Image denoising processing method and system fusing wavelet frame and sharpening operator

Abstract

The invention relates to the technical field of image noise processing, and provides an image denoising processing method and system fusing a wavelet frame and a sharpening operator, and the method comprises the steps: carrying out the sharpening processing of an obtained to-be-processed image; wavelet domain transformation is carried out on the sharpened image, and a denoising preprocessing mathematical model is constructed with the purpose of minimizing the weighted sum of a deblurring error item, an image sharpening consistency item and a wavelet domain sparse regular item; solving of the denoising preprocessing mathematical model is decomposed into a plurality of sub-problems, an alternating direction multiplier method is adopted for solving, and a restored image after denoising processing is obtained. According to the method, wavelet multi-scale analysis and sharpening operator edge enhancement are fused, an optimization model for minimizing the deblurring error, sharpening consistency and sparse regularization is constructed, and an alternating direction multiplier method and soft threshold processing are adopted, so that the image denoising effect and the detail retention capability are effectively improved; the problems of detail loss, fuzzy aggravation and low calculation efficiency in the image denoising processing process are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image noise processing, in particular to an image denoising processing method and system fusing wavelet framework and sharpening operator. BACKGROUND

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

[0003] In the field of image processing, removing noise and enhancing details are the basic steps to improve image quality, improve visual experience, and improve the accuracy of subsequent analysis (such as target detection and recognition). In practical applications, images are often contaminated by different types of noise (such as impulse noise). The generation of impulse noise is closely related to image acquisition, transmission, storage, etc. It is a common type of noise in image processing, which appears as isolated pixels in the image. It has multidimensional damage to image quality and also limits the accuracy of subsequent image algorithms. Therefore, removing impulse noise is crucial.

[0004] At present, the existing image noise processing method has the following defects: (1) Traditional linear filtering methods include mean filtering and median filtering. While removing noise, they often ignore high-frequency information in the image, such as texture and sharp edges. In a high noise density environment (such as more than 50%), median filtering is prone to misjudging image effective details as noise, resulting in blurred edges and lost details.

[0005] (2) For image deblurring, the common way is to introduce a regularization term, which can regularize the solution process and optimize the feasible solution set. In the deblurring method, the Rudin-Osher-Fatemi (ROF) model is most widely used. The Rudin-Osher-Fatemi (ROF) model proposes a method based on total variation (TV) regularization, which can preserve image edges to some extent. However, the ROF model has the defect of being unable to distinguish between high-frequency oscillations (clear details) and smooth regions (fuzzy parts), which limits the deblurring effect.

[0006] Another deblurring method is based on the diffusion method of an impact filter, including backward diffusion, forward diffusion, and anisotropic diffusion. Although it can enhance image edges, it often amplifies noise in the sharpening process due to its inherent instability, affecting image quality. At the same time, the existing wavelet domain threshold method often misjudges high-frequency details as noise when processing impulse noise, resulting in excessive suppression, causing image blurring or artifacts (such as Gibbs phenomenon). Hard thresholding methods can damage image continuity, while soft thresholding methods can easily introduce bias, further weakening detail expression. SUMMARY

[0007] The present application proposes an image denoising processing method and system fusing wavelet framework and sharpening operator to solve the above problems, which fuses multi-scale analysis of wavelet framework and edge enhancement of sharpening operator, and solves the problems of traditional methods in detail reservation and calculation efficiency, etc.

[0008] In order to achieve the above object, the present application adopts the following technical scheme: The present application provides an image denoising processing method fusing wavelet framework and sharpening operator in the first aspect, comprising the following steps: The obtained image to be processed is sharpened based on a sharpening operator to obtain a sharpened image; The sharpened image is subjected to wavelet domain transformation, a denoising preprocessing mathematical model is constructed with the goal of minimizing the weighted sum of a deblurring error term, an image sharpening consistency term and a wavelet domain sparse regularization term; The solution of the denoising preprocessing mathematical model is decomposed into several sub-problems, the denoising preprocessing mathematical model is solved by using an alternating direction multiplier method, the image to be processed is decomposed into detail components of different scales, sparse constraint solving is performed, soft threshold processing is used to remove noise and redundant information, and a restored image after denoising processing is obtained.

[0009] The present application provides an image denoising processing system fusing wavelet framework and sharpening operator in the second aspect, comprising: A sharpening module is configured to sharpen the obtained image to be processed based on a sharpening operator to obtain a sharpened image; A model construction module is configured to subject the sharpened image to wavelet domain transformation, and construct a denoising preprocessing mathematical model with the goal of minimizing the weighted sum of a deblurring error term, an image sharpening consistency term and a wavelet domain sparse regularization term; A solution module is configured to decompose the solution of the denoising preprocessing mathematical model into several sub-problems, solve the denoising preprocessing mathematical model by using an alternating direction multiplier method, decompose the image to be processed into detail components of different scales, perform sparse constraint solving, use soft threshold processing to remove noise and redundant information, and obtain a restored image after denoising processing.

[0010] The present application provides an image denoising processing system fusing wavelet framework and sharpening operator in the third aspect, comprising an image acquisition device and a processor. The image acquisition device is used to acquire an image to be processed. The processor is configured to perform the steps of the image denoising processing method fusing wavelet framework and sharpening operator.

[0011] Compared with the prior art, the present application has the following advantages: (1) The present application first enhances the edges and high-frequency details based on a sharpening operator, effectively realizing better thread detail expression; then, through wavelet domain transformation, the sharpened image is decomposed into multiple scale dimension detail components, in the wavelet domain, an optimization model is constructed, and the deblurring error term (anti-blurring), the sharpening consistency term (to ensure that the edges are not damaged) and the sparse regularization term (to promote better sparse expression) are balanced; then, the alternating direction multiplier method (ADMM) is used to decompose the processing model into multiple sub-problems that are easy to solve, to gradually improve the processing effect and improve the data processing efficiency; finally, the decomposed detail components are sparsely constrained, and the soft threshold processing is used to select the appropriate threshold to remove noise and redundant information, and to realize the performance optimization of the restored image.

[0012] (2) The image denoising processing method of the present application can effectively maintain the image edges and detail organization in a high-noise environment, and avoid the loss of details caused by traditional cleaning methods; at the same time, focusing on the sparsity of the wavelet domain, the method effectively reduces the blurring and pseudo-phenomenon caused by excessive suppression, improves the visual effect after processing, and improves the subsequent analysis accuracy; the application of the alternating direction multiplier method helps to increase the stability of the solution, reduce the computational complexity, and effectively support the real-time image processing requirements.

[0013] The advantages of the present application and the advantages of the additional aspects will be described in detail in the following specific embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0014] The drawings accompanying the specification of the present application serve to provide a further understanding of the present application, and the illustrative embodiments of the present application and their descriptions serve to explain the present application, and do not constitute a limitation thereof.

[0015] Figure 1 is a flowchart of the image denoising processing method of embodiment 1 of the present application; Figure 2 is a test example processing process schematic diagram of the image denoising processing method of embodiment 1 of the present application; Figure 3 is an original gray test image in the test example of embodiment 1 of the present application; Figure 4 is a noise image after pulse noise and blur processing of the original gray test image in the test example of embodiment 1 of the present application; Figure 5 is a sharpened image after sharpening enhancement of the noise image in the test example of embodiment 1 of the present application; Figure 6is the final recovered image in the test example of embodiment 1 of the present application, solved by the denoising preprocessing mathematical model; Figure 7 is the difference image including image details of edge enhancement amount in the test example of embodiment 1 of the present application. DETAILED DESCRIPTION

[0016] The present application will be further described below in conjunction with the accompanying drawings and embodiments.

[0017] It should be noted that the following detailed description is exemplary in nature and is intended to provide further description of the present application. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0018] It should be noted that the terms used herein are only intended to describe specific embodiments and are not intended to limit exemplary embodiments according to the present application. As used herein, the singular form is intended to include the plural form unless the context clearly indicates otherwise, and it should also be understood that when the terms "comprise" and / or "include" are used in the specification, they indicate the presence of the features, steps, operations, devices, components and / or combinations thereof. It should be noted that the various embodiments and features in the present application can be combined with each other without conflict. The embodiments will be described in detail below in conjunction with the accompanying drawings.

[0019] Embodiment 1 In one or more embodiments of the technical solutions disclosed, Figures 1 to 7 As shown in the technical solutions disclosed in one or more embodiments, an image denoising processing method fusing a wavelet framework and a sharpening operator includes the following steps: Step 1, based on the sharpening operator, sharpening processing is performed on the obtained image to be processed to obtain a sharpened image; Step 2, wavelet domain transformation is performed on the sharpened image, and a denoising preprocessing mathematical model is constructed to minimize the weighted sum of the deblurring error term, the image sharpening consistency term and the wavelet domain sparse regularization term as the target; Step 3, the solution of the denoising preprocessing mathematical model is decomposed into several sub-problems, the denoising preprocessing mathematical model is solved by using the alternating direction multiplier method, and the image to be processed is decomposed into different scale detail components, sparse constraint solving is performed, soft threshold processing is used to remove noise and redundant information, and a restored image after denoising processing is obtained; In the above scheme, more image details are obtained by using an image sharpening operator, which can be used as a preprocessing step. The edge enhancement is realized by the sharpening operator, which can enhance the image features and edge information in advance, highlight the main structural features of the image, and solve the problem of blurred image edges and lost details after processing by traditional filtering algorithms. In the constructed denoising preprocessing mathematical model, the image sharpening consistency term is set to make the denoised image as close as possible to the structure of the sharpened image, so as to ensure that the feature information is not excessively smoothed, and the high-frequency coefficients corresponding to the edges are dynamically enhanced by the wavelet frame sharpening operator to preserve the details in the denoising process. Then, the wavelet frame method is used to further improve the image restoration result. The image to be processed is decomposed into different scale detail components, and sparse constraint solving is performed. The soft threshold value processing is used to remove noise and redundant information. In step 1, the image to be processed is sharpened, including the following steps: Step 11, the obtained image to be processed is blurred to obtain a blurred image; Optionally, Gaussian filtering, mean filtering and bilateral filtering methods can be used. Preferably, in this embodiment, the Gaussian filtering method is used to obtain the blurred image. The convolution operation of the image to be processed and the Gaussian kernel is performed to realize the smoothing of the local region, suppress the high-frequency details, and achieve the blurring effect. Step 12, calculate the difference between the original image to be processed and the blurred image to obtain image details including edge enhancement; Step 13, superimpose the image details to the original image to be processed to obtain an enhanced image; In this embodiment, the important features and edge detail regions of the image can be determined in advance through the above sharpening process. By highlighting these key regions, it is easier to protect and identify meaningful information during subsequent wavelet denoising or regularization model optimization, and to avoid excessive suppression or elimination of useful details.

[0020] Further technical solutions, a denoising preprocessing mathematical model is constructed using wavelet transform, as follows: ; Wherein, is a blurring operator, is a restored image without noise and blurring, is a noise image, i.e. an image to be processed; is an image sharpening operator, is a sharpened image that can enhance the key features of the image; is a wavelet transform, and satisfies , is an identity matrix, denotes norm; are two positive parameters.

[0021] The embodiment constructs and proposes a de-noising preprocessing mathematical model as a convex optimization model, which can ensure the successful implementation of a convergent efficient numerical algorithm, and provides a powerful tool for the actual operation of image deblurring.

[0022] The de-noising preprocessing mathematical model includes the following items: (1) Deblurring error term : The L1 norm of the difference between the blurred image and the noise image is used to ensure the consistency of the restored image and the noise image after deblurring operation, and to ensure the consistency as much as possible; (2) Image sharpening consistency term : By introducing the L1 norm of the difference between the original observation image and the sharpened image, and multiplying it by the regularization coefficient μ, as an additional constraint, it is used to enhance the structural detail consistency between the restored image and the sharpened noise image; (3) Wavelet domain sparse regularization term : The wavelet transform is used to realize the sparse representation of the image, which is used to suppress noise and highlight real details. It can better distinguish noise and signal, and facilitate the realization of more accurate de-noising effect; In the embodiment, in order to solve the problem that the sharpening image as a preprocessing step may cause the image to be over-sharpened and distorted, the image sharpening consistency term is added to the constructed de-noising preprocessing mathematical model. By minimizing the image sharpening consistency term, the restored image can retain the overall structural features of the original observation image while fully referencing the edge and detail information revealed by the sharpened image. This design effectively buffers the abnormal enhancement effect that may exist in the sharpened image, while improving the edge definition and structural consistency of the restored result.

[0023] In step 3, the alternating direction multiplier method is used to solve the de-noising preprocessing mathematical model, which decomposes the solution of the de-noising preprocessing mathematical model into several sub-problems, including the following steps: Step 31, based on the sharpened image, the de-noising preprocessing mathematical model is simplified and equivalently transformed, and the original problem of the de-noising preprocessing mathematical model is converted into a constrained problem, that is, the de-noising preprocessing mathematical model is converted into an equivalent form with constraints; Let the sharpened image The de-noising preprocessing mathematical model can be rewritten as: ; The above de-noising preprocessing mathematical model is rewritten in an equivalent way, and the equivalent form with constraints is as follows: ; ; wherein d is a wavelet domain sparsity constraint variable; z is a detail consistency constraint variable; t is a deblurring error constraint variable; Step 32, using the augmented Lagrange method, a Lagrange multiplier is introduced for each constraint, and a penalty function term is introduced to strengthen the constraint condition, and an augmented Lagrange expression of the denoising preprocessing mathematical model is constructed, and a plurality of sub-problems are obtained, and the specific augmented Lagrange expression is as follows: ; wherein, , is a penalty parameter, , is a Lagrange multiplier.

[0024] Specifically, the sub-problems obtained by the augmented Lagrange method include a restored image solving sub-problem (u sub-problem), a wavelet domain sparsity constraint solving sub-problem (d sub-problem), a detail consistency solving sub-problem (z sub-problem), a deblurring error solving sub-problem (t sub-problem), and a Lagrange multiplier (b, a, c) solving sub-problem; In this embodiment, the augmented Lagrange method is used to introduce a Lagrange multiplier for each constraint condition, and a penalty function term is introduced to strengthen the constraint condition, thereby constructing an augmented Lagrange expression of the denoising preprocessing mathematical model. Combined with the multi-scale gradient constraint optimization reconstruction process, the adaptability of the local gradient change characteristics under different scales is regulated, and the generation problem of common artifacts (such as Gibbs phenomenon, over-sharpening texture distortion, etc.) in traditional methods is effectively suppressed. While ensuring the clarity of the image details, false textures and edge burrs can be suppressed, so that the final restored image achieves significant improvement in subjective visual effect and objective evaluation index (such as PSNR, SSIM), and the overall image result is more natural, continuous and structurally complete, meeting the actual application requirements of high-quality image denoising and sharpening.

[0025] After introducing the augmented Lagrange function, the entire denoising preprocessing mathematical model is solved step by step by the alternating direction multiplier method (ADMM). The alternating direction multiplier method is used to solve the denoising preprocessing mathematical model. Specifically, each time the update is in the order of restored image u, wavelet domain sparsity constraint variable d, Lagrange multiplier b, detail consistency constraint variable z, Lagrange multiplier c, deblurring error constraint variable t, and Lagrange multiplier a, wherein the restored image u is solved and updated by fast Fourier transform, the wavelet domain sparsity constraint variable d, the detail consistency constraint variable z, and the deblurring error constraint variable t are solved by a soft threshold operator, and the steps are as follows: Step 33.1 initialization: initialize the optimization variables corresponding to each sub-problem, including the image restoration variable wavelet sparsity variable detail consistency variable deblurring error variable and corresponding initial value of Lagrange multiplier ; Step 33.2, perform fast Fourier transform on the sharpened image, fix other variables in the augmented Lagrange expression except for the image u, solve the augmented Lagrange expression, and obtain the restored image of the (k+1)th iteration based on the variable solution of the kth iteration sub-problem ; The sub-problem about u is to solve the formula iteratively by fixing other variables to minimize the objective function about u:

[0026] ; wherein, is the value of the solved Lagrange multiplier of the kth iteration, is the value of the wavelet domain sparsity constraint variable of the kth iteration; is the value of the detail consistency constraint variable of the kth iteration; is the value of the deblurring error constraint variable of the kth iteration; For the sub-problem of , if it is differentiable, the optimality condition of the above formula satisfies: ; Solve by using fast Fourier transform to obtain: ; wherein, denotes two-dimensional discrete Fourier transform, is a complex conjugate; denotes element-wise multiplication.

[0027] Step 33.3, fix other variables in the augmented Lagrange expression except for the wavelet domain sparsity constraint variable d to the current updated value, and solve the wavelet domain sparsity constraint variable of the (k+1)th iteration by using a soft threshold operator; In this step, the current updated value refers to the updated variable after the execution of the previous step. Specifically, in this step, the previous step is step 33.2, and the current updated value of the restored image is the restored image of the (k+1)th iteration , and other variables are not updated in step 33.2, so the remaining variables are the variable values of the kth iteration: ; the explanation of the current updated value is applicable in subsequent steps; Specifically, fixing all variables in the augmented Lagrangian expression except for the wavelet domain sparsity constraint variable d, the augmented Lagrangian expression is solved. By applying the soft thresholding operator, based on the restored image of the (k+1)th iteration and the variables of other variables in the subproblem of the kth iteration (…), the results are obtained. The solution yields the wavelet sparse representation for the (k+1)th iteration. ; Regarding the sparsity constraint variables in the wavelet domain Solving the subproblems, including the terms of the augmented Lagrangian expression with respect to the wavelet domain sparsity constraint variables, yields the subproblem-solving function for d:

[0028] ; Using soft threshold operator Solve :

[0029] in,

[0030]

[0031] During the solution process, let: ; ; Step 33.4 Based on the obtained wavelet domain sparsity constraint variables of the (k+1)th iteration and the restored image of the (k+1)th iteration By updating through addition, we obtain the Lagrange multipliers for the (k+1)th iteration. The updated formula is as follows: ; Step 33.5 Fix all variables in the augmented Lagrange expression except for the detail consistency constraint variable z as the current updated values, solve the augmented Lagrange expression, and use the soft threshold operator to obtain the detail consistency constraint variable for the (k+1)th iteration. ;

[0032] ; Using soft threshold operator Solve :

[0033] Step 33.6 Based on the obtained detail consistency constraint variables of the (k+1)th iteration and the deblurred image of the k+1th iteration and the sharpened image , the Lagrange multiplier of the k+1th iteration is obtained by addition update , the update formula is: ; Step 33.7, the current updated values of the variables in the augmented Lagrange expression except the deblurring error constraint variable t are fixed, the augmented Lagrange expression is solved, the soft threshold operator is used to solve the deblurring error constraint variable of the k+1th iteration; For the sub-problem of , the sub-problem about , the solution function is:

[0034] ; The soft threshold operator is used to solve : ; Step 33.8, based on the obtained deblurring error constraint variable and the deblurred image of the k+1th iteration and the image to be processed , the Lagrange multiplier of the k+1th iteration is obtained by addition update ;

[0035] Step 33.8, based on the obtained deblurred image, the objective function is calculated, when the objective function value does not meet the requirement, the next loop iteration is performed in step 33.2, until the objective function value meets the stopping iteration condition, the deblurred image u is obtained. The stopping iteration condition is: ; Wherein, represents the function value of the deblurred image of the kth iteration.

[0036] The embodiment innovatively combines image sharpening with regularization-based methods in the image deblurring task, and the effect obtained by this combination is significantly better than that of any single method. A large number of experimental studies are carried out, and the experimental process is described below. In the test verification of the whole method, a gray test image can be obtained, the test image is added with pulse noise and blur processing, and the process of steps 1 to 3 is performed on the image to be processed, that is, the noise image. First, the noise image is preprocessed by the image sharpening operator of step 1, and then the sharpened image is processed by the denoising preprocessing mathematical model constructed in step 2. The denoising preprocessing mathematical model is iteratively solved based on the alternating direction multiplier method to obtain a restored image, that is, the original gray test image.

[0037] The specific test example processing process is shown in the figure Figures 3 to 7 The original image of the test verification is a gray test image, as shown in Figure 3 Figure 4 is an image processed by adding pulse noise and blur, and is a noise image to be processed by the method process of the embodiment. As shown in Figure 7 The difference image obtained after step 12 is shown in Figure 5 is a sharpened image after sharpening enhancement of step 1. Figure 6 The image is finally restored by solving the denoising preprocessing mathematical model. The restored image of Figure 6 and the original image before adding noise Figure 3 can be compared. It can be seen that through the denoising and deblurring scheme of the embodiment, the details in the original image can be preserved, and a clearer restored image can be obtained.

[0038] The method of the embodiment is suitable for a variety of mixed noise interference and high requirement for detail feature image processing tasks. For example, in medical image processing (such as MRI, CT and ultrasound images), the influence of superimposed pulse noise and Gaussian noise can be effectively suppressed, and the lesion edge and tissue details can be accurately preserved; in the field of satellite remote sensing and multispectral imaging, sensor pulse noise and atmospheric scattering Gaussian noise can be removed at the same time, and the accuracy of feature classification and interpretation can be improved; in industrial detection, the method highlights the features of small defects such as cracks and scratches on the metal surface; in low-light photography and consumer electronics applications, the method effectively balances the removal of pulse-type thermal noise and Gaussian-type read noise and image detail preservation under high ISO conditions; in real-time video monitoring and traffic recognition field, deblurring and denoising processing can be performed simultaneously to improve the accuracy of target recognition and tracking in dynamic environment. Through adaptive application in different scenes, the excellent detail preservation and robust denoising performance are fully demonstrated.

[0039] Embodiment 2 Based on embodiment 1, an image denoising processing system fusing a wavelet framework and a sharpening operator is provided in the embodiment, comprising: A sharpening module configured to sharpen the acquired image to be processed based on a sharpening operator to obtain a sharpened image.​ A model construction module is configured to perform wavelet domain transformation on the sharpened image to construct a denoising pre-processing mathematical model aiming at minimizing a weighted sum of a deblurring error term, an image sharpening consistency term and a wavelet domain sparsity regularization term; A solution module is configured to decompose the solution of the denoising pre-processing mathematical model into several sub-problems, solve the denoising pre-processing mathematical model by using an alternating direction multiplier method, decompose the image to be processed into detail components of different scales, perform sparse constraint solving, remove noise and redundant information by using soft threshold processing, and obtain a restored image after denoising processing.

[0040] It should be noted that each module in the embodiment corresponds to each step in Embodiment 1 one by one, and the specific implementation process is the same, which will not be repeated here.

[0041] Embodiment 3 Based on Embodiment 1, the embodiment provides an image denoising processing system fusing a wavelet framework and a sharpening operator, comprising an image acquisition device and a processor. The image acquisition device is used to acquire an image to be processed. The image acquisition device can be a camera or the like. The processor is configured to perform the steps of the image denoising processing method fusing the wavelet framework and the sharpening operator according to Embodiment 1.

[0042] The above only describes the preferred embodiments of the present application and is not used to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

[0043] The above describes the specific embodiments of the present application in combination with the accompanying drawings, but is not a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications or changes made on the basis of the technical solutions of the present application without creative labor are still within the protection scope of the present application.

Claims

1. An image denoising method integrating wavelet framework and sharpening operator, characterized in that, Includes the following steps: The acquired image to be processed is sharpened based on the sharpening operator to obtain the sharpened image; The sharpened image is subjected to wavelet domain transformation. The goal is to minimize the weighted sum of the deblurring error term, the image sharpening consistency term, and the wavelet domain sparse regularization term to construct a denoising preprocessing mathematical model. The denoising preprocessing mathematical model is decomposed into several sub-problems. The alternating direction multiplier method is used to solve the denoising preprocessing mathematical model. The image to be processed is decomposed into detail components of different scales and sparse constraint is solved. Soft thresholding is used to remove noise and redundant information to obtain the denoised restored image.

2. The image denoising method fusing wavelet framework and sharpening operator as described in claim 1, characterized in that, Sharpening the image to be processed includes the following steps: The acquired image to be processed is blurred to obtain a blurred image; Calculate the difference between the original image to be processed and the blurred image to obtain image details including edge enhancement. The enhanced image is obtained by overlaying image details onto the original image to be processed.

3. The image denoising method fusing wavelet framework and sharpening operator as described in claim 1, characterized in that, The mathematical model for denoising preprocessing includes: Deblurring error term: The L1 norm of the difference between the blurred image and the noisy image is calculated to ensure the consistency between the restored image and the noisy image after the deblurring operation. Image sharpening consistency term: An additional constraint is introduced by multiplying the L1 norm of the difference between the restored and sharpened images by a regularization coefficient μ. Wavelet domain sparse regularization: Wavelet transform is used to restore the sparse representation of the image, which is used to suppress noise and highlight details.

4. The image denoising method fusing wavelet framework and sharpening operator as described in claim 3, characterized in that, The mathematical model for denoising preprocessing is as follows: ； in, It is a fuzzy operator. It is a restored image with noise and blur removed. The image is a noisy image, i.e., the image to be processed. For image sharpening operators, To sharpen the image; It is a wavelet transform and satisfies , It is the identity matrix. Represented as Norm; There are two positive parameters.

5. The image denoising method fusing wavelet framework and sharpening operator as described in claim 1, characterized in that, The alternating direction multiplier method is used to solve the denoising preprocessing mathematical model, which is decomposed into several sub-problems, including the following steps: Based on the sharpened image, the mathematical model of denoising preprocessing is simplified and equivalently transformed, and the original problem of the mathematical model of denoising preprocessing is transformed into a constrained problem, that is, the mathematical model of denoising preprocessing is transformed into an equivalent form with constraints. The augmented Lagrange method is adopted, and Lagrange multipliers are introduced for each constraint. At the same time, a penalty function term is introduced to strengthen the constraint conditions. The augmented Lagrange expression of the denoising preprocessing mathematical model is constructed, and multiple subproblems are obtained.

6. The image denoising method fusing wavelet framework and sharpening operator as described in claim 5, characterized in that: The subproblems obtained by the augmented Lagrange method include solving the image restoration problem, solving the wavelet domain sparsity constraint problem, solving the detail consistency problem, solving the deblurring error problem, and solving the Lagrange multiplier problem.

7. The image denoising method fusing wavelet framework and sharpening operator as described in claim 1, characterized in that: The alternating direction multiplier method is used to solve the denoising preprocessing mathematical model. Specifically, each update is performed in the following order: restored image u, wavelet domain sparsity constraint variable d, Lagrange multiplier b, detail consistency constraint variable z, Lagrange multiplier c, deblurring error constraint variable t, and Lagrange multiplier a. The restored image u is solved and updated using Fast Fourier Transform, while the wavelet domain sparsity constraint variable d, detail consistency constraint variable z, and deblurring error constraint variable t are solved using the soft threshold operator.

8. An image denoising system integrating wavelet framework and sharpening operator, characterized in that, include: The sharpening module is configured to sharpen the acquired image to be processed based on sharpening operators to obtain a sharpened image. The model building module is configured to perform wavelet domain transformation on the sharpened image, with the goal of minimizing the weighted sum of the deblurring error term, the image sharpening consistency term, and the wavelet domain sparse regularization term, and to construct a denoising preprocessing mathematical model. The solution module is configured to decompose the solution of the denoising preprocessing mathematical model into several sub-problems, use the alternating direction multiplier method to solve the denoising preprocessing mathematical model, decompose the image to be processed into detail components of different scales, perform sparse constraint solution, and use soft thresholding to remove noise and redundant information to obtain the denoised restored image.

9. The image denoising system integrating wavelet framework and sharpening operator as described in claim 8, characterized in that, The mathematical model for denoising preprocessing includes: Deblurring error term: The L1 norm of the difference between the blurred image and the noisy image is calculated to ensure the consistency between the restored image and the noisy image after the deblurring operation. Image sharpening consistency term: An additional constraint is introduced by multiplying the L1 norm of the difference between the restored and sharpened images by a regularization coefficient μ. Wavelet domain sparse regularization: Wavelet transform is used to restore the sparse representation of the image, which is used to suppress noise and highlight details.

10. An image denoising system integrating wavelet framework and sharpening operator, characterized in that: Includes image acquisition device and processor; Image acquisition device, used to acquire images to be processed; The processor is configured to perform the steps of an image denoising processing method that integrates wavelet framework and sharpening operator as described in any one of claims 1-7.

Images