 91 is configured to store a computer program, the processor 92 is communicatively connected to the memory 91, and the processor 92 is configured to invoke the computer program to execute the QR code image super-resolution reconstruction method and / or the QR code image recognition method according to the present application.

[0096] Optionally, the electronic device 9 also includes a display 93. The display 93 is communicatively connected to the memory 91 and the processor 92, and serves as a GUI interface for the QR code image super-resolution reconstruction method and / or QR code image recognition method.

[0097] In some embodiments, the processor 92 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc. Alternatively, the processor 92 may be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0098] This application also provides a computer-readable storage medium storing a computer program, which is executed by a processor to implement the QR code image super-resolution reconstruction method and / or the QR code image recognition method according to this application. Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing a processor. The program can be stored in a computer-readable storage medium, which is a non-transitory medium, such as random access memory, read-only memory, flash memory, hard disk, solid-state drive, magnetic tape, floppy disk, optical disk, and any combination thereof. The above storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., digital video disc (DVD)), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0099] This application embodiment may also provide a computer program product comprising one or more computer instructions. When the computer instructions are loaded and executed on a computing device, all or part of the processes or functions described in this application embodiment are generated. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means.

[0100] When the computer program product is executed by a computer, the computer performs the method described in the foregoing method embodiments. The computer program product can be a software installation package; when the foregoing method is required, the computer program product can be downloaded and executed on the computer.

[0101] The descriptions of the processes or structures corresponding to the above figures each have their own emphasis. For parts of a process or structure that are not described in detail, please refer to the relevant descriptions of other processes or structures.

[0102] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this application should still be covered by the claims of this application.

Claims

1. A method for super-resolution reconstruction of QR code images, characterized in that, include: Feature encoding is performed on the QR code image to be super-reconstructed to obtain feature information corresponding to the QR code image, including extracting feature information corresponding to the QR code image through a feature encoder network. The feature information includes QR code image features, self-similarity features, and edge features. The feature information is fused to obtain fused feature information, including calculating the feature information through a feature fusion network, and using convolution operation to fuse the feature information corresponding to the multidimensional features into pixel-level weights as the fused feature information; Based on the fused feature information, feature reconstruction is performed to obtain a super-resolution reconstructed feature map; and Feature decoding is performed based on the super-reconstruction feature map to obtain the super-reconstructed QR code image.

2. The QR code image super-resolution reconstruction method according to claim 1, characterized in that, Extracting feature information corresponding to the QR code image through a feature encoder network includes: The QR code image features are extracted from the QR code image, and the QR code image features include features at multiple levels, including primary features, intermediate features and high-level features; The self-similarity features are extracted from the QR code image, and these features are used to reconstruct the details of the QR code image; and The edge features are extracted from the QR code image.

3. The QR code image super-resolution reconstruction method according to claim 1, characterized in that, Feature encoding of the QR code image to be super-reconstructed to obtain feature information corresponding to the QR code image includes: The multi-frame QR code images to be super-reconstructed are feature-encoded separately to obtain the feature information corresponding to each frame of the QR code image.

4. The QR code image super-resolution reconstruction method according to claim 3, characterized in that, The fusion of the feature information to obtain fused feature information includes: Align multiple frames of the QR code image; and The feature information corresponding to the aligned QR code image is fused into a single feature to obtain the fused feature information.

5. The QR code image super-resolution reconstruction method according to claim 4, characterized in that, Aligning multiple frames of the QR code image includes aligning multiple frames of the QR code image based on the eye feature and / or dot feature.

6. The QR code image super-resolution reconstruction method according to claim 1, characterized in that, The feature reconstruction based on the fused feature information to obtain the super-resolution reconstructed feature map includes: The super-resolution reconstructed feature map is calculated based on the feature information by using convolutional operations through a feature reconstruction network.

7. The QR code image super-resolution reconstruction method according to claim 1, characterized in that, The process of obtaining a super-reconstructed QR code image based on the super-reconstructed feature map includes: Super-resolution reconstruction feature decoding information is obtained based on the super-resolution reconstruction feature map through a feature decoder network; and The super-resolution reconstruction image of the QR code is calculated based on the super-resolution reconstruction feature decoding information and the QR code image.

8. A QR code image recognition method, characterized in that, include: Acquire the collected QR code image, and use the QR code image super-resolution reconstruction method according to any one of claims 1 to 7 to obtain the super-resolution reconstructed QR code image; and Identify the feature code of the super-resolution reconstructed image based on the QR code.

9. A QR code image super-resolution reconstruction system, characterized in that, include: A feature encoder network is configured to perform feature encoding on the QR code image to be super-reconstructed to obtain feature information corresponding to the QR code image; A feature fusion network is configured to fuse the feature information to obtain fused feature information; A feature reconstruction network is configured to perform feature reconstruction based on the fused feature information to obtain a super-resolution reconstructed feature map; and A feature decoder network is configured to perform feature decoding based on the super-reconstructed feature map to obtain a super-reconstructed QR code image. The feature encoder network is configured to extract feature information corresponding to the QR code image, the feature information including QR code image features, self-similarity features and edge features, and the feature fusion network is configured to calculate the feature information to fuse the feature information corresponding to the multidimensional features into pixel-level weights as the fused feature information using convolution operations.

10. The QR code image super-resolution reconstruction system according to claim 9, characterized in that, The feature encoder network, feature fusion network, feature reconstruction network, and feature decoder network are sequentially cascaded and integrated into a deep learning model.

11. A QR code image recognition system, characterized in that, include: The super-resolution reconstruction module is configured as follows: Acquire the collected QR code image; The QR code image is feature-encoded to obtain feature information corresponding to the QR code image; The feature information is fused to obtain fused feature information; Based on the fused feature information, feature reconstruction is performed to obtain a super-resolution reconstructed feature map; as well as Feature decoding is performed based on the super-reconstructed feature map to obtain the super-reconstructed QR code image; and The recognition module is configured to recognize the feature codes of the super-resolution reconstructed image of the QR code. The super-resolution reconstruction module is configured to extract feature information corresponding to the QR code image through a feature encoder network. The feature information includes QR code image features, self-similarity features, and edge features. The super-resolution reconstruction module is configured to calculate the feature information through a feature fusion network to fuse the feature information corresponding to the multi-dimensional features into pixel-level weights as the fused feature information using convolution operations.

12. An electronic device, characterized in that, include: Memory, configured to store computer programs; as well as The processor is configured to invoke the computer program to execute the QR code image super-resolution reconstruction method according to any one of claims 1 to 7 and / or the QR code image recognition method according to claim 8.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that, The computer program is executed to implement the QR code image super-resolution reconstruction method according to any one of claims 1 to 7 and / or the QR code image recognition method according to claim 8.

Images