Image denoising model generation method and device, equipment, storage medium and chip
By acquiring the training data set and correcting the initial image denoising model, the target image denoising model is generated, which solves the problem that the small computing power model cannot take into account both details and noise in the image denoising task, and achieves a better denoising effect.
Patent Information
- Application Number
- CN202411466717.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-18
- Publication Date
- 2025-07-25
AI Technical Summary
The existing small computing power artificial intelligence model cannot take into account the details and noise performance of the image in image denoising tasks, and the denoising effect is poor.
By acquiring the training data set, including the first image and the corresponding label image, the target image denoising model is generated using the initial image denoising model, and the initial image denoising model is corrected by determining the difference between the label image and the second image and the edge image difference.
The denoising effect of the target image denoising model on the image is improved, so that it can take into account the details and noise performance of the image at the same time.
Smart Images
Figure CN120374432A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of artificial intelligence technology, and in particular, to a method, apparatus, device, storage medium, and chip for generating an image denoising model. Background Art
[0002] With the development of artificial intelligence and the improvement of the computing power of edge devices, artificial intelligence models are gradually applied to the image denoising tasks of edge devices. However, limited by the power consumption requirements of edge devices, edge devices usually can only use small-computing-power artificial intelligence models to perform image denoising tasks.
[0003] In related technologies, small-computing-power artificial intelligence models have fewer parameters and less training data used. Therefore, when using small-computing-power artificial intelligence models to denoise images, it is impossible to well balance the details of the images and the performance of the noise, and the denoising effect is poor. Summary of the Invention
[0004] The present disclosure aims to at least partly solve one of the technical problems in the related technologies.
[0005] A first aspect embodiment of the present disclosure provides a method for generating an image denoising model, including:
[0006] Obtaining a training data set, where the training data set includes a first image and a corresponding labeled image;
[0007] Inputting the first image into an initial image denoising model to obtain a second image;
[0008] Determining a first edge image corresponding to the labeled image and a second edge image corresponding to the second image;
[0009] According to the difference between the labeled image and the second image and the difference between the first edge image and the second edge image, correcting the initial image denoising model to obtain a target image denoising model.
[0010] A second aspect embodiment of the present disclosure provides an apparatus for generating an image denoising model, including:
[0011] An obtaining module, configured to obtain a training data set, where the training data set includes a first image and a corresponding labeled image;
[0012] A processing module, configured to input the first image into an initial image denoising model to obtain a second image;
[0013] A determining module, configured to determine a first edge image corresponding to the labeled image and a second edge image corresponding to the second image;
[0014] A correction module, configured to correct the initial image denoising model according to the difference between the labeled image and the second image, and the difference between the first edge image and the second edge image, so as to obtain a target image denoising model.
[0015] According to a third aspect embodiment of the present disclosure, an electronic device is provided, including: a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the method for generating an image denoising model as proposed in the first aspect embodiment of the present disclosure is implemented.
[0016] According to a fourth aspect embodiment of the present disclosure, a computer-readable storage medium is provided, storing a computer program. When the computer program is executed by a processor, the method for generating an image denoising model as proposed in the first aspect embodiment of the present disclosure is implemented.
[0017] In a fifth aspect of the present disclosure, a chip is provided, including one or more interface circuits and one or more processors; the interface circuit is configured to receive a signal and send the signal to the processor, and the signal includes computer instructions stored in a memory. When the processor executes the computer instructions, the chip executes the steps of the method described in the first aspect.
[0018] In a sixth aspect of the present disclosure, an image processing method is provided, the method includes,
[0019] Inputting an image to be processed into the target image denoising model to output a target image, where the target image denoising model is obtained based on the method described in the first aspect. In an example of the present disclosure, the above target image denoising model can be placed into a chip.
[0020] In an example of the present disclosure, the above chip may specifically include an Image Signal Processor (ISP), a Graphic Processing Unit (GPU), etc.
[0021] In an embodiment of the present disclosure, the above image model can be implanted into an Artificial Intelligence (AI) server, and the AI server is used for model training or model training.
[0022] The method, device, equipment, storage medium, and chip for generating an image denoising model provided by the present disclosure have the following beneficial effects:
[0023] In an embodiment of the present disclosure, first, a training data set is obtained, where the training data set includes a first image and a corresponding label image. Then, the first image is input into an initial image denoising model to obtain a second image, and a first edge image corresponding to the label image and a second edge image corresponding to the second image are determined. Finally, the initial image denoising model is corrected according to the difference between the label image and the second image and the difference between the first edge image and the second edge image to obtain a target image denoising model. Thus, based on the noise difference between the second image denoised by the initial image denoising model and the label image, and the detail difference between the first edge image corresponding to the label image and the second edge image corresponding to the second image, the image denoising model is trained, so that the trained target image denoising model can take into account both the details of the image and the performance of the noise, improving the denoising effect of the target image denoising model on the image.
[0024] Additional aspects and advantages of the present disclosure will be given in part in the following description, become apparent in part from the following description, or be learned through the practice of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] The above and / or additional aspects and advantages of the present disclosure will become apparent and be readily understood from the following description of the embodiments in conjunction with the accompanying drawings, where:
[0026] Figure 1 is a schematic flowchart of a method for generating an image denoising model provided by an embodiment of the present disclosure;
[0027] Figure 2 is a schematic flowchart of a method for generating an image denoising model provided by another embodiment of the present disclosure;
[0028] Figure 3 is a schematic flowchart of a method for determining a second loss value provided by an embodiment of the present disclosure;
[0029] Figure 4 is a schematic flowchart of a method for generating an image denoising model provided by another embodiment of the present disclosure;
[0030] Figure 5 is a schematic structural diagram of an image denoising model generation device provided by another embodiment of the present disclosure;
[0031] Figure 6 shows a block diagram of an exemplary electronic device suitable for implementing the embodiments of the present disclosure. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0032] Embodiments of the present disclosure will be described in detail below. Examples of the embodiments are shown in the accompanying drawings, where like or similar reference numerals denote like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the accompanying drawings are exemplary and are intended to explain the present disclosure, and should not be construed as limiting the present disclosure.
[0033] A method, apparatus, device, storage medium, and chip for generating an image denoising model according to an embodiment of the present disclosure will be described below with reference to the accompanying drawings.
[0034] In the embodiments of the present disclosure, the method for generating the image denoising model is configured in a device for generating the image denoising model as an example. The device for generating the image denoising model can be applied to any electronic device so that the electronic device can execute the function of generating the image denoising model. In an embodiment of the present disclosure, the above method can be completed by a processing chip.
[0035] Figure 1 A flowchart of a method for generating an image denoising model provided by an embodiment of the present disclosure.
[0036] As Figure 1 shown, the method for generating the image denoising model may include the following steps:
[0037] Step 101: Obtain a training data set, where the training data set includes a first image and a corresponding labeled image.
[0038] Among them, the first image may be an image containing noise, and the corresponding labeled image is an image without noise.
[0039] It should be noted that images directly captured by an image sensor usually contain white noise, and it is difficult to capture an image without noise. Therefore, an artificially synthesized image without noise can be used as the labeled image, and then noise is added to the labeled image to obtain the first image containing noise. In some embodiments, one labeled image may correspond to one or more first images, and the present disclosure does not limit this.
[0040] In some embodiments, the first image and the labeled image may be in RGB format, Portable Network Graphics (PNG) format, etc. The present disclosure does not limit this.
[0041] In some embodiments, the first image and the labeled image may also be in RAW format. It should be noted that the RAW format is a lossless compression format, and the data is the original file without being processed by the camera, retaining more image details and original information. Therefore, denoising an image in RAW format can more easily restore a clear image.
[0042] Step 102: Input the first image into the initial image denoising model to obtain a second image.
[0043] The second image is the image obtained after the initial image denoising model denoises the first image.
[0044] The initial image denoising model can be an untrained model or a pre-trained model. The present disclosure does not limit this.
[0045] In some embodiments, the initial image denoising model can be a Convolutional Neural Networks (CNN), a Recurrent Neural Network (RNN), etc. The present disclosure does not limit this.
[0046] In some embodiments, the initial image denoising model can be a small computing power model. In some embodiments, the number of parameters in the initial image denoising model is less than a first value.
[0047] Step 103: Determine the first edge image corresponding to the labeled image and the second edge image corresponding to the second image.
[0048] In some embodiments, the Sobel operator, Laplacian operator, etc. can be used to perform edge detection on the labeled image to obtain the corresponding first edge image.
[0049] In some embodiments, the Sobel operator, Laplacian operator, etc. can be used to perform edge detection on the second image to obtain the corresponding second edge image.
[0050] In some embodiments, edge detection can also be performed on the labeled image and the second image respectively based on a fractional order differential operator to obtain the first edge image corresponding to the labeled image and the second edge image corresponding to the second image. Thus, the edge information of the image is extracted using the fractional order differential operator, thereby improving the effect of detail performance.
[0051] In some embodiments, the fractional order differential operator can be a Caputo operator, Tiansi operator, etc. The present disclosure does not limit this.
[0052] For example, the fractional order differential operator can be as shown in Table 1:
[0053] Table 1
[0054]
[0055]
[0056] Among them, v represents the order of the differential operator. In some embodiments, the value of V can be 0.5, 0.6, etc. The present disclosure does not limit this. Among them, the 0.5-order differential operator has the strongest robustness to noise, and richer and clearer edge information can be obtained when extracting edges.
[0057] Step 104: Modify the initial image denoising model according to the difference between the label image and the second image, and the difference between the first edge image and the second edge image, so as to obtain the target image denoising model.
[0058] Among them, the difference between the label image and the second image can represent the deviation between the second image obtained by denoising the first image by the initial image denoising model and the label image without noise.
[0059] Among them, the difference between the first edge image and the second edge image can represent the deviation between the edge of the second image and the edge of the label image without noise.
[0060] In the embodiments of the present disclosure, modifying the initial image denoising model based on the difference between the label image and the second image can improve the denoising ability of the image denoising model for images; modifying the initial image denoising model based on the difference between the first edge image and the second edge image can improve the ability of the image denoising model to retain image details.
[0061] In some embodiments, determine the first loss value according to the difference between the label image and the second image, determine the second loss value according to the difference between the first edge image and the second edge image, determine the target loss value according to the first loss value and the second loss value, and finally modify the initial image denoising model based on the target loss value to obtain the target image denoising model.
[0062] In some embodiments, the difference between the label image and the second image can be calculated based on the first loss function to obtain the first loss value; the difference between the first edge image and the second edge image can be calculated based on the second loss function to obtain the second loss value.
[0063] In some embodiments, the first loss function and the second loss function can be the same or different. For example, both the first loss function and the second loss function can be the Mean Absolute Error (MAE) function. The present disclosure does not limit this.
[0064] In some embodiments, the sum of the first loss value and the second loss value can be determined as the target loss value.
[0065] Alternatively, the first weight corresponding to the first loss value and the second weight corresponding to the second loss value can also be determined. Then, based on the first weight and the second weight, the first loss value and the second loss value are weighted to obtain the target loss value.
[0066] In the embodiments of the present disclosure, based on the noise difference between the second image denoised by the initial image denoising model and the label image, and the detail difference between the first edge image corresponding to the label image and the second edge image corresponding to the second image, the image denoising model is trained. Even if the image processing model is a small computing power model, it can also take into account both the details of the image and the performance of the noise, and can improve the denoising effect of the small computing power target image denoising model deployed on the edge side on the image.
[0067] In the embodiments of the present disclosure, first, a training data set is obtained, where the training data set includes a first image and the corresponding label image. Then, the first image is input into the initial image denoising model to obtain a second image, and the first edge image corresponding to the label image and the second edge image corresponding to the second image are determined. Finally, based on the difference between the label image and the second image and the difference between the first edge image and the second edge image, the initial image denoising model is corrected to obtain the target image denoising model. Thus, based on the noise difference between the second image denoised by the initial image denoising model and the label image, and the detail difference between the first edge image corresponding to the label image and the second edge image corresponding to the second image, the image denoising model is trained, so that the trained target image denoising model can take into account both the details of the image and the performance of the noise, and improves the denoising effect of the target image denoising model on the image.
[0068] Figure 2 It is a schematic flow chart of a method for generating an image denoising model provided by an embodiment of the present disclosure. As Figure 2 shown, the method for generating the image denoising model may include the following steps:
[0069] Step 201, obtain a training data set, where the training data set includes a first image and the corresponding label image, and the first image and the label image are in RAW format.
[0070] Step 202, input the first image into the initial image denoising model to obtain a second image.
[0071] Among them, the specific implementation forms of steps 201 to 202 can refer to the detailed descriptions in other embodiments of the present disclosure, and will not be specifically elaborated here.
[0072] Step 203, convert the label image and the second image into RGB format respectively to obtain the third image corresponding to the label image and the fourth image corresponding to the second image.
[0073] In some embodiments, the label image and the second image can be respectively converted into RGB format through a differentiable ISP to obtain a third image corresponding to the label image and a fourth image corresponding to the second image.
[0074] In some embodiments, the ISP may include several modules such as a Demosaic module, a WhiteBalance Correction (WBC), a Color Correction Matrix (CCM), and a Gamma correction.
[0075] Step 204: Convert the third image and the fourth image into YUV format respectively to obtain a fifth image corresponding to the third image and a sixth image corresponding to the fourth image.
[0076] In some embodiments, based on the conversion formula between RGB and YUV, the third image and the fourth image in RGB format can be respectively converted into YUV format.
[0077] Step 205: Perform edge detection on the luminance channel in the fifth image based on a fractional differential operator to obtain a first edge image.
[0078] Among them, the luminance channel is the Y channel.
[0079] Step 206: Perform edge detection on the luminance channel in the sixth image based on a fractional differential operator to obtain a second edge image.
[0080] In the embodiments of the present disclosure, by performing edge detection on the luminance channel through a fractional differential operator, the luminance information in the Y channel can be fully utilized, and at the same time, the interference of the color information in the UV channels can be avoided, thereby improving the accuracy of edge detection.
[0081] Step 207: Modify the initial image denoising model according to the difference between the label image and the second image and the difference between the first edge image and the second edge image to obtain a target image denoising model.
[0082] Among them, the specific implementation form of step 207 can refer to the detailed descriptions in other embodiments of the present disclosure and will not be specifically elaborated here.
[0083] Figure 3 It is a schematic flowchart of a method for determining a second loss value provided by an embodiment of the present disclosure; as Figure 3As shown in the figure, the RAW format label image is converted into a third image in RGB format through ISP, the third image in RGB format is converted into a fifth image in YUV format, and edge detection is performed on the luminance channel in the fifth image based on a fractional differential operator to obtain a first edge image; the first image in RAW format is input into an initial image denoising model to obtain a second image in RAW format, which is converted into a fourth image in RGB format through ISP, the fourth image in RGB format is converted into a sixth image in YUV format, and edge detection is performed on the luminance channel in the sixth image based on a fractional differential operator to obtain a second edge image; finally, a second loss value is determined according to the first edge image and the second edge image.
[0084] In the embodiments of the present disclosure, a training data set is first obtained. The training data set includes a first image in RAW format and a corresponding label image. The first image is input into an initial image denoising model to obtain a second image. The label image and the second image are respectively converted into RGB format to obtain a third image corresponding to the label image and a fourth image corresponding to the second image. The third image and the fourth image are respectively converted into YUV format to obtain a fifth image corresponding to the third image and a sixth image corresponding to the fourth image. Edge detection is performed on the luminance channel in the fifth image based on a fractional differential operator to obtain a first edge image, and edge detection is performed on the luminance channel in the sixth image based on a fractional differential operator to obtain a second edge image. The initial image denoising model is corrected according to the difference between the label image and the second image and the difference between the first edge image and the second edge image to obtain a target image denoising model. Thus, denoising of an image in RAW format can retain more image details and original information. Moreover, converting the image in RAW format to YUV format and performing edge detection on the luminance channel (Y channel) based on a fractional differential operator can improve the accuracy of edge detection and further improve the denoising effect of the target image denoising model on the image.
[0085] Figure 4 is a schematic flowchart of a method for generating an image denoising model provided by an embodiment of the present disclosure; as Figure 4 shown, the method for generating the image denoising model may include the following steps:
[0086] Step 401, obtain a training data set, where the training data set includes a first image and a corresponding label image, and the first image and the label image are in RAW format.
[0087] Step 402, input the first image into an initial image denoising model to obtain a second image.
[0088] Step 403: Convert the label image and the second image into RGB format respectively to obtain the third image corresponding to the label image and the fourth image corresponding to the second image.
[0089] Among them, for the specific implementation forms of steps 401 to 403, reference can be made to the detailed descriptions in other embodiments of the present disclosure, and no further details will be elaborated here.
[0090] Step 404: Perform edge detection on the third image based on the fractional differential operator to obtain the first edge image.
[0091] In some embodiments, the third image can be first converted into a grayscale image, and then edge detection is performed on the grayscale image corresponding to the third image based on the fractional differential operator to obtain the first edge image.
[0092] Step 405: Perform edge detection on the fourth image based on the fractional differential operator to obtain the second edge image.
[0093] In some embodiments, the fourth image can be first converted into a grayscale image, and then edge detection is performed on the grayscale image corresponding to the fourth image based on the fractional differential operator to obtain the second edge image.
[0094] Step 406: Modify the initial image denoising model according to the difference between the label image and the second image and the difference between the first edge image and the second edge image to obtain the target image denoising model.
[0095] Among them, for the specific implementation form of step 406, reference can be made to the detailed descriptions in other embodiments of the present disclosure, and no further details will be elaborated here.
[0096] In the embodiments of the present disclosure, a training data set is obtained, where the training data set includes the first image in RAW format and the corresponding label image. The first image is input into the initial image denoising model to obtain the second image. The label image and the second image are respectively converted into RGB format to obtain the third image corresponding to the label image and the fourth image corresponding to the second image. Edge detection is performed on the third image based on the fractional differential operator to obtain the first edge image, and edge detection is performed on the fourth image to obtain the second edge image. Finally, the initial image denoising model is modified according to the difference between the label image and the second image and the difference between the first edge image and the second edge image to obtain the target image denoising model. Thus, denoising the RAW format image can retain more image details and original information, and converting the RAW format image to RGB format and performing edge detection on the RGB format image based on the fractional differential operator can improve the accuracy and computational amount of edge detection, while improving the denoising effect of the target image denoising model on the image and improving the training efficiency.
[0097] An embodiment of the present disclosure further provides an image processing method, including inputting an image to be processed into a target image denoising model to output a target image. The target image denoising model is generated according to the method for generating an image denoising model proposed in the embodiments of the present disclosure. Since the target image denoising model can take into account both the details of the image and the performance of the noise, improving the denoising effect of the target image denoising model on the image. Therefore, when using the target image denoising model to denoise the image to be processed, a target image with a better denoising effect can be obtained.
[0098] To implement the above embodiments, the present disclosure further proposes a device for generating an image denoising model.
[0099] Figure 5 It is a schematic structural diagram of the device for generating an image denoising model provided by the embodiments of the present disclosure.
[0100] As Figure 5 shown, the device 500 for generating an image denoising model may include:
[0101] An acquisition module 501, configured to acquire a training data set, where the training data set includes a first image and a corresponding labeled image;
[0102] A processing module 502, configured to input the first image into an initial image denoising model to obtain a second image;
[0103] A determination module 503, configured to determine a first edge image corresponding to the labeled image and a second edge image corresponding to the second image;
[0104] A correction module 504, configured to correct the initial image denoising model according to the difference between the labeled image and the second image and the difference between the first edge image and the second edge image to obtain a target image denoising model.
[0105] In some embodiments, the determination module 503 is configured to:
[0106] Based on a fractional-order differential operator, perform edge detection on the labeled image and the second image respectively to obtain a first edge image corresponding to the labeled image and a second edge image corresponding to the second image.
[0107] In some embodiments, the first image and the labeled image are in RAW format.
[0108] In some embodiments, the determination module 503 is configured to:
[0109] Convert the labeled image and the second image into RGB format respectively to obtain a third image corresponding to the labeled image and a fourth image corresponding to the second image;
[0110] Based on a fractional-order differential operator, edge detection is performed on the third image to obtain a first edge image;
[0111] Based on a fractional-order differential operator, edge detection is performed on the fourth image to obtain a second edge image.
[0112] In some embodiments, a determination module 503 is configured to:
[0113] Convert the label image and the second image into the RGB format respectively to obtain a third image corresponding to the label image and a fourth image corresponding to the second image;
[0114] Convert the third image and the fourth image into the YUV format respectively to obtain a fifth image corresponding to the third image and a sixth image corresponding to the fourth image;
[0115] Based on a fractional-order differential operator, edge detection is performed on the luminance channel in the fifth image to obtain a first edge image;
[0116] Based on a fractional-order differential operator, edge detection is performed on the luminance channel in the sixth image to obtain a second edge image.
[0117] In some embodiments, a correction module 504 is configured to:
[0118] Determine a first loss value according to the difference between the label image and the second image;
[0119] Determine a second loss value according to the difference between the first edge image and the second edge image;
[0120] Determine a target loss value according to the first loss value and the second loss value;
[0121] Based on the target loss value, correct the initial image denoising model to obtain a target image denoising model.
[0122] For the functions and specific implementation principles of the above-mentioned modules in the embodiments of the present disclosure, reference may be made to the above-mentioned method embodiments, and details are not described herein again.
[0123] The generating device of the image denoising model according to an embodiment of the present disclosure first obtains a training data set, where the training data set includes a first image and a corresponding labeled image. Then, the first image is input into an initial image denoising model to obtain a second image, and a first edge image corresponding to the labeled image and a second edge image corresponding to the second image are determined. Finally, the initial image denoising model is corrected according to the difference between the labeled image and the second image and the difference between the first edge image and the second edge image to obtain a target image denoising model. Thus, based on the noise difference between the second image denoised by the initial image denoising model and the labeled image, and the detail difference between the first edge image corresponding to the labeled image and the second edge image corresponding to the second image, the image denoising model is trained, so that the obtained target image denoising model can take into account both the details of the image and the performance of the noise, improving the denoising effect of the target image denoising model on the image.
[0124] To implement the above embodiment, the present disclosure also proposes a chip, including one or more interface circuits and one or more processors; the interface circuit is configured to receive a signal and send the signal to the processor, and the signal includes computer instructions stored in a memory. When the processor executes the computer instructions, the chip executes the steps of the method for generating an image denoising model.
[0125] In an example of the present disclosure, the above chip may specifically include an image signal processor (ISP), a graphics processing unit (GPU), etc.
[0126] To implement the above embodiment, the present disclosure also proposes an electronic device, including: a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the method for generating an image denoising model as proposed in the foregoing embodiments of the present disclosure is implemented.
[0127] Figure 6 A block diagram of an exemplary electronic device suitable for implementing the embodiments of the present disclosure is shown. Figure 6 The illustrated electronic device 12 is merely an example and should not impose any limitation on the functions and usage scope of the embodiments of the present disclosure.
[0128] As Figure 6 shown, the electronic device 12 is presented in the form of a general-purpose computing device. The components of the electronic device 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, and a bus 18 connecting different system components (including the system memory 28 and the processing unit 16).
[0129] Bus 18 represents one or more of several types of bus architectures, including a memory bus or memory controller, a peripheral bus, an Accelerated Graphics Port, a processor, or a local bus using any of the multiple bus architectures. By way of example, these architectures include, but are not limited to, Industry Standard Architecture (ISA) bus, Micro Channel Architecture (MAC) bus, Enhanced ISA bus, Video Electronics Standards Association (VESA) local bus, and Peripheral Component Interconnection (PCI) bus.
[0130] Electronic device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by electronic device 12, including volatile and nonvolatile media, removable and non-removable media.
[0131] Memory 28 may include computer system readable media in the form of volatile memory, such as Random Access Memory (RAM) 30 and / or cache memory 32. Electronic device 12 may further include other removable / non-removable, volatile / nonvolatile computer system storage media. By way of example only, storage system 34 can be used for reading and writing on non-removable, nonvolatile magnetic media ( Figure 6 not shown, typically referred to as a "hard disk drive"). Although Figure 6 not shown in, a disk drive for reading and writing on a removable nonvolatile disk (such as a "floppy disk"), and an optical disk drive for reading and writing on a removable nonvolatile optical disk (such as Compact Disc Read Only Memory (CD-ROM), Digital Video Disc Read Only Memory (DVD-ROM), or other optical media) can be provided. In these cases, each drive can be connected to bus 18 through one or more data media interfaces. Memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the embodiments of the present disclosure.
[0132] A program / utilities 40 having a set (at least one) of program modules 42 can be stored, for example, in a memory 28. Such program modules 42 include, but are not limited to, an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. The program modules 42 generally execute the functions and / or methods in the embodiments described in this disclosure.
[0133] The electronic device 12 can also communicate with one or more external devices 14 (such as a keyboard, a pointing device, a display 24, etc.), and can also communicate with one or more devices that enable a user to interact with the electronic device 12, and / or communicate with any device that enables the electronic device 12 to communicate with one or more other computing devices (such as a network card, a modem, etc.). Such communication can be carried out through an input / output (I / O) interface 22. Moreover, the electronic device 12 can also communicate with one or more networks (such as a Local Area Network (LAN), a Wide Area Network (WAN), and / or a public network, such as the Internet) through a network adapter 20. As shown in the figure, the network adapter 20 communicates with other modules of the electronic device 12 through a bus 18. It should be understood that although not shown in the figure, other hardware and / or software modules can be used in combination with the electronic device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.
[0134] The processing unit 16 executes various functional applications and data processing by running programs stored in the system memory 28, such as implementing the methods mentioned in the foregoing embodiments.
[0135] To implement the above embodiments, the present disclosure also proposes a computer-readable storage medium storing a computer program which, when executed by a processor, implements the method for generating an image denoising model as proposed in the foregoing embodiments of the present disclosure.
[0136] In the description of this specification, the descriptions referring to terms such as "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.
[0137] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one of such features. In the description of the present disclosure, "a plurality of" means at least two, such as two, three, etc., unless otherwise specifically defined.
[0138] Any process or method description, whether in a flowchart or described otherwise herein, can be understood to represent a module, segment, or portion of code including one or more executable instructions for implementing a customized logic function or process, and the scope of the preferred embodiments of the present disclosure includes additional implementations, where the functions may be executed in a substantially simultaneous manner or in an order opposite to that shown or discussed, according to the functions involved, which should be understood by those skilled in the art to which the embodiments of the present disclosure pertain.
[0139] Logic and / or steps represented in a flowchart or otherwise described herein can, for example, be considered as a definitional sequence of executable instructions for implementing logical functions and can be embodied specifically in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device. As used in this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in connection with the instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electrical connection portion having one or more wirings (electronic device), a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable medium on which the program can be printed, as the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpretation, or otherwise processing as appropriate, and then storing it in a computer memory.
[0140] It should be understood that various parts of the present disclosure can be implemented by hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.
[0141] Those of ordinary skill in the art of this technology can understand that all or part of the steps carried by the method of implementing the above embodiments can be completed by instructing relevant hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiments.
[0142] In addition, in various embodiments of the present disclosure, each functional unit may be integrated into a processing module, or each unit may exist physically alone, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.
[0143] The above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disc, etc. Although the embodiments of the present disclosure have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present disclosure. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present disclosure.
Claims
1. A method for generating an image denoising model, characterized in that, The method includes: Obtain a training data set, where the training data set includes a first image and a corresponding labeled image; Input the first image into an initial image denoising model to obtain a second image; Determine a first edge image corresponding to the labeled image and a second edge image corresponding to the second image; Modify the initial image denoising model according to the difference between the labeled image and the second image and the difference between the first edge image and the second edge image to obtain a target image denoising model.
2. The method according to claim 1, wherein The determining the first edge image corresponding to the labeled image and the second edge image corresponding to the second image includes: Based on a fractional differential operator, perform edge detection on the labeled image and the second image respectively to obtain the first edge image corresponding to the labeled image and the second edge image corresponding to the second image.
3. The method according to claim 1, wherein The first image and the labeled image are in RAW format.
4. The method according to claim 3, characterized in that, The determining the first edge image corresponding to the labeled image and the second edge image corresponding to the second image includes: Convert the labeled image and the second image into RGB format respectively to obtain a third image corresponding to the labeled image and a fourth image corresponding to the second image; Based on a fractional differential operator, perform edge detection on the third image to obtain the first edge image; Based on the fractional differential operator, perform edge detection on the fourth image to obtain the second edge image.
5. The method according to claim 3, characterized in that The determining the first edge image corresponding to the labeled image and the second edge image corresponding to the second image includes: Convert the labeled image and the second image into RGB format respectively to obtain a third image corresponding to the labeled image and a fourth image corresponding to the second image; Convert the third image and the fourth image into YUV format respectively to obtain a fifth image corresponding to the third image and a sixth image corresponding to the fourth image; Based on a fractional differential operator, perform edge detection on the luminance channel of the fifth image to obtain the first edge image; Based on the fractional differential operator, perform edge detection on the luminance channel of the sixth image to obtain the second edge image.
6. The method according to any one of claims 1-5, characterized in that, The modifying the initial image denoising model according to the difference between the labeled image and the second image and the difference between the first edge image and the second edge image to obtain a target image denoising model includes: Determine a first loss value according to the difference between the labeled image and the second image; Determine a second loss value according to the difference between the first edge image and the second edge image; Determine a target loss value according to the first loss value and the second loss value; Based on the target loss value, modify the initial image denoising model to obtain the target image denoising model.
7. An image processing method, the method includes: Input an image to be processed into a target image denoising model to output a target image, where the target image denoising model is obtained according to the method of any one of claims 1-6.
8. An apparatus for generating an image denoising model, characterized in that, The device includes: An acquisition module, configured to acquire a training data set, where the training data set includes a first image and a corresponding label image; A processing module, configured to input the first image into an initial image denoising model to obtain a second image; A determination module, configured to determine a first edge image corresponding to the label image and a second edge image corresponding to the second image; A correction module, configured to correct the initial image denoising model according to the difference between the label image and the second image and the difference between the first edge image and the second edge image to obtain a target image denoising model.
9. The device according to claim 8, characterized in that, The determination module is configured to: Based on a fractional-order differential operator, perform edge detection on the label image and the second image respectively to obtain a first edge image corresponding to the label image and a second edge image corresponding to the second image.
10. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the method for generating an image denoising model according to any one of claims 1-6, or implements the image processing method according to claim 7.
11. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the method for generating an image denoising model according to any one of claims 1-6, or implements the image processing method according to claim 7.
12. A chip, characterized in that, It includes one or more interface circuits and one or more processors; the interface circuit is configured to receive a signal and send the signal to the processor, and the signal includes computer instructions. When the processor executes the computer instructions, the chip executes the steps of the method according to any one of claims 1-6.