A Wavelet Total Variation Denoising Method Based on High Resolution

Abstract

The present invention discloses a wavelet total variation denoising method based on high resolution, which includes performing wavelet transform on a high-resolution image to obtain wavelet coefficients at different scales and ranges; based on the obtained wavelet coefficients, using the local features and gradient information of the image to adjust the regularization parameters for different noise regions in the image, and returning the adjustment results for different noise regions; performing inverse wavelet transform according to the adjustment results of different noise regions to remove noise and reconstruct the denoised high-resolution image; decomposing the image into frequency components at different scales through wavelet decomposition, effectively extracting image information in different scales and frequency ranges; by dynamically adjusting the regularization parameters, on the one hand, noise is suppressed, and on the other hand, the situation of over-smoothing is avoided, retaining the original texture and edge information of the image; the established double-loop mechanism increases the accuracy of the inverse transform and ensures the image detail effect of features in different noise regions.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular, to a wavelet total variation denoising method based on high resolution. Background Art

[0002] In recent years, with the rapid development of image processing technology, the acquisition and processing of high-resolution images have been widely applied in various fields, such as medical imaging, remote sensing monitoring, intelligent transportation, etc. High-resolution images contain richer detailed information and can provide more accurate analysis and recognition capabilities. However, with the popularization of high-resolution images, the problem of image noise has become increasingly prominent. Image noise not only affects the visual effect of images but also interferes with subsequent image analysis and processing. Therefore, how to effectively remove image noise has become an important research direction in the field of image processing.

[0003] Currently, the methods for denoising high-resolution images mainly include methods based on the spatial domain and methods based on the frequency domain. Methods based on the spatial domain, such as mean filtering, median filtering, etc., mainly rely on local information of images for processing, but are prone to losing image details and blurring during the denoising process. Methods based on the frequency domain, such as Fourier transform, wavelet transform, etc., can better retain the detailed information of images. Among them, wavelet transform has become an important tool in image denoising research due to its good time-frequency localization characteristics. However, traditional wavelet denoising methods do not perform well in dealing with complex noise and often have difficulty in simultaneously achieving noise suppression and detail retention. Summary of the Invention

[0004] The purpose of this part is to outline some aspects of the embodiments of the present invention and briefly introduce some preferred embodiments. Simplifications or omissions may be made in this part, as well as in the abstract and title of the present application, to avoid obscuring the purpose of this part, the abstract, and the title, and such simplifications or omissions shall not be used to limit the scope of the present invention.

[0005] In view of the above existing problems, the present invention is proposed. Therefore, the present invention provides a wavelet total variation denoising method based on high resolution to solve the problems raised in the background art.

[0006] To solve the above technical problems, the present invention provides the following technical solution: A wavelet total variation denoising method based on high resolution, comprising:

[0007] Obtain a high-resolution image, perform wavelet transform on the high-resolution image to obtain wavelet coefficients of different scales and ranges;

[0008] Based on the obtained wavelet coefficients, using the local features and gradient information of the image, adjust the regularization parameters for different noise regions in the image, and return the adjustment results for different noise regions;

[0009] Perform inverse wavelet transform to remove noise according to the adjustment results of the different noise regions, and reconstruct the high-resolution image after denoising.

[0010] As a preferred scheme of the wavelet total variation denoising method based on high resolution according to the present invention, wherein: obtaining a high-resolution image includes:

[0011] The high resolution is 4K, 5K, 6K, 7K, 8K, 10K, 12K, and 16K.

[0012] As a preferred scheme of the wavelet total variation denoising method based on high resolution according to the present invention, wherein: performing wavelet transform on the high-resolution image to obtain wavelet coefficients of different scales and ranges, including:

[0013] Obtain the wavelet type, and at the same time set the wavelet decomposition layer number and the surrounding frequency band range, and obtain the scale frequency of wavelet decomposition in each layer under different wavelet types;

[0014] According to the scale frequency, calculate the wavelet coefficients inside and outside the corresponding scale frequency range, and construct a wavelet coefficient packet, and judge whether the corresponding wavelet coefficient is included in the surrounding frequency band range;

[0015] If it is included in the surrounding frequency band range, further judge whether it is a wavelet coefficient within the frequency range. If so, update the scale frequency with the wavelet coefficient to obtain wavelet coefficients of different scales and ranges; if not, reset the wavelet decomposition layer number; if it is not included in the surrounding frequency band range, use the value of the frequency band range as the assignment parameter for constructing the wavelet coefficient packet, obtain the maximum and minimum values of the wavelet coefficient packet, and reset the surrounding frequency band range with the maximum and minimum values of the wavelet coefficient packet.

[0016] As a preferred scheme of the wavelet total variation denoising method based on high resolution according to the present invention, wherein: based on the obtained wavelet coefficients, using the local features and gradient information of the image, adjusting the regularization parameters for different noise regions in the image includes:

[0017] Calculate the weights of the local features of the image and perform feature summation;

[0018] Calculate the derivatives of the gradient information, and the derivatives of the gradient information include the first derivative, the second Laplacian operator, and the third derivative;

[0019] Adjust the regularization parameters for different noise regions by combining wavelet coefficients, summed image features, and the derivatives of gradient information at different scales and ranges to obtain the minimum noise in different regions.

[0020] As a preferred embodiment of the wavelet total variation denoising method based on high resolution according to the present invention, further comprising:

[0021] The different region noises are divided into low-noise regions, medium-noise regions, high-noise regions, and noise edge regions;

[0022] The ranges of the low-noise region, medium-noise region, and high-noise region respectively correspond to the representation ranges of the first-order derivative of gradient information, the second-order Laplacian operator, and the third-order derivative.

[0023] As a preferred embodiment of the wavelet total variation denoising method based on high resolution according to the present invention, wherein: the noise edge region includes:

[0024] It is composed of the first-order derivative of gradient information and wavelet basis functions.

[0025] As a preferred embodiment of the wavelet total variation denoising method based on high resolution according to the present invention, wherein: perform inverse wavelet transform to remove noise according to the adjustment results of the different noise regions, and reconstruct the high-resolution image after denoising, including:

[0026] Set four double loops corresponding to the low-noise region, medium-noise region, high-noise region, and noise edge region. Each double loop includes an outer loop and an inner loop. The outer loop controls the number of layers of the inverse wavelet transform, and the inner loop performs the inverse wavelet transform.

[0027] As a preferred embodiment of the wavelet total variation denoising method based on high resolution according to the present invention, further comprising:

[0028] Obtain the height and width of the high-resolution image;

[0029] Initialize matrices A and B with the same size as the image height and width to store the transformation process;

[0030] The outer loop includes the original transform in the vertical direction and the original transform in the horizontal direction;

[0031] The inner loop includes the inverse transform in the vertical direction and the inverse transform in the horizontal direction.

[0032] As a preferred embodiment of the wavelet total variation denoising method based on high resolution according to the present invention, wherein: the outer loop includes the original transform in the vertical direction and the original transform in the horizontal direction, including:

[0033] For the original transformation in the horizontal direction, for each row, calculate the mean of adjacent pixels and store it in A;

[0034] For the original transformation in the vertical direction, for each column, calculate the difference of adjacent pixels and store it in B. Update the inverse wavelet transform result of the current outer loop layer through (A + B) / 2, and enter the inner loop.

[0035] As a preferred solution of the wavelet total variation denoising method based on high resolution according to the present invention, wherein: the inner loop includes an inverse transform in the vertical direction and an inverse transform in the horizontal direction, including:

[0036] Zero the stored values of A and B;

[0037] For the inverse transform in the vertical direction of the inner loop, for each column, calculate the reconstructed value of the original pixels and store it in A;

[0038] For the inverse transform in the horizontal direction of the inner loop, for each row, calculate the reconstructed value of the original pixels and store it in B;

[0039] Update the inverse wavelet transform result of the current inner loop through |A - B|, and return to the outer loop.

[0040] Compared with the prior art, the beneficial effects of the invention are as follows: By obtaining a high-resolution image, performing wavelet transform on the high-resolution image to obtain wavelet coefficients of different scales and ranges; based on the obtained wavelet coefficients, using the local features and gradient information of the image to adjust the regularization parameters for different noise regions in the image, and returning the adjustment results of different noise regions; performing inverse wavelet transform according to the adjustment results of different noise regions to remove noise and reconstruct the denoised high-resolution image; decomposing the image into frequency components of different scales through wavelet decomposition, effectively extracting image information in different scales and frequency ranges; by dynamically adjusting the regularization parameters, on the one hand, noise is suppressed, and on the other hand, the situation of over-smoothing is avoided, and the original texture and edge information of the image are retained; the established double-loop mechanism increases the accuracy of the inverse transform and ensures the image detail effect of the characteristics of different noise regions. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings. Among them:

[0042] Figure 1 is the overall flowchart of the wavelet total variation denoising method based on high resolution according to an embodiment of the present invention;

[0043] Figure 2 Comparison graph of SNR, PSNR, SSIM noise standard deviation for the wavelet total variation denoising method based on high resolution according to an embodiment of the present invention;

[0044] Figure 3 Comparison graph of SNR, PSNR, SSIM noise metrics for the wavelet total variation denoising method based on high resolution according to an embodiment of the present invention. Detailed implementation manners

[0045] To make the above objects, features and advantages of the present invention more apparent and understandable, the following provides a detailed description of the specific implementation manners of the present invention with reference to the accompanying drawings of the specification. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0046] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0047] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure or characteristic that may be included in at least one implementation manner of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that excludes other embodiments.

[0048] The present invention is described in detail with reference to schematic diagrams. When detailing the embodiments of the present invention, for ease of explanation, the cross-sectional views showing the device structure will be enlarged locally in a non-general proportion, and the schematic diagrams are only examples and should not limit the scope of protection of the present invention herein. In addition, in actual production, three-dimensional spatial dimensions including length, width and depth should be included.

[0049] At the same time, in the description of the present invention, it should be noted that the orientation or positional relationship indicated by terms such as "upper, lower, inner and outer" is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance.

[0050] Unless otherwise clearly specified and defined in the present invention, the terms "installation, connection, and coupling" shall be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may also be a mechanical connection, an electrical connection, or a direct connection, or it may be indirectly connected through an intermediate medium, or it may be the communication inside two components. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0051] Embodiment 1

[0052] Referring to Figure 1 , which is the first embodiment of the present invention. This embodiment provides a wavelet total variation denoising method based on high resolution, including:

[0053] S1. Obtain a high-resolution image, perform wavelet transform on the high-resolution image to obtain wavelet coefficients of different scales and ranges;

[0054] Furthermore, the obtained high resolutions are 4K (3840×2160 pixels, total pixel number is 8,294,400), 5K (5120×2880 pixels, total pixel number is 14,745,600), 6K (6144×3160 pixels, total pixel number is 19,401,600), 7K (7168×3584 pixels, total pixel number is 25,694,464), 8K (7680×4320 pixels, total pixel number is 33,177,600), 10K (10240×4320 pixels, total pixel number is 44,236,800), 12K (12288×6480 pixels, total pixel number is 79,626,240), and 16K (15360×8640 pixels, total pixel number is 133,225,600);

[0055] Furthermore, obtain the wavelet type in the high-resolution image, and at the same time set the wavelet decomposition layer number L and the surrounding frequency band range B to obtain the scale frequencies of wavelet decomposition in each layer under different wavelet types;

[0056] Furthermore, according to the scale frequencies, calculate the wavelet coefficients inside and outside the corresponding scale frequency range, construct a wavelet coefficient packet, and determine whether the corresponding wavelet coefficients are included in the surrounding frequency band range;

[0057] Furthermore, if it is included in the surrounding frequency band range, then determine whether it is a wavelet coefficient within the frequency range. If so, update the scale frequencies with the wavelet coefficients to obtain wavelet coefficients of different scales and ranges; if not, reset the wavelet decomposition layer number; if it is not included in the surrounding frequency band range, then use the value of the surrounding frequency band range as the assignment parameter for constructing the wavelet coefficient packet, obtain the maximum and minimum values of the wavelet coefficient packet, and reset the surrounding frequency band range with the maximum and minimum values of the wavelet coefficient packet;

[0058] It should be noted that by judging the conditions to update the surrounding frequency band range or scale frequency, the detailed information of the image can be analyzed at different scale frequencies, avoiding the waiting requests for sequential processing or synchronous processing, and improving the flexibility of image processing;

[0059] Specifically, the wavelet transform formula is:

[0060]

[0061] where W j is the j-th wavelet coefficient, and W k (F) is the wavelet transform at the k-th layer;

[0062] Specifically, the wavelet packet coefficient packet is:

[0063]

[0064] where W p is the wavelet coefficient packet, f i is the i-th scale frequency, and F is the image;

[0065] S2. Based on the obtained wavelet coefficients, use the local features and gradient information of the image to adjust the regularization parameters for different noise regions in the image, and return the adjustment results for different noise regions;

[0066] Furthermore, calculate the weights of the local features of the image and perform feature summation;

[0067] Specifically, let the local feature of the image be F(x, y) and the weight be W(x, y), then we get:

[0068]

[0069] where S is the result of feature summation, M is the width of the image, N is the height of the image, and W(x, y) represents the weight of the image at coordinates x, y;

[0070] Even further, calculate the derivatives of the gradient information, and the derivatives of the gradient information include the first derivative, the second Laplacian operator, and the third derivative;

[0071] Specifically, let the gradient information of the image be Then the first derivative is expressed as:

[0072]

[0073] where represents the Gaussian gradient;

[0074] Specifically, the second Laplacian operator is expressed as:

[0075]

[0076] Among them, Δ represents the Laplace operator;

[0077] Specifically, the third derivative is expressed as:

[0078]

[0079] Furthermore, by combining wavelet coefficients of different scales and ranges, the sum of image features and the derivatives of gradient information, the regularization parameters for different noise regions are adjusted to obtain the minimum noise in different regions;

[0080] It should be noted that although adjusting the regularization parameters for different noise regions can suppress noise, it will affect the details of the original high-resolution image, so it is necessary to eliminate the influence through inverse wavelet transform;

[0081] Further, different regions of noise are divided into low-noise regions, medium-noise regions, high-noise regions, and noise edge regions;

[0082] Specifically, the ranges of the low-noise region, medium-noise region, and high-noise region respectively correspond to the ranges of the first derivative of gradient information, the second Laplace operator, and the third derivative;

[0083] Furthermore, the noise edge region is composed of the first derivative of gradient information and wavelet basis functions;

[0084] Specifically, the adjustment of the regularization parameter is expressed as:

[0085]

[0086] Among them, α represents the weight coefficient of the low-noise region, g z (x) represents the wavelet basis function of the z-th wavelet type; β represents the weight coefficient of the medium-noise region; γ represents the weight coefficient of the high-noise region; λ i is the result of the adjustment of the regularization parameter;

[0087] Specifically, the minimum noise in different regions is expressed as:

[0088]

[0089] Among them, n represents the total number of regions;

[0090] S3. Perform inverse wavelet transform according to the adjustment results of different noise regions to remove noise and reconstruct the denoised high-resolution image;

[0091] Further, four double loops are set, corresponding to the low-noise region, medium-noise region, high-noise region, and noise edge region. Each double loop includes an outer loop and an inner loop. The outer loop controls the number of layers of the inverse wavelet transform, and the inner loop performs the inverse wavelet transform.

[0092] Specifically, both double loops adopt the form of For loop nesting, with the step size being the minimum noise of different regions. Each time the outer loop is looped once, the region is switched. For example, when the current inner loop finishes execution and transfers to the outer loop, the step size corresponding to the minimum noise of the low-noise region becomes the step size corresponding to the minimum noise of the medium-noise region, and so on.

[0093] Further, the steps of the inverse wavelet transform are as follows:

[0094] Obtain the height and width of the high-resolution image.

[0095] Initialize matrices A and B with the same size as the image height and width to store the transformation process.

[0096] Specifically, the outer loop includes the original transformation in the vertical direction and the original transformation in the horizontal direction. Among them, the original transformation refers to the process of wavelet transform in the above formula. The inner loop includes the inverse transformation in the vertical direction and the inverse transformation in the horizontal direction. Considering the vertical and horizontal directions is because of the representation form of the image in the original coordinates (x, y), and the inverse transformation is the way to reverse-solve the regularization adjustment result for the original coordinates.

[0097] Furthermore, for the original transformation in the horizontal direction, for each row, calculate the mean value of adjacent pixels and store it in A.

[0098] Furthermore, for the original transformation in the vertical direction, for each column, calculate the difference between adjacent pixels and store it in B. Then, update the result of the inverse wavelet transform of the current outer loop layer through (A + B) / 2 and enter the inner loop.

[0099] Further, set the values stored in A and B to zero.

[0100] Specifically, the reason for setting to zero is that after updating the result of the inverse wavelet transform of the current outer loop layer, there are still residual data in A and B at this time. In order not to affect the accuracy of the final inverse wavelet transform result, it is necessary to clear them.

[0101] Furthermore, for the inverse transformation in the vertical direction of the inner loop, for each column, calculate the reconstructed value of the original pixel and store it in A.

[0102] Furthermore, for the inverse transformation in the horizontal direction of the inner loop, for each row, calculate the reconstructed value of the original pixel and store it in B.

[0103] Further, update the inverse wavelet transform result of the current inner loop through |A - B|, and return to the outer loop;

[0104] Specifically, the reconstructed value of the original pixel refers to the value obtained by performing wavelet transform on the pixel value of the original high-resolution image.

[0105] Embodiment 2

[0106] Refer to Figure 2 and Figure 3 , which is the second embodiment of the present invention. This embodiment provides a wavelet total variation denoising method based on high resolution, including: To verify the effectiveness of the wavelet total variation denoising method based on high resolution, this embodiment selects four groups of high-resolution images with different noise levels for experiments; The test images are complex medical images with resolutions of 4K, 6K, 8K, and 12K respectively, and different degrees of Gaussian noise and salt-and-pepper noise are added to each group of images;

[0107] Obtain high-resolution medical images with resolutions of 4K (3840×2160 pixels), 6K (6144×3160 pixels), 8K (7680×4320 pixels), and 12K (12288×6480 pixels) respectively;

[0108] Add Gaussian noise and salt-and-pepper noise to each group of high-resolution images respectively, with noise standard deviations of 0.01, 0.03, 0.05, and 0.07 respectively. The Gaussian noise is normally distributed, and the densities of the salt-and-pepper noise are 0.01, 0.03, 0.05, and 0.07 respectively;

[0109] Perform wavelet transform on the high-resolution images after adding noise, select Daubechies wavelet as the wavelet basis function, and decompose the images into 4 layers to obtain wavelet coefficients with different scales and ranges;

[0110] Utilize the local features and gradient information of the images to adjust the regularization parameters for different noise regions; Perform inverse wavelet transform according to the adjustment results of different noise regions to remove noise and reconstruct the denoised high-resolution images; During the inverse wavelet transform process, adopt a double-loop mechanism to process the low-noise region, medium-noise region, high-noise region, and noise edge region respectively to ensure the retention of image details and edge information; Evaluate the quality of the images before and after denoising, use signal-to-noise ratio (SNR), peak signal-to-noise ratio (PSNR), and structural similarity (SSIM) indicators for evaluation, and record them in the experimental data table, referring to Table 1;

[0111] Table 1

[0112]

[0113]

[0114] As can be seen from Table 1, the SNR before denoising is relatively low at high noise levels, and the SNR after denoising is significantly improved, with an average increase of more than 5.0 dB, indicating that the method of the present invention effectively suppresses noise through regularization adjustment; the PSNR before denoising decreases as the noise standard deviation increases, and the PSNR after denoising is significantly improved, with an average increase of more than 5.3 dB, indicating that the double-loop mechanism of the method of the present invention can eliminate the influence of image details brought by regularization adjustment; the SSIM before denoising significantly decreases at high noise levels, and the SSIM after denoising is greatly improved, with an average increase of more than 0.12, fully proving the effectiveness of the method of the present invention.

[0115] And in combination with Figure 2 and Figure 3 it can be seen that by dynamically adjusting the regularization parameter, the present invention not only suppresses noise, but also avoids over-smoothing of the image, and ensures the accuracy of the inverse wavelet transform during the double-loop process.

[0116] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) containing computer-usable program code. The solutions in the embodiments of the present application can be implemented in various computer languages. For example, object-oriented programming languages such as Java and interpreted scripting languages such as JavaScript.

[0117] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of the flows and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the specified functions in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0118] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means that implement the function specified in the flowchart(s) Figure 1 a flowchart or flowcharts and / or block(s) Figure 1 a block or blocks.

[0119] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable apparatus provide steps for implementing the function specified in the flowchart(s) Figure 1 a flowchart or flowcharts and / or block(s) Figure 1 a block or blocks.

[0120] Although the preferred embodiments of the present application have been described, additional changes and modifications can be made by those skilled in the art once they learn of the basic inventive concept. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications that fall within the scope of the present application.

[0121] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application is also intended to include these modifications and variations.

Claims

1. A wavelet total variation denoising method based on high resolution, characterized in that: include: Acquire a high-resolution image, and perform wavelet transformation on the high-resolution image to obtain wavelet coefficients of different scales and ranges; The high-resolution image is subjected to wavelet transformation to obtain wavelet coefficients of different scales and ranges, including: Get the wavelet type, set the wavelet decomposition layer number and the surrounding frequency band range, and get the scale frequency of the wavelet decomposition in each layer number under different wavelet types; According to the scale frequency, the wavelet coefficients within and outside the corresponding scale frequency range are calculated, and a wavelet coefficient package is constructed to determine whether the corresponding wavelet coefficients are included in the surrounding frequency band range; If it is included in the surrounding frequency band, then determine whether it is a wavelet coefficient within the frequency range. If so, re-update the scale frequency through the wavelet coefficient to obtain wavelet coefficients of different scales and ranges; if not, reset the number of wavelet decomposition layers; if it is not included in the surrounding frequency band, use the value of the frequency band as the assignment parameter for constructing a wavelet coefficient package, obtain the maximum and minimum values ​​of the wavelet coefficient package, and reset the surrounding frequency band with the maximum and minimum values ​​of the wavelet coefficient package; Based on the obtained wavelet coefficients, the local features and gradient information of the image are used to adjust the regularization parameters of different noise areas in the image, and the adjustment results of different noise areas are returned; Based on the obtained wavelet coefficients, the local features and gradient information of the image are used to adjust the regularization parameters of different noise areas in the image, including: Calculate the weights of local features of the image and sum the features; Calculating the derivative of the gradient information, wherein the derivative of the gradient information includes a first-order derivative, a second-order Laplace operator, and a third-order derivative; By combining the wavelet coefficients of different scales and ranges, the summed image features and the derivative of the gradient information, the regularization parameters of different noise regions are adjusted to obtain the minimum noise in different regions; Performing inverse wavelet transform to remove noise according to the adjustment results of the different noise regions, and reconstructing a high-resolution image after denoising; Performing inverse wavelet transform to remove noise according to the adjustment results of the different noise regions, and reconstructing a high-resolution image after noise removal, including: Four double loops are set, corresponding to the low noise area, the medium noise area, the high noise area and the noise edge area. Each double loop includes an outer loop and an inner loop. The outer loop controls the number of layers of the inverse wavelet transform, and the inner loop performs the inverse wavelet transform.

2. The wavelet total variation denoising method based on high resolution as claimed in claim 1, characterized in that: Get high-resolution images, including: The high resolutions are 4K, 5K, 6K, 7K, 8K, 10K, 12K and 16K.

3. The wavelet total variation denoising method based on high resolution as claimed in claim 1, characterized in that: Based on the obtained wavelet coefficients, the local features and gradient information of the image are used to adjust the regularization parameters of different noise areas in the image, including: The different noise areas are divided into a low noise area, a medium noise area, a high noise area and a noise edge area; The low noise region, the medium noise region, and the high noise region ranges correspond to the representation ranges of the first-order derivative, the second-order Laplace operator, and the third-order derivative of the gradient information, respectively.

4. The wavelet total variation denoising method based on high resolution as claimed in claim 3, characterized in that: The noise edge area includes: It consists of the first-order derivative of gradient information and wavelet basis function.

5. The wavelet total variation denoising method based on high resolution as claimed in claim 1, characterized in that: The method further comprises: performing inverse wavelet transform to remove noise according to the adjustment results of the different noise regions, and reconstructing a high-resolution image after the noise is removed. Get the height and width of the high-resolution image; Initialize matrices A and B with the same height and width as the image to store the transformation process; The outer loop contains the original transformation in the vertical direction and the original transformation in the horizontal direction; The inner loop includes inverse vertical transform and inverse horizontal transform.

6. The wavelet total variation denoising method based on high resolution as claimed in claim 5, characterized in that: The outer loop includes the original transformation in the vertical direction and the original transformation in the horizontal direction, including: The original transformation in the horizontal direction, for each row, calculates the mean of adjacent pixels and stores it in A; The original transformation in the vertical direction calculates the difference between adjacent pixels for each column, stores it in B, updates the inverse wavelet transform result of the current outer loop layer by (A+B) / 2, and enters the inner loop.

7. The wavelet total variation denoising method based on high resolution as claimed in claim 5, characterized in that: The inner loop includes vertical inverse transform and horizontal inverse transform, including: Reset the values ​​stored in A and B to zero; The vertical inverse transform of the inner loop calculates the reconstructed value of the original pixel for each column and stores it in A; The horizontal inverse transformation of the inner loop calculates the reconstructed value of the original pixel for each row and stores it in B; Through |AB|, the inverse wavelet transform result of the current inner loop is updated and returned to the outer loop.

Images