Extreme overexposure image repair method, system, medium, program product, and device

Abstract

The application belongs to the technical field of computer vision and image processing, and provides an extreme overexposure image repairing method, system, medium, program product and equipment. The method comprises the following steps: obtaining an overexposure image, extracting effective pixel features, semi-effective pixel features and invalid pixel features in the overexposure image; constructing the dependency relationship between the semi-effective pixel features and the effective pixel features in the overexposure image and the three channels of pixels based on a channel attention module of a Transformer, adjusting the global exposure consistency of the overexposure image, and obtaining a global exposure correction image; processing the global exposure correction image by using a mask guided highlight completion module, and outputting a final repaired image from the last mask guided highlight completion module, so that a high-quality and visually consistent repaired result can be generated.

Description

Technical Field

[0001] This invention belongs to the field of computer vision and image processing technology, and particularly relates to a method, system, medium, program product and device for repairing extremely overexposed images. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Image inpainting and enhancement techniques have always been a research hotspot in computer vision and image processing, especially for the task of restoring overexposed images, with the goal of recovering details lost in highlight areas due to overexposure. In extremely overexposed scenes, highlight areas often suffer complete loss of detail due to pixel saturation, posing significant challenges to traditional inpainting methods. Existing technologies mainly fall into the following research directions: exposure correction and image inpainting, but these methods still have many shortcomings in extremely overexposed scenes.

[0004] Exposure correction aims to adjust the overall or local exposure levels of an image to improve its visual quality. Existing methods include end-to-end models based on global optimization and learning-driven approaches. Traditional optimization methods balance image brightness by adjusting the image's histogram distribution or illumination model. These methods, such as Retinex-based enhancement models, can improve unevenly exposed images to some extent, but they cannot recover lost details in overexposed highlight areas. Some convolutional neural network (CNN) based methods, such as Deep Photo Enhancer, adjust the image's exposure levels end-to-end. While these methods perform well in improving the overall quality of low-light or overexposed images, they tend to make global adjustments and struggle to accurately model highlight areas. Image inpainting tasks focus on filling in missing or occluded parts of an image to generate visually plausible content. However, image inpainting techniques lack targeted modeling of the specific characteristics of overexposed areas, especially when simultaneously processing valid and invalid pixels, making it difficult to coordinate the transition between them, often resulting in a lack of global consistency in the generated results. Generative Adversarial Network (GAN) and diffusion model methods perform well when dealing with small-scale missing areas, but for large-scale missing overexposed areas, the generated results are prone to inconsistencies in texture or unreasonable content.

[0005] Therefore, although existing exposure correction and image restoration methods have solved the problem of overexposed images to some extent, they still face the following major challenges for extreme overexposed scenes: (1) Exposure correction methods mainly make global adjustments to effective pixels and cannot solve the problem of missing details in the highlight area; (2) Image restoration methods lack modeling of the characteristics of the highlight area, resulting in inconsistent textures or insufficient details in the generated results; (3) Existing methods do not handle the transition between the highlight area and the background in a natural way, and the generated restoration content lacks global consistency; (4) There is a lack of high-quality datasets for extreme overexposed scenes, which limits the generalization ability and practical application effect of the model. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention provides a method, system, medium, program product, and device for repairing extremely overexposed images, which can generate high-quality and visually consistent repair results.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] The first aspect of the present invention provides a method for restoring extremely overexposed images.

[0009] In one or more embodiments, a method for restoring an extremely overexposed image is provided, comprising:

[0010] Acquire an overexposed image, and extract the effective pixel features, semi-effective pixel features, and invalid pixel features from the overexposed image;

[0011] The Transformer-based channel attention module constructs the dependencies between the semi-effective pixel features and effective pixel features in the overexposed image and the three pixel channels, respectively, and adjusts the global exposure consistency of the overexposed image to obtain a globally exposed corrected image.

[0012] The global exposure-corrected image is processed using at least one mask-guided highlight completion module, and the final repaired image is obtained from the output of the last mask-guided highlight completion module. If there are at least two mask-guided highlight completion modules, the input image of the first mask-guided highlight completion module is the global exposure-corrected image, and the input images of the remaining mask-guided highlight completion modules are the intermediate repaired images output by the previous mask-guided highlight completion module.

[0013] As one implementation method, in each of the mask-guided highlight completion modules, under the guidance of the highlight region mask, the highlight region in the input image is identified and the semi-effective pixel features and invalid pixel features therein are extracted and combined to form highlight region features; the highlight region features are processed by the fast Fourier convolution module to generate a residual image, which is then added to the input image to obtain the corresponding repaired image.

[0014] In one implementation, if there are at least two mask-guided specular completion modules, each intermediate repaired image is processed sequentially by a 1×1 convolutional layer and a sigmoid activation function to generate a soft mask, which is then input into the next mask-guided specular completion module to form a specular region mask to identify the specular region in the input image.

[0015] In one implementation, the fast Fourier convolution module includes two branches: a local branch and a global branch. In each branch, the highlight region features are used as input features. After processing by the local branch and the global branch respectively, local features and global features are obtained. The local features and global features are then connected to form detail features.

[0016] As one implementation method, the input features are divided into two groups by the channel dimension before inputting the local and global branches;

[0017] In the local branch, both sets of features are convolved, and then the output features are convolved again to form local features;

[0018] In the global branch, one set of features is processed by convolution, and another set of features is processed by a spectrum transformer to extract spectral features. Then, the two sets of processed output features are concatenated to form global features.

[0019] As one implementation, if there are at least two mask-guided specular completion modules, intermediate supervision is applied to the intermediate repaired image in the mask-guided specular completion module to optimize the mask quality; the intermediate supervision function is:

[0020]

[0021] Among them, G lr It is the real value image downsampled by a set ratio; N is the number of mask-guided specular completion modules, which is a positive integer greater than or equal to 2; This represents the intermediate repair image of the i-th module.

[0022] A second aspect of the present invention provides an extreme overexposed image restoration system.

[0023] In one or more embodiments, an extreme overexposed image restoration system includes:

[0024] A pixel classification unit is used to acquire overexposed images and extract effective pixel features, semi-effective pixel features, and invalid pixel features from the overexposed images.

[0025] The exposure correction unit is used to construct the dependencies between the semi-effective pixel features and effective pixel features in the overexposed image and the three channels of the pixel, respectively, based on the Transformer-based channel attention module, and adjust the global exposure consistency of the overexposed image to obtain a globally exposed corrected image.

[0026] The highlight reconstruction unit is used to process the global exposure-corrected image using at least one mask-guided highlight completion module, and the final repaired image is output by the last mask-guided highlight completion module. If there are at least two mask-guided highlight completion modules, the input image of the first mask-guided highlight completion module is the global exposure-corrected image, and the input images of the remaining mask-guided highlight completion modules are the intermediate repaired images output by the previous mask-guided highlight completion module.

[0027] A third aspect of the present invention provides a computer-readable storage medium.

[0028] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the extreme overexposure image restoration method described above.

[0029] A fourth aspect of the present invention provides a computer program product.

[0030] A computer program product includes a computer program / instructions that, when executed by a processor, implement the steps in the extreme overexposure image restoration method described above.

[0031] A fifth aspect of the present invention provides an electronic device.

[0032] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps in the extreme overexposure image restoration method described above.

[0033] Compared with the prior art, the beneficial effects of the present invention are:

[0034] (1) This invention uses two stages: exposure correction and highlight reconstruction. In the exposure correction stage, the channel attention module of Transformer is used to construct the dependency relationship between the semi-effective pixel features and effective pixel features in the overexposed image and the three channels of the pixel, respectively, to adjust the global exposure consistency of the overexposed image. In the highlight reconstruction stage, the semi-effective pixel features and ineffective pixel features in the highlight area are processed to generate real details, which solves the problem that existing methods are difficult to take into account both global and local repair.

[0035] (2) This invention introduces a mask-guided specular completion module. Under the guidance of the specular region mask, the specular region is accurately located. Combined with the fast Fourier convolution module, the specular region features are processed, which enhances the perception of global information. At the same time, the local context is used to generate content with rich details, so that the repaired specular region can be naturally integrated with the surrounding environment in terms of texture, color and lighting, which significantly improves the fusion effect between the specular region and the background.

[0036] (3) The present invention also applies intermediate supervision to the intermediate repair image, optimizes the mask quality, and ensures that the transition between the generated content and the original image is natural and delicate, thereby improving the repair effect of overexposed areas. Attached Figure Description

[0037] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0038] Figure 1 This is a schematic flowchart of an extreme overexposed image restoration method according to an embodiment of the present invention;

[0039] Figure 2 This is a schematic diagram of the two stages of exposure correction and highlight reconstruction in the extreme overexposure image restoration process of this invention embodiment;

[0040] Figure 3 This is a schematic diagram illustrating the two stages of exposure correction and highlight reconstruction in the extreme overexposed image restoration process according to an embodiment of the present invention.

[0041] Figure 4 This is a comparison of the effects of extreme overexposed image restoration tasks in embodiments of the present invention;

[0042] Figure 5 This is a comparative example of the performance differences between the extreme overexposure image restoration method of this invention and existing methods in extreme overexposure image restoration tasks.

[0043] Figure 6 This is Example 2, a comparison of the performance differences between the extreme overexposure image restoration method of this invention and existing methods in extreme overexposure image restoration tasks;

[0044] Figure 7 This is a schematic diagram of the structure of an extreme overexposed image restoration system according to an embodiment of the present invention;

[0045] Figure 8 This is a schematic diagram of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0046] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0047] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0048] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0049] Figure 1 This is a flowchart illustrating an extreme overexposed image restoration method according to an embodiment of the present invention, as shown below. Figure 1 The extreme overexposure image restoration method shown in this embodiment may include:

[0050] S101: Obtain an overexposed image and extract the effective pixel features, semi-effective pixel features, and invalid pixel features from the overexposed image.

[0051] In practice, effective pixel features, semi-effective pixel features, and invalid pixel features in overexposed images can be extracted using convolution modules or other existing image classification algorithms.

[0052] S102: The channel attention module based on Transformer constructs the dependency relationship between the semi-effective pixel features and effective pixel features in the overexposed image and the three channels of the pixel, respectively, and adjusts the global exposure consistency of the overexposed image to obtain a globally exposed corrected image.

[0053] Step S102 is the exposure correction stage. Combined with... Figure 2 and Figure 3 The Transformer-based channel attention module achieves adjustment of global exposure consistency through feature extraction, encoding, and decoding operations.

[0054] Specifically, for example, given input features F in The query vector Q, key K, and value V are obtained through a 1×1 convolution followed by a depthwise convolution, and then reshaped into their corresponding shapes. The channel attention map C can be represented as: C = Softmax(K×Q / λ), where the × operation represents matrix multiplication, and λ is a learnable scaling parameter used to ensure numerical stability. The modulated vector... It can be done through formula The calculated output features are obtained through a reshaping operation followed by a 1×1 convolutional layer. The exposure correction stage utilizes a Transformer-based channel attention module to model the three-channel dependencies of pixels and adjusts pixel values ​​using global feature distribution.

[0055] The Transformer-based channel attention module models the dependencies between channels and adjusts the brightness and exposure levels of the image, thereby achieving consistent overall image exposure.

[0056] S103: The global exposure-corrected image is processed using at least one mask-guided highlight completion module, and the final repaired image is obtained by the output of the last mask-guided highlight completion module; wherein, if there are at least two mask-guided highlight completion modules, the input image of the first mask-guided highlight completion module is the global exposure-corrected image, and the input images of the remaining mask-guided highlight completion modules are the intermediate repaired images output by the previous mask-guided highlight completion module.

[0057] Step S103 is the highlight reconstruction stage, focusing on repairing details lost due to overexposure in the image. Due to the frequency domain transformation, Fast Fourier Convolution (FFT) enables the network to have a receptive field covering the entire image in early layers. Specifically, in each mask-guided highlight completion module, guided by a highlight region mask, highlight regions in the input image are identified, and semi-effective and ineffective pixel features are extracted and combined to form highlight region features. These features are then processed using the FFT module to generate a residual image, which is then added to the input image to obtain the corresponding repaired image.

[0058] In the highlight reconstruction stage, a fast Fourier convolution module is used to enhance the network's ability to perceive global information and generate high-quality details by combining local context.

[0059] If there are at least two mask-guided specular completion modules, each intermediate repaired image is processed sequentially by a 1×1 convolutional layer and a sigmoid activation function to generate a soft mask, which is then input into the next mask-guided specular completion module to form a specular region mask to identify the specular region in the input image.

[0060] Specifically, the fast Fourier convolution module includes two branches: a local branch and a global branch. In each branch, the highlight region features are used as input features. After processing by the local branch and the global branch respectively, local features and global features are obtained. The local features and global features are then connected to form detail features.

[0061] In some specific embodiments, the input features are divided into two groups by the channel dimension before inputting the local and global branches;

[0062] In the local branch, both sets of features are convolved, and then the output features are convolved again to form local features;

[0063] In the global branch, one set of features is processed by convolution, and another set of features is processed by a spectrum transformer to extract spectral features. Then, the two sets of processed output features are concatenated to form global features.

[0064] In the spectrum converter, spatial features are first converted to the spectral domain, updated to spectral data, and then converted back to the spatial format.

[0065] The combination of global and local features enables the model to balance the needs of global and local features during the restoration process, effectively improving the restoration quality of highlight areas and ensuring the authenticity and consistency of the generated results in terms of texture, color, and lighting.

[0066] If there are at least two mask-guided specular completion modules, intermediate supervision is applied to the intermediate repaired image within the mask-guided specular completion module to optimize the mask quality; the intermediate supervision function is:

[0067]

[0068] Among them, G lr It is the real value image downsampled by a set ratio (e.g., 1 / 8); N is the number of mask-guided specular completion modules, which is a positive integer greater than or equal to 2; Let represent the intermediate repaired image of the i-th module. By introducing an intermediate supervision mechanism, efficient feature transfer at each stage is ensured, further improving the visual effect and realism of the final repaired result.

[0069] This embodiment precisely locates highlight areas under the guidance of a highlight area mask, enhances global perception capabilities by combining fast Fourier convolution, and refines generated content using local convolution, allowing the repaired highlight areas to blend naturally with the surrounding environment in terms of texture, color, and lighting. The entire repair process, through two collaborative stages, ensures both the uniformity of global exposure and the restoration of detailed information in the highlight areas, thereby generating a high-quality repaired image.

[0070] The final restored image of this embodiment was validated on the EOR4K dataset, demonstrating the significant effect of this extreme overexposed image restoration method on image quality improvement.

[0071] Figure 4This image demonstrates the effectiveness of the method of this invention in the task of restoring extremely overexposed images. From left to right, the images show the input overexposed image, the intermediate result after the first stage of exposure correction, the generated highlight area mask, and the final restoration result. The first image is the original overexposed input image, where excessive brightness caused the complete loss of some details. The second image shows the result after the first stage of exposure correction, which adjusted the exposure levels of effective and semi-effective pixels in the image, making the overall brightness more balanced, but the details in the highlight areas were not fully restored. The third image shows the highlight area mask generated in the second stage. This mask guides the highlight completion module to automatically mark the areas requiring key restoration, and the mask guides the highlight completion module to focus on generating content in the highlight areas. The fourth image is the final restoration result, showing that the details in the originally overexposed areas have been effectively restored, the textures of the person's face and clothing are clearer, and the transition between the lighting and color of the background is natural, presenting visual consistency. This demonstrates that the method of this invention, with multi-stage collaborative restoration and mask guidance, can achieve high-quality restoration of overexposed images.

[0072] Figure 5 and Figure 6 This paper demonstrates the performance difference between the method of this invention and existing methods in the task of restoring extremely overexposed images. Compared with methods such as FECNet, WaveFill, and HDRTVNet, which exhibit problems such as blurring, color distortion, and texture inconsistency in highlight area restoration, this method demonstrates superior detail recovery capabilities, generating realistic and delicate restored content while ensuring a natural transition between highlight areas and the surrounding environment. Whether in people, objects, or natural scenes, this invention can restore lost texture and color details, and the generated results are closer to the real target image, demonstrating higher visual consistency and generalization ability.

[0073] Figure 7 This is a schematic diagram of an extreme overexposed image restoration system according to an embodiment of the present invention. This embodiment is similar to... Figure 1 Corresponding to the extreme overexposure image restoration methods, such as Figure 7 As shown, the extreme overexposure image restoration system in this embodiment may include:

[0074] Pixel classification unit 701 is used to acquire an overexposed image and extract effective pixel features, semi-effective pixel features and invalid pixel features from the overexposed image.

[0075] The exposure correction unit 702 is used to construct the dependency relationship between the semi-effective pixel features and effective pixel features in the overexposed image and the three channels of the pixel, respectively, based on the Transformer-based channel attention module, and adjust the global exposure consistency of the overexposed image to obtain a globally exposed corrected image.

[0076] The highlight reconstruction unit 703 is used to process the global exposure-corrected image using at least one mask-guided highlight completion module, and the final repaired image is output by the last mask-guided highlight completion module. If there are at least two mask-guided highlight completion modules, the input image of the first mask-guided highlight completion module is the global exposure-corrected image, and the input images of the remaining mask-guided highlight completion modules are the intermediate repaired images output by the previous mask-guided highlight completion module.

[0077] In each of the mask-guided highlight completion modules, under the guidance of the highlight region mask, the highlight region in the input image is identified and the semi-effective pixel features and invalid pixel features are extracted and combined to form the highlight region features. The highlight region features are then processed by the Fast Fourier Convolution module to generate a residual image, which is then added to the input image to obtain the corresponding repaired image.

[0078] If there are at least two mask-guided specular completion modules, each intermediate repaired image is processed sequentially by a 1×1 convolutional layer and a sigmoid activation function to generate a soft mask, which is then input into the next mask-guided specular completion module to form a specular region mask to identify the specular region in the input image.

[0079] Specifically, the fast Fourier convolution module includes two branches: a local branch and a global branch. In each branch, the highlight region features are used as input features. After processing by the local branch and the global branch respectively, local features and global features are obtained. The local features and global features are then connected to form detail features.

[0080] In some specific embodiments, the input features are divided into two groups by the channel dimension before inputting the local and global branches;

[0081] In the local branch, both sets of features are convolved, and then the output features are convolved again to form local features;

[0082] In the global branch, one set of features is processed by convolution, and another set of features is processed by a spectrum transformer to extract spectral features. Then, the two sets of processed output features are concatenated to form global features.

[0083] If there are at least two mask-guided specular completion modules, intermediate supervision is applied to the intermediate repaired image within the mask-guided specular completion module to optimize the mask quality; the intermediate supervision function is:

[0084]

[0085] Among them, G lrIt is the real value image downsampled by a set ratio; N is the number of mask-guided specular completion modules, which is a positive integer greater than or equal to 2; This represents the intermediate repair image of the i-th module.

[0086] It should be noted here that, Figure 7 The various modules in the extreme overexposure image restoration system, and Figure 1 The steps in the extreme overexposed image restoration method correspond one-to-one, and their specific implementation processes are the same, so they will not be repeated here.

[0087] Reference Figure 8 A schematic diagram of an electronic device is provided. It should be noted that... Figure 8 The illustrated electronic device 800 is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0088] like Figure 8 As shown, the electronic device 800 includes a central processing unit (CPU) 801, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 802 or a program loaded from a storage section 808 into a random access memory (RAM) 803. The RAM 803 also stores various programs and data required for system operation. The CPU 801, ROM 802, and RAM 803 are interconnected via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0089] The following components are connected to I / O interface 805: an input section 806 including a keyboard, mouse, etc.; an output section 807 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 808 including a hard disk, etc.; and a communication section 809 including a network interface card such as a local area network (LAN) card, modem, etc. The communication section 809 performs communication processing via a network such as the Internet. A drive 810 is also connected to I / O interface 805 as needed. A removable medium 811, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 810 as needed so that computer programs read from it can be installed into storage section 808 as needed.

[0090] When the central processing unit 801 in the electronic device of this embodiment executes the program, it achieves the following: Figure 1 The steps in the extreme overexposed image restoration method shown.

[0091] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program including functions for executing... Figure 1 The program code for the method shown. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 809, and / or installed from a removable medium 811. When the computer program is executed by the central processing unit 801, it performs the various functions defined in the apparatus of this application.

[0092] in, Figure 1 The computer program instructions corresponding to the method shown may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in the process. Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0093] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0094] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for restoring extremely overexposed images, characterized in that, include: Acquire an overexposed image, and extract the effective pixel features, semi-effective pixel features, and invalid pixel features from the overexposed image; The Transformer-based channel attention module constructs the dependencies between the semi-effective pixel features and effective pixel features in the overexposed image and the three pixel channels, respectively, and adjusts the global exposure consistency of the overexposed image to obtain a globally exposed corrected image. The global exposure-corrected image is processed using at least one mask-guided highlight completion module, and the final repaired image is obtained by the output of the last mask-guided highlight completion module. If there are at least two mask-guided highlight completion modules, the input image of the first mask-guided highlight completion module is the global exposure-corrected image, and the input images of the remaining mask-guided highlight completion modules are the intermediate repaired images output by the previous mask-guided highlight completion module. In each of the mask-guided specular completion modules, under the guidance of the specular region mask, the specular region in the input image is identified and the semi-effective pixel features and invalid pixel features are extracted and combined to form the specular region features; the specular region features are processed by the fast Fourier convolution module to generate a residual image, which is then added to the input image to obtain the corresponding repaired image; The Fast Fourier Convolution module includes two branches: a local branch and a global branch. In each branch, the highlight region features are used as input features. After processing by the local branch and the global branch respectively, local features and global features are obtained. Then, the local features and global features are connected to form detail features. Before inputting the local and global branches, the input features are divided into two groups based on the channel dimension; In the local branch, both sets of features are convolved, and then the output features are convolved again to form local features; In the global branch, one set of features is processed by convolution, and another set of features is processed by a spectrum transformer to extract spectral features. Then, the two sets of processed output features are concatenated to form global features. If there are at least two mask-guided specular completion modules, each intermediate repaired image will be sequentially processed... The convolutional layer and sigmoid activation function process generate a soft mask, which is then input into the next mask-guided specular completion module to form a specular region mask to identify the specular regions in the input image. The entire restoration process involves two collaborative stages, ensuring both the uniformity of global exposure and the restoration of detail in highlight areas, thereby generating a high-quality restored image.

2. The method for restoring extremely overexposed images as described in claim 1, characterized in that, If there are at least two mask-guided specular completion modules, intermediate supervision is applied to the intermediate repaired image within the mask-guided specular completion module to optimize the mask quality; the intermediate supervision function is: ； in, It is a real value image downsampled by a set ratio; It represents the number of mask-guided specular completion modules, which is a positive integer greater than or equal to 2; Indicates the first Intermediate repair images for each module.

3. An extreme overexposure image restoration system employing the method described in claim 1, characterized in that, include: A pixel classification unit is used to acquire overexposed images and extract effective pixel features, semi-effective pixel features, and invalid pixel features from the overexposed images. The exposure correction unit is used to construct the dependencies between the semi-effective pixel features and effective pixel features in the overexposed image and the three channels of the pixel, respectively, based on the Transformer-based channel attention module, and adjust the global exposure consistency of the overexposed image to obtain a globally exposed corrected image. The highlight reconstruction unit is used to process the global exposure-corrected image using at least one mask-guided highlight completion module, and the final repaired image is output by the last mask-guided highlight completion module. If there are at least two mask-guided highlight completion modules, the input image of the first mask-guided highlight completion module is the global exposure-corrected image, and the input images of the remaining mask-guided highlight completion modules are the intermediate repaired images output by the previous mask-guided highlight completion module.

4. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps in the extreme overexposure image restoration method as described in any one of claims 1-2.

5. A computer program product, comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps in the extreme overexposure image restoration method as described in any one of claims 1-2.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the extreme overexposure image restoration method as described in any one of claims 1-2.

Images