Image denoising method, device and electronic equipment
By evaluating noise components and decomposing multi-bands in image data, and combining this with preset noise reduction intensity adjustment, the problems of uncontrollable noise reduction intensity and consistent frequency band adjustment in existing technologies are solved, thereby improving image quality.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- AXERA SEMICON (SHANGHAI) CO LTD
- Filing Date
- 2023-02-28
- Publication Date
- 2026-06-26
AI Technical Summary
In existing technologies, adjustable image noise reduction techniques based on conditional subnetworks result in unpredictable noise reduction strength, making it difficult to achieve controllability. Furthermore, noise reduction methods that are consistent across the entire frequency band cannot achieve frequency band separation and adjustment, leading to unsightly noise patterns after noise reduction.
Noise components are evaluated by acquiring image data, multi-band decomposition is performed, and each frequency band is adjusted based on different preset noise reduction intensities to reconstruct the spatial noise map and determine the target noise reduction map.
It achieves precise noise reduction of image data, improves image quality, ensures controllable noise reduction intensity and frequency band separation adjustment, and reduces noise residue.
Smart Images
Figure CN116188311B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of noise reduction technology, and more particularly to a method, apparatus and electronic device for image noise reduction. Background Technology
[0002] Currently, images inevitably contain some noise due to environmental limitations. Among related technologies, adjustable image noise reduction techniques are mainly based on conditional sub-network technology. However, due to the inherent uninterpretability of conditional sub-networks, the noise reduction intensity cannot be controlled in a white-box manner as expected, making it difficult to achieve adjustable noise reduction intensity. Furthermore, the methods for adjusting the noise reduction intensity of related images all adjust the entire frequency band consistently, failing to achieve frequency band-specific adjustments. This results in unsightly residual noise patterns on the denoised image. Therefore, a more intelligent image noise reduction method is urgently needed. Summary of the Invention
[0003] This application aims to address, at least to some extent, the technical problems in the related art.
[0004] According to a first aspect of this application, an image denoising method is provided, the method comprising: acquiring image data to be denoised; evaluating noise components in the image data to obtain a noise component map corresponding to the image data; performing multi-band decomposition on the noise component map to obtain noise maps of multiple frequency bands corresponding to the noise component map; performing different denoising intensity transformations on the noise maps of the multiple frequency bands based on multiple preset different denoising intensities to obtain multiple candidate noise maps with different denoising intensities; reconstructing a spatial noise map corresponding to the image data based on the multiple candidate noise maps with different denoising intensities; and determining a target denoised map corresponding to the image data based on the spatial noise map and the image data.
[0005] In one embodiment of this application, before obtaining the image data to be denoised, the method further includes: obtaining candidate image data to be denoised, and performing anomaly preprocessing on the candidate image data to obtain image data without anomalies.
[0006] In one embodiment of this application, the step of performing multi-band decomposition on the noise component map to obtain noise maps of multiple frequency bands corresponding to the noise component map includes: performing multi-band decomposition on the noise component map to obtain component maps of each frequency band; and determining noise maps of multiple frequency bands corresponding to the noise component map based on the component maps of each frequency band.
[0007] In one embodiment of this application, the step of performing different noise reduction intensity transformations on the noise maps of the multiple frequency bands based on a preset multiple different noise reduction intensity to obtain multiple candidate noise maps with different noise reduction intensity includes: determining noise reduction parameters corresponding to each noise reduction intensity based on a preset multiple different noise reduction intensity; and performing different noise reduction intensity transformations on the noise maps of the multiple frequency bands through the noise reduction parameters to obtain multiple candidate noise maps with different noise reduction intensity.
[0008] The image denoising method provided in this application involves acquiring image data to be denoised, evaluating noise components in the image data to obtain a noise component map corresponding to the image data, performing multi-band decomposition on the noise component map to obtain noise maps of multiple frequency bands corresponding to the noise component map, applying different denoising intensity transformations to the noise maps of multiple frequency bands based on multiple preset different denoising intensities to obtain multiple candidate noise maps with different denoising intensities, reconstructing the spatial domain noise map corresponding to the image data based on the multiple candidate noise maps with different denoising intensities, and processing the image data according to the spatial domain noise map to obtain the target denoised map corresponding to the image data. Thus, by adjusting the image data by frequency band and processing the image data in combination with the adjusted spatial domain noise map, the target denoised map corresponding to the image data is accurately determined, thereby improving the image quality.
[0009] According to a second aspect of the embodiments of this application, an image denoising apparatus is provided. The apparatus includes: an acquisition module for acquiring image data to be denoised; an evaluation module for evaluating noise components in the image data to obtain a noise component map corresponding to the image data; a decomposition module for performing multi-band decomposition on the noise component map to obtain noise maps of multiple frequency bands corresponding to the noise component map; a transformation module for performing different noise reduction intensity transformations on the noise maps of the multiple frequency bands based on multiple preset different noise reduction intensities to obtain multiple candidate noise maps with different noise reduction intensities; a reconstruction module for reconstructing a spatial noise map corresponding to the image data based on the multiple candidate noise maps with different noise reduction intensities; and a determination module for determining a target denoised map corresponding to the image data based on the spatial noise map and the image data.
[0010] In one embodiment of this application, the apparatus further includes: a preprocessing module, configured to acquire candidate image data to be denoised, and perform anomaly preprocessing on the candidate image data to obtain image data without anomalies.
[0011] In one embodiment of this application, the decomposition module is specifically used to: perform multi-band decomposition on the noise component map to obtain component maps of each frequency band; and determine noise maps of multiple frequency bands corresponding to the noise component map based on the component maps of each frequency band.
[0012] In one embodiment of this application, the transformation module is specifically used to: determine the noise reduction parameters corresponding to each noise reduction level based on a preset plurality of different noise reduction levels; and perform different noise reduction level transformations on the noise maps of multiple frequency bands through the noise reduction parameters to obtain a plurality of candidate noise maps with different noise reduction levels.
[0013] The image denoising apparatus provided in this application acquires image data to be denoised, evaluates the noise components of the image data to obtain a noise component map corresponding to the image data, performs multi-band decomposition on the noise component map to obtain noise maps of multiple frequency bands corresponding to the noise component map, performs different noise reduction intensity transformations on the noise maps of multiple frequency bands based on multiple preset different noise reduction intensities to obtain multiple candidate noise maps with different noise reduction intensities, reconstructs the spatial noise map corresponding to the image data based on the multiple candidate noise maps with different noise reduction intensities, and processes the image data according to the spatial noise map to obtain the target denoised map corresponding to the image data. Thus, by adjusting the image data by frequency band and processing the image data in combination with the adjusted spatial noise map, the target denoised map corresponding to the image data is accurately determined, thereby improving the image quality.
[0014] According to a third aspect of this application, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method described in the first aspect.
[0015] According to a fourth aspect of this application, a non-transitory computer-readable storage medium is provided storing computer instructions for causing the computer to perform the method described in the first aspect.
[0016] According to a fifth aspect of this application, a computer program product is provided, which, when executed by a processor, implements the method described in the first aspect.
[0017] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description
[0018] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:
[0019] Figure 1 This is a flowchart illustrating an image noise reduction method according to an exemplary embodiment.
[0020] Figure 2This is a flowchart illustrating an image noise reduction method according to another exemplary embodiment;
[0021] Figure 3 This is a flowchart illustrating an image noise reduction method according to another exemplary embodiment;
[0022] Figure 4 This is a structural diagram of an image noise reduction device according to an exemplary embodiment;
[0023] Figure 5 This is a schematic diagram of the structure of an image noise reduction device according to an exemplary embodiment. Detailed Implementation
[0024] The embodiments of this application are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.
[0025] The image noise reduction method, apparatus, and electronic device provided in this application will now be described in detail with reference to the accompanying drawings.
[0026] Figure 1 This is a schematic flowchart illustrating an image denoising method according to an exemplary embodiment. It should be noted that the execution entity of this image denoising method is an image denoising device, which can be installed in an electronic device. In some instances, the electronic device can be a server; this embodiment does not specifically limit the electronic device.
[0027] like Figure 1 As shown, the noise reduction method for this image may include the following steps:
[0028] Step 101: Obtain the image data to be denoised.
[0029] Optionally, the image data to be denoised can be obtained by a camera (sensor) or image data in some raw image files (RAW), but it is not limited to these, and this embodiment does not specifically limit it.
[0030] In some embodiments, before acquiring the image data to be denoised, the method further includes: acquiring candidate image data to be denoised, and performing anomaly preprocessing on the candidate image data to obtain image data without anomalies.
[0031] Specifically, RAW image data can be acquired and processed through Image Signal Processing (ISP) to remove abnormal image data and obtain processed image data in different data formats (RGB, YUV).
[0032] Abnormal image data may include, but is not limited to, missing image data, and this embodiment does not specifically limit it.
[0033] Step 102: Evaluate the noise components of the image data to obtain the noise component map corresponding to the image data.
[0034] In some embodiments, a neural network can be used to evaluate the noise components of the image data to obtain a noise component map corresponding to the image data, but this is not limited to this.
[0035] Specifically, a U-shaped convolutional neural network can be used to evaluate noise components in image data. This convolutional neural network can accurately evaluate noise component maps of the same size as the input image data based on the pixels of the input image data.
[0036] The U-shape-based convolutional neural network is trained using (noise, clean) paired data and supervised by a loss function (L1 loss). Empirically, the training noise intensity needs to be greater than the actual noise intensity, thus giving the U-shape convolutional neural network a larger adjustment space.
[0037] Step 103: Perform multi-band decomposition on the noise component map to obtain noise maps of multiple frequency bands corresponding to the noise component map.
[0038] In some embodiments, the noise component map can be decomposed into multiple frequency bands based on the frequency band decomposition technique of Laplace pyramid decomposition to obtain noise maps of multiple frequency bands corresponding to the noise component map, but it is not limited to this.
[0039] Specifically, when the noise component map is decomposed into multiple frequency bands using the Laplace pyramid decomposition technique, the noise component map can be decomposed into component maps of multiple frequency bands, thereby determining the noise map of each frequency band and achieving accurate determination of the noise map of multiple frequency bands.
[0040] Step 104: Based on multiple preset noise reduction intensities, apply different noise reduction intensities to the noise maps of multiple frequency bands to obtain multiple candidate noise maps with different noise reduction intensities.
[0041] Optionally, multiple preset noise reduction levels can be configured by the user, but are not limited to this.
[0042] Optionally, multiple different noise reduction strengths can be the same or different, and this embodiment does not specifically limit this.
[0043] In some embodiments, a greater noise reduction strength results in a greater noise reduction strength for the noise maps of multiple frequency bands, while a smaller noise reduction strength results in a smaller noise reduction strength for the noise maps of multiple frequency bands.
[0044] Step 105: Based on multiple candidate noise maps with different noise reduction intensities, reconstruct the spatial noise map corresponding to the image data.
[0045] In some embodiments, multiple candidate noise maps with different noise reduction strengths can be reconstructed using the Laplacian pyramid, but this is not the only option.
[0046] Specifically, the Laplacian pyramid can be used to reconstruct candidate noise maps with different noise reduction levels across multiple frequency bands layer by layer to establish the spatial noise map corresponding to the image data.
[0047] One approach to perform layer-by-layer reconstruction of candidate noise maps with different denoising strengths across multiple frequency bands is to add the corresponding residuals stored in the Laplacian pyramid to each image layer.
[0048] Step 106: Determine the target denoised map corresponding to the image data based on the spatial noise map and the image data.
[0049] In some embodiments, determining the target denoised image corresponding to the image data based on the spatial noise map and the image data can be achieved by adding the received image data to the spatial noise map to achieve a denoising effect, thereby obtaining a noise-free target denoised image.
[0050] The image denoising method provided in this application involves acquiring image data to be denoised, evaluating noise components in the image data to obtain a noise component map corresponding to the image data, performing multi-band decomposition on the noise component map to obtain noise maps of multiple frequency bands corresponding to the noise component map, applying different denoising intensity transformations to the noise maps of multiple frequency bands based on multiple preset different denoising intensities to obtain multiple candidate noise maps with different denoising intensities, reconstructing the spatial domain noise map corresponding to the image data based on the multiple candidate noise maps with different denoising intensities, and processing the image data according to the spatial domain noise map to obtain the target denoised map corresponding to the image data. Thus, by adjusting the image data by frequency band and processing the image data in combination with the adjusted spatial domain noise map, the target denoised map corresponding to the image data is accurately determined, thereby improving the image quality.
[0051] To clearly illustrate the previous embodiment, this embodiment also provides an image noise reduction method. Figure 2This is a schematic flowchart illustrating another image noise reduction method provided in an embodiment of this application.
[0052] like Figure 2 As shown, the noise reduction method for this image may include the following steps:
[0053] Step 201: Obtain the image data to be denoised.
[0054] Step 202: Evaluate the noise components of the image data to obtain the noise component map corresponding to the image data.
[0055] It should be noted that for a detailed description of steps 201 to 202, please refer to the relevant descriptions in the embodiments of this application, which will not be repeated here.
[0056] Step 203: Perform multi-band decomposition on the noise component map to obtain the component maps of each frequency band.
[0057] In some embodiments, when the noise component map is decomposed into multiple frequency bands using a frequency band decomposition technique based on Laplace pyramid decomposition, the resulting component maps for each frequency band are Laplace component maps and Gaussian component maps for each frequency band.
[0058] Step 204: Based on the component maps of each frequency band, determine the noise maps of multiple frequency bands corresponding to the noise component maps.
[0059] In some embodiments, when multi-band decomposition of the noise component map is performed using a frequency band decomposition technique based on Laplacian pyramid decomposition, noise maps of multiple frequency bands corresponding to the noise component map can be established based on the Laplacian component map and Gaussian component map of each frequency band.
[0060] Step 205: Based on multiple preset noise reduction intensities, apply different noise reduction intensity transformations to the noise maps of multiple frequency bands to obtain multiple candidate noise maps with different noise reduction intensities.
[0061] Step 206: Based on multiple candidate noise maps with different noise reduction intensities, reconstruct the spatial noise map corresponding to the image data.
[0062] Step 207: Determine the target denoised map corresponding to the image data based on the spatial noise map and the image data.
[0063] The image denoising method provided in this application embodiment acquires image data to be denoised, evaluates noise components in the image data to obtain a noise component map corresponding to the image data, decomposes the noise component map into multiple frequency bands to obtain component maps of each frequency band, determines multiple noise maps of multiple frequency bands corresponding to the noise component map based on the component maps of each frequency band, performs different noise reduction intensity transformations on the noise maps of multiple frequency bands based on multiple preset different noise reduction intensities to obtain multiple candidate noise maps with different noise reduction intensities, reconstructs the spatial noise map corresponding to the image data based on the multiple candidate noise maps with different noise reduction intensities, and processes the image data according to the spatial noise map to obtain the target denoised map corresponding to the image data. Thus, by adjusting the image data by frequency band and processing the image data in combination with the adjusted spatial noise map, multi-frequency band control of the image data is achieved to accurately determine the target denoised map corresponding to the image data.
[0064] Furthermore, to clearly illustrate the above embodiment, this embodiment also provides an image noise reduction method. Figure 3 This is a schematic flowchart illustrating another image noise reduction method provided in an embodiment of this application.
[0065] like Figure 3 As shown, the image noise reduction method includes the following steps:
[0066] Step 301: Obtain the image data to be denoised.
[0067] Step 302: Evaluate the noise components of the image data to obtain the noise component map corresponding to the image data.
[0068] Step 303: Perform multi-band decomposition on the noise component map to obtain noise maps of multiple frequency bands corresponding to the noise component map.
[0069] It should be noted that for a detailed description of steps 301 to 303, please refer to the relevant descriptions in the embodiments of this application, which will not be repeated here.
[0070] Step 304: Based on multiple preset noise reduction intensities, determine the noise reduction parameters corresponding to each noise reduction intensity.
[0071] In some embodiments, the noise reduction parameters corresponding to each noise reduction level can be data of the look-up table (Lut) type, but are not limited to this.
[0072] It is understandable that the greater the preset noise reduction intensity, the larger the corresponding noise reduction parameter; conversely, the smaller the preset noise reduction intensity, the smaller the corresponding noise reduction parameter.
[0073] Step 305: Apply different noise reduction intensity transformations to the noise maps of multiple frequency bands using noise reduction parameters to obtain multiple candidate noise maps with different noise reduction intensities.
[0074] In some embodiments, one implementation method for applying different noise reduction levels to noise maps of multiple frequency bands using noise reduction parameters can be as follows: when the noise reduction parameters corresponding to each noise reduction level are LUT control parameters, a lookup table is performed on the noise maps of multiple frequency bands according to the LUT control parameters to achieve the purpose of adjusting different noise reduction levels, thereby obtaining multiple candidate noise maps with different noise reduction levels.
[0075] Step 306: Based on multiple candidate noise maps with different noise reduction intensities, reconstruct the spatial noise map corresponding to the image data.
[0076] Step 307: Determine the target denoised map corresponding to the image data based on the spatial noise map and the image data.
[0077] The image denoising method provided in this application involves acquiring image data to be denoised, evaluating noise components in the image data to obtain a noise component map corresponding to the image data, performing multi-band decomposition on the noise component map to obtain noise maps of multiple frequency bands corresponding to the noise component map, determining denoising parameters corresponding to each denoising intensity based on multiple preset different denoising intensities, applying different denoising intensity transformations to the noise maps of multiple frequency bands through the denoising parameters to obtain multiple candidate noise maps with different denoising intensities, reconstructing the spatial domain noise map corresponding to the image data based on the multiple candidate noise maps with different denoising intensities, and processing the image data according to the spatial domain noise map to obtain the target denoised map corresponding to the image data. Thus, by adjusting the image data by frequency band and processing the image data in conjunction with the adjusted spatial domain noise map, denoising control of different denoising parameters in multiple frequency bands of the image data is achieved to accurately determine the target denoised map corresponding to the image data and improve image quality.
[0078] In summary, it is understandable that, in order to better implement the image denoising method of this application, this application also proposes an image denoising apparatus to implement the image denoising method, such as... Figure 4 As shown, Figure 4 The diagram shows the structure of an image denoising device, which includes an input module, a CNN noise estimation module, a frequency division intensity white-box adjustment component, a denoising module, and an output module. The frequency division intensity white-box adjustment component includes a noise frequency band decomposition module, a noise intensity adjustment module, and a noise reconstruction module.
[0079] Specifically, such as Figure 4As shown, the image data to be denoised is input into the CNN noise estimation module via the input module. The CNN noise estimation module evaluates the noise components of the received image data to obtain the noise component map corresponding to the image data, and sends the noise component map to the noise frequency band decomposition module in the frequency division intensity white box adjustment component. The noise frequency band decomposition module performs multi-band decomposition on the noise component map to obtain noise maps of multiple frequency bands (frequency band 1, frequency band 2, ..., frequency band n) corresponding to the noise component map, and sends the noise maps of multiple frequency bands to the noise intensity adjustment module in the frequency division intensity white box adjustment component. The noise intensity adjustment module is based on... Multiple preset noise reduction parameters (noise reduction parameter 1, noise reduction parameter 2, noise reduction parameter 3, ..., noise reduction parameter n) are used to apply different noise reduction force transformations to noise maps in multiple frequency bands to obtain multiple candidate noise maps with different noise reduction forces. These candidate noise maps with different noise reduction forces are then sent to the noise reconstruction module in the frequency division force white-box adjustment component. The noise reconstruction module reconstructs the spatial noise map corresponding to the image data and sends the spatial noise map to the noise reduction module. Based on the spatial noise map and the image data, the noise reduction module determines the target noise reduction map corresponding to the image data and then outputs the target noise reduction map through the output module.
[0080] To achieve the above embodiments, this application also proposes an image noise reduction device.
[0081] Figure 5 This is a schematic diagram of the structure of an image noise reduction device according to an exemplary embodiment.
[0082] like Figure 5 As shown, the image noise reduction device 500 includes an acquisition module 501, an evaluation module 502, a decomposition module 503, a transformation module 504, a reconstruction module 505, and a determination module 506, wherein:
[0083] The acquisition module 501 is used to acquire the image data to be denoised;
[0084] Evaluation module 502 is used to evaluate the noise components of the image data to obtain a noise component map corresponding to the image data;
[0085] The decomposition module 503 is used to perform multi-band decomposition on the noise component map to obtain noise maps of multiple frequency bands corresponding to the noise component map.
[0086] The transformation module 504 is used to transform the noise maps of the multiple frequency bands with different noise reduction intensities based on multiple preset noise reduction intensities, so as to obtain multiple candidate noise maps with different noise reduction intensities.
[0087] The reconstruction module 505 is used to reconstruct the spatial noise map corresponding to the image data based on the multiple candidate noise maps with different noise reduction intensities.
[0088] The determining module 506 is used to determine the target denoised map corresponding to the image data based on the spatial noise map and the image data.
[0089] In one embodiment of this application, the apparatus further includes: a preprocessing module, configured to acquire candidate image data to be denoised, and perform anomaly preprocessing on the candidate image data to obtain image data without anomalies.
[0090] In one embodiment of this application, the decomposition module 503 is specifically used to: perform multi-band decomposition on the noise component map to obtain component maps of each frequency band;
[0091] Based on the component maps of each frequency band, noise maps of multiple frequency bands corresponding to the noise component maps are determined.
[0092] In one embodiment of this application, the transformation module 504 is specifically used for:
[0093] Based on multiple preset noise reduction intensities, determine the noise reduction parameters corresponding to each noise reduction intensity;
[0094] By applying different noise reduction levels to the noise maps of multiple frequency bands using the noise reduction parameters, multiple candidate noise maps with different noise reduction levels can be obtained.
[0095] It should be noted that the explanation of the above-described image denoising method embodiments also applies to the image denoising device in this embodiment, and this embodiment will not repeat the above.
[0096] The image denoising apparatus of this application acquires image data to be denoised, evaluates the noise components of the image data to obtain a noise component map corresponding to the image data, performs multi-band decomposition on the noise component map to obtain noise maps of multiple frequency bands corresponding to the noise component map, performs different noise reduction intensity transformations on the noise maps of multiple frequency bands based on multiple preset different noise reduction intensities to obtain multiple candidate noise maps with different noise reduction intensities, reconstructs the spatial noise map corresponding to the image data based on the multiple candidate noise maps with different noise reduction intensities, and processes the image data according to the spatial noise map to obtain the target denoised map corresponding to the image data. Thus, by adjusting the image data by frequency band and processing the image data in combination with the adjusted spatial noise map, the target denoised map corresponding to the image data is accurately determined, thereby improving the image quality.
[0097] To implement the above embodiments, this application also proposes an electronic device, comprising:
[0098] At least one processor; and
[0099] A memory communicatively connected to the at least one processor; wherein,
[0100] The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the aforementioned method.
[0101] To implement the above embodiments, this application also proposes a non-transitory computer-readable storage medium storing computer instructions for causing the computer to perform the aforementioned method.
[0102] To implement the above embodiments, this application also proposes a computer program product, including a computer program that, when executed by a processor, implements the method described above.
[0103] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.
[0104] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.
[0105] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.
[0106] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.
[0107] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.
[0108] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.
[0109] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.
[0110] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.
Claims
1. A method for denoising images, characterized in that, The method includes: Acquire the image data to be denoised; The image data is evaluated for noise components to obtain a noise component map corresponding to the image data; Decomposing the noise component map into multiple frequency bands to obtain noise maps of multiple frequency bands corresponding to the noise component map includes: decomposing the noise component map into multiple frequency bands to obtain component maps of each frequency band; and determining the noise maps of multiple frequency bands corresponding to the noise component map based on the component maps of each frequency band. Based on multiple preset noise reduction intensities, different noise reduction intensities are applied to the noise maps of the multiple frequency bands to obtain multiple candidate noise maps with different noise reduction intensities. Based on the multiple candidate noise maps with different noise reduction intensities, the spatial noise map corresponding to the image data is reconstructed. In this process, the candidate noise maps with different noise reduction intensities in multiple frequency bands are reconstructed layer by layer using the Laplacian pyramid to establish the spatial noise map corresponding to the image data. The method of reconstructing the candidate noise maps with different noise reduction intensities in multiple frequency bands layer by layer is to add the residual stored in the Laplacian pyramid to each image layer. Based on the spatial noise map and the image data, determine the target denoising map corresponding to the image data.
2. The method as described in claim 1, characterized in that, Before acquiring the image data to be denoised, the method further includes: Acquire candidate image data to be denoised, and perform anomaly preprocessing on the candidate image data to obtain anomaly-free image data.
3. The method as described in claim 1, characterized in that, The method involves applying different noise reduction intensity transformations to the noise maps of the multiple frequency bands based on multiple preset noise reduction intensities to obtain multiple candidate noise maps with different noise reduction intensities, including: Based on multiple preset noise reduction intensities, determine the noise reduction parameters corresponding to each noise reduction intensity; By applying different noise reduction levels to the noise maps of multiple frequency bands using the noise reduction parameters, multiple candidate noise maps with different noise reduction levels can be obtained.
4. An image noise reduction device, characterized in that, The device includes: The acquisition module is used to acquire the image data to be denoised; An evaluation module is used to evaluate the noise components of the image data to obtain a noise component map corresponding to the image data. The decomposition module is used to perform multi-band decomposition on the noise component map to obtain noise maps of multiple frequency bands corresponding to the noise component map. Specifically, it is used to: perform multi-band decomposition on the noise component map to obtain component maps of each frequency band; and determine the noise maps of multiple frequency bands corresponding to the noise component map based on the component maps of each frequency band. The transformation module is used to transform the noise maps of the multiple frequency bands with different noise reduction intensities based on multiple preset noise reduction intensities, so as to obtain multiple candidate noise maps with different noise reduction intensities. The reconstruction module is used to reconstruct the spatial noise map corresponding to the image data based on the multiple candidate noise maps with different noise reduction intensities. The reconstruction operation is performed layer by layer on the candidate noise maps with different noise reduction intensities in multiple frequency bands through the Laplacian pyramid to establish the spatial noise map corresponding to the image data. The method of performing the layer by layer reconstruction operation on the candidate noise maps with different noise reduction intensities in multiple frequency bands is to add the corresponding image pairs to the residuals stored in the Laplacian pyramid for each layer. The determination module is used to determine the target denoised map corresponding to the image data based on the spatial noise map and the image data.
5. The apparatus as described in claim 4, characterized in that, The device further includes: The preprocessing module is used to acquire candidate image data to be denoised and to perform anomaly preprocessing on the candidate image data to obtain image data without anomalies.
6. The apparatus as claimed in claim 4, characterized in that, The transformation module is specifically used for: Based on multiple preset noise reduction intensities, determine the noise reduction parameters corresponding to each noise reduction intensity; By applying different noise reduction levels to the noise maps of multiple frequency bands using the noise reduction parameters, multiple candidate noise maps with different noise reduction levels can be obtained.
7. An electronic device, characterized in that, include: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-3.
8. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-3.