An image denoising method based on an adaptive total variation image denoising model

Abstract

The application discloses an image denoising method based on an adaptive total variation image denoising model, which comprises the following steps: obtaining an image to be denoised; preprocessing the obtained image to be denoised to obtain an original image matrix; introducing the original image matrix into the adaptive total variation image denoising model based on preset model parameters, and obtaining multiple denoised images through multiple iterations of denoising; and selecting a denoised image with the best denoising effect from the multiple denoised images as the denoised image of the image to be denoised, wherein the adaptive function of the adaptive total variation image denoising model is used to control the denoising intensity in the denoising process. The application can denoise an image and obtain a denoised image.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image inpainting, and particularly relates to an image denoising method based on an adaptive total variation image denoising model. BACKGROUND

[0002] Since images are polluted by different degrees of noise in the process of acquisition, transmission and display, image inpainting is an important processing link.

[0003] In the field of image inpainting, the total variation image denoising method is an important branch. In the total variation image denoising method, there are several important total variation denoising models. Among them, the harmonic model is isotropic diffusion, so the harmonic model is not conducive to edge preservation, causing edge blur; the TV model is anisotropic diffusion, and has the effects of preserving edges and smoothing noise, but when the noise is too large, the staircase effect occurs; the Zhang Hongying model is an adaptive total variation model, which performs anisotropic diffusion at the edge and isotropic diffusion at the flat area, and can suppress the staircase effect to a certain extent. However, since the gradient module is used as an edge detection operator, some detailed information such as edges and textures cannot be fully represented, and the denoising effect of the model is not ideal.

[0004] Therefore, the present application provides an image denoising method based on an adaptive total variation image denoising model. SUMMARY

[0005] The present application aims to overcome the deficiencies in the prior art, and provides an image denoising method based on an adaptive total variation image denoising model, which can denoise images and obtain denoised images.

[0006] To achieve the above-mentioned purpose, the present application adopts the following technical scheme:

[0007] The present application provides an image denoising method based on an adaptive total variation image denoising model, comprising the following steps:

[0008] Obtaining a to-be-denoised image;

[0009] Preprocessing the obtained to-be-denoised image to obtain an original image matrix;

[0010] Based on the preset model parameters, the original image matrix is imported into the adaptive total variation image denoising model, and after multiple iterations of denoising, multiple denoised images are obtained, and the denoised image with the best denoising effect is selected from the multiple denoised images as the denoised image of the to-be-denoised image;

[0011] The adaptive function of the adaptive total variation image denoising model is used to control the denoising intensity in the denoising process.

[0012] Further, the original image matrix is a two-dimensional matrix of m1*m2, and each element in the two-dimensional matrix corresponds to a pixel point of the image to be denoised.

[0013] wherein m1 is the number of rows of the two-dimensional matrix, and m2 is the number of columns of the two-dimensional matrix.

[0014] Further, the original image matrix is introduced into the adaptive total variation image denoising model based on the preset model parameters, and a plurality of denoised images are obtained through multiple iterations of denoising, including:

[0015] The following iteration steps are performed:

[0016] n, u, u ηη and are obtained, and the initial values or the updated values corresponding to the last iteration step and the model parameters are obtained;

[0017] The denoised image matrix of the current iteration number is updated by the following formula:

[0018]

[0019] In the formula, the superscript n represents the initial state or the updated state of the last iteration step, the superscript n+1 represents the current updated state, n is the iteration number, and the initial value of n is 0, u is the denoised image matrix, and the initial value of u is u0, u0 is the original image matrix, is the second derivative of the denoised image matrix in the edge direction, and The initial value of u is the second derivative of the original image matrix in the edge direction, u ηη is the second derivative of the denoised image matrix in the gradient direction, and u ηη The initial value of u is the second derivative of the original image matrix in the gradient direction, is the gradient modulus of the denoised image matrix, and The initial value of u is the gradient modulus of the original image matrix, P is an adaptive function, and λ and dt are model parameters.

[0020] It is determined whether the iteration number meets the standard: if it meets the standard, the iteration step is stopped and the denoised image matrix of the current updated state is output to obtain the corresponding denoised image, otherwise, the denoised image matrix of the current updated state is output to obtain the corresponding denoised image, and n, u, u ηη and

[0021] u n =u n+1

[0022]

[0023]

[0024]

[0025] n = n + 1

[0026] wherein u x is the partial derivative of the denoised image matrix with respect to x, u y is the partial derivative of the denoised image matrix with respect to y, u xy is the second order partial derivative of the denoised image matrix with respect to xy, u yy is the second order partial derivative of the denoised image matrix with respect to y, u xx is the second order partial derivative of the denoised image matrix with respect to x.

[0027] Further, u x , u y u y , u xy , u xx and u yy are updated by the following equations:

[0028]

[0029]

[0030]

[0031]

[0032]

[0033] wherein (, ) is an element, and x is an integer belonging to [0, m1], y is an integer belonging to [0, m2], wherein if x = 0 and / or y = 0, the element (x-1, y-1) is set to 0; if y = 0 and / or x = m1, the element (x+1, y-1) is set to 0; if y = m2 and / or x = m1, the element (x+1, y+1) is set to 0; if x = 0 and / or y = m2, the element (x-1, y+1) is set to 0; if x = m1, the element (x+1, y) is set to 0; if x = 0, the element (x-1, y) is set to 0; if y = m2, the element (x, y+1) is set to 0; if y = 0, the element (x, y-1) is set to 0.

[0034] Further, the judging whether the iteration number is up to standard comprises:

[0035] when n is greater than or equal to 300, the iteration number is up to standard;

[0036] when n is less than 300, the iteration number is not up to standard.

[0037] Further, the self-adaptive function comprises the following formula:

[0038]

[0039]

[0040] Wherein, qulvtidu(·) is the texture feature of the denoising image, K(·) is the second derivative level set curvature, d1(·) is the normalized difference curvature, and V1(·) is the normalized local variance.

[0041] Further, the normalized difference curvature comprises the following formula:

[0042]

[0043] In the formula, D1(·) is the difference curvature, min(·) is the minimum value of the function, and max(·) is the maximum value of the function.

[0044] Further, the second derivative level set curvature comprises the following formula:

[0045]

[0046] Further, the normalized local variance comprises the following formula:

[0047]

[0048] In the formula, sigma(·) is the variance.

[0049] Further, the variance comprises the following formula:

[0050]

[0051] In the formula, is the average value of the texture feature of the local field of the denoising image, N is the pixel range parameter of the local field, and N is an odd positive integer smaller than m1 and m2, and the element (x+i, y+j) is 0 if x+i<0 and / or x+i>m1 and / or y+j<0 and / or y+j>m2.

[0052] Compared with the prior art, the present application has the following beneficial effects:

[0053] The method of the present application is simple, and the original image matrix is introduced into the self-adaptive total variation image denoising model to obtain the denoising image of the image to be denoised, and the self-adaptive total variation image denoising model can not only realize image denoising, but also make the denoising image retain more detailed features. BRIEF DESCRIPTION OF DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiment description will be briefly introduced.

[0055] Figure 1 An embodiment flow chart of the image denoising method based on the adaptive total variation image denoising model of the present application;

[0056] Figure 2 A denoised image obtained after the method of the present application is used to denoise a test image with a noise standard deviation of 15;

[0057] Figure 3 A denoised image obtained after the method of the present application is used to denoise a test image with a noise standard deviation of 20;

[0058] Figure 4 A denoised image obtained after the method of the present application is used to denoise a test image with a noise standard deviation of 25;

[0059] Figure 5 The PSNR values of the denoised images obtained after the method of the present application is used to denoise each test image. DETAILED DESCRIPTION

[0060] In the description of the present application, it should be understood that the terms "center", "longitudinal", "transverse", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. In addition, the terms "first", "second" and the like are only for the purpose of description and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, the features defined with "first", "second" and the like can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise specified.

[0061] In the description of the present application, it should be noted that unless otherwise specified and limited, the terms "mounting", "connecting", "connection" should be understood broadly, for example, it can be fixed connection, or detachable connection, or integrally connected; it can be mechanical connection, or electrical connection; it can be directly connected, or indirectly connected through an intermediate medium, or the communication between the two elements inside. For those skilled in the art, the specific meaning of the above terms in the present application can be understood through specific circumstances.

[0062] Example 1:

[0063] The embodiment provides an image denoising method based on an adaptive total variation image denoising model.

[0064] Reference Figure 1 The image denoising method based on the adaptive total variation image denoising model comprises the following steps:

[0065] S1 obtains an image to be denoised.

[0066] S2 pre-processes the obtained image to be denoised to obtain an original image matrix.

[0067] In an application, the original image matrix is a two-dimensional matrix of m1*m2, and each element in the two-dimensional matrix corresponds to a pixel point of the image to be denoised.

[0068] Wherein, m1 is the number of rows of the two-dimensional matrix, and m2 is the number of columns of the two-dimensional matrix.

[0069] S3 imports the original image matrix into the adaptive total variation image denoising model based on preset model parameters, obtains multiple denoised images through multiple iterations of denoising, and selects a denoised image with the best denoising effect from the multiple denoised images as the denoised image of the image to be denoised.

[0070] Wherein, the adaptive function of the adaptive total variation image denoising model is used to control the denoising intensity in the denoising process.

[0071] In an application, a denoised image matrix is obtained each time the denoising is iterated once, and the corresponding denoised image and the iteration number are output. Then, the denoised image with the best denoising effect is selected from the multiple denoised images to determine the denoised image of the image to be denoised.

[0072] The method steps are simple, the original image matrix is imported into the adaptive total variation image denoising model, the denoised image of the image to be denoised can be obtained, and the adaptive total variation image denoising model can not only realize image denoising but also make the denoised image retain more detailed features.

[0073] Embodiment 2

[0074] Based on the embodiment 1, the embodiment details the steps of importing the original image matrix into the adaptive total variation image denoising model, and obtaining multiple denoised images through multiple iterations of denoising.

[0075] Importing the original image matrix into the adaptive total variation image denoising model based on preset model parameters, and obtaining multiple denoised images through multiple iterations of denoising comprises:

[0076] The following iteration steps are performed:

[0077] S31 obtains n, u, uηη and the initial value or the updated value corresponding to the last iteration step of and the model parameters;

[0078] S32 updates the denoising image matrix of the current iteration number by the following formula:

[0079]

[0080] In the formula, the superscript n represents the initial state or the updated state of the last iteration step, the superscript n+1 represents the current updated state, n is the iteration number, the initial value of n is 0, u is the denoising image matrix, and the initial value of u is u0, u0 is the original image matrix, is the second-order derivative of the denoising image matrix in the edge direction, and the initial value of is the second-order derivative of the original image matrix in the edge direction, u ηη is the second-order derivative of the denoising image matrix in the gradient direction, and ηη the initial value of is the second-order derivative of the original image matrix in the gradient direction, is the gradient modulus of the denoising image matrix, and the initial value of is the gradient modulus of the original image matrix, P is an adaptive function, and λ and dt are both model parameters;

[0081] S33 judges whether the iteration number meets the requirement:

[0082] When n is greater than or equal to 300, the iteration number meets the requirement, and then the iteration step is stopped and the denoising image matrix of the current updated state is output to obtain the corresponding denoising image. 300 is a preset total iteration number, which can be adjusted by those skilled in the art according to actual needs.

[0083] When n is less than 300, the iteration number does not meet the requirement, and then the denoising image matrix of the current updated state is output to obtain the corresponding denoising image, and n, u, u ηη and

[0084] u n = u n+1

[0085]

[0086]

[0087]

[0088] n = n + 1

[0089] In the formula, u x is the partial derivative of the denoising image matrix with respect to x, uy the first order partial derivative of the denoised image matrix with respect to y, u xy the second order partial derivative of the denoised image matrix with respect to xy, u yy the second order partial derivative of the denoised image matrix with respect to y, u xx the second order partial derivative of the denoised image matrix with respect to x.

[0090] In the application, when n is less than 300, the iteration number is not up to the standard, and u is updated by the following formula: x y y xy xx yy

[0091]

[0092]

[0093]

[0094]

[0095]

[0096] wherein (x, y) is an element, and x is an integer belonging to [0, m1], and y is an integer belonging to [0, m2], wherein if x = 0 and / or y = 0, the element (x-1, y-1) is 0; if y = 0 and / or x = m1, the element (x+1, y-1) is 0; if y = m2 and / or x = m1, the element (x+1, y+1) is 0; if x = 0 and / or y = m2, the element (x-1, y+1) is 0; if x = m1, the element (x+1, y) is 0; if x = 0, the element (x-1, y) is 0; if y = m2, the element (x, y+1) is 0; if y = 0, the element (x, y-1) is 0.

[0097] Embodiment 3:

[0098] Based on Embodiment 1 or 2, this embodiment details the method for obtaining the adaptive function.

[0099] S301 first, the second order derivative level set curvature is determined by the following formula:

[0100]

[0101] wherein K(·) is the second order derivative level set curvature.

[0102] S302 then, the texture feature of the denoised image is extracted by the following formula: ​​​​​​

[0103]

[0104] In the formula, qulvtidu(·) is the texture feature of the denoising image.

[0105] The traditional method represents the texture feature of the image by the gradient modulus of the image matrix, and the gradient modulus is the first derivative. Compared with the gradient modulus, the second derivative level set curvature can express more abundant texture features and edge details, so the embodiment comprehensively represents the texture feature of the denoising image by the second derivative level set curvature and the gradient modulus of the image matrix, so as to better consider the texture detail features in the image.

[0106] S303Then, the variance is determined by the following formula:

[0107]

[0108] In the formula, σ(·) is the variance, is the average value of the texture feature of the local region of the denoising image, N is the pixel range parameter of the local region, and N is an odd positive integer smaller than m1 and m2, and the element (x+i, y+j) is 0 if x+i<0 and / or x+i>m1 and / or y+j<0 and / or y+j>m2.

[0109] S304Next, the normalized local variance is determined by the following formula:

[0110]

[0111] In the formula, V1(·) is the normalized local variance, min(·) is the minimum value of the function, and max(·) is the maximum value of the function.

[0112] Based on the characteristics of the normalized local variance, in the edge region, the texture feature of the denoising image changes sharply, so the normalized local variance tends to 1; in the flat region, the texture feature of the denoising image changes slowly, so the normalized local variance tends to 0. Therefore, the normalized local variance can represent the change of the texture feature of the local denoising image. S305Then, the normalized differential curvature is determined by the following formula:

[0113]

[0114] In the formula, d1(·) is the normalized differential curvature, and D1(·) is the differential curvature.

[0115] In the application, the differential curvature is determined by the following formula:

[0116]

[0117] In practical application, based on the property of differential curvature, differential curvature can distinguish noise points, flat areas and edge areas of an image:

[0118] In noise points, and |u ηη are very large and almost equal, the value of D1(x, y) is almost 0, and the value of d1(x, y) is almost 0, therefore, when d1(x, y)≤0.001, it is represented as an isolated noise point;

[0119] In flat areas, and |u ηη are very large and almost equal, the value of D1(x, y) is very small, about 5.

[0120] In edge areas, |u ηη is very large, is very small, and therefore the value of D1(x, y) is greater than 10, even much greater than 0.

[0121] 306Finally, the adaptive function is determined by the following formula:

[0122]

[0123] In the formula, P is the adaptive function.

[0124] The piecewise adaptive function can produce different denoising effects in different areas of the denoised image:

[0125] In edge areas, the values of V1(·) and qulvtidu(·) are very large, so that the adaptive function approaches 1, the denoising strength is reduced, and anisotropic denoising is performed to preserve the detailed features of the edge. In application, the introduction of the exponential 1+d1(x, y) can make the adaptive function more sensitive to the edge area, and avoid the loss of detailed features of the edge.

[0126] In flat areas, the values of V1(·) and qulvtidu(·) are very small, so that the adaptive function is equal to 2, the denoising strength is increased, and noise is smoothed.

[0127] In transition areas, one of the values of V1(·) and qulvtidu(·) is very large, and the other is slightly small, so that the adaptive function is between 1-2, and moderate denoising is performed to protect the detailed texture and achieve the denoising effect.

[0128] Example 4

[0129] In this embodiment, the harmonic model, the TV model, the Zhang Hongying model and the adaptive total variation image denoising model are initialized, and the total iteration number of the four models is preset to be 300, the model parameter dt=0.1, λ=0.05, and N=7.

[0130] Based on Examples 1-3, this example uses image denoising methods based on harmonic models, TV models, Zhang Hongying models, and adaptive total variation image denoising models to denoise test images with added Gaussian white noise, and extracts the denoised image with the best denoising effect and the corresponding number of iterations for each method.

[0131] refer to Figures 2-4 It can be seen that when denoising test images with noise standard deviations of 15, 20 and 25 respectively, the denoised image obtained by the image denoising method based on the adaptive total variation image denoising model of this application has the best denoising effect compared with the other three methods: fewer noise points, clear edge contours and no staircase effect.

[0132] In addition, analyze one by one Figures 2-4 The PSNR, SSIM, and optimal number of iterations G for each denoising map are detailed in Table 1.

[0133] Table 1 shows the PSNR, SSIM, and optimal number of iterations G for the best denoised image obtained by four image denoising methods.

[0134]

[0135] As shown in Table 1, the PSNR and SSIM values ​​of the image denoising method based on the adaptive total variation image denoising model in this application are significantly higher than those of the other three methods. In terms of the number of iterations, the number of iterations required for the denoised image obtained by the image denoising method based on the adaptive total variation image denoising model in this application is significantly less than that of the image denoising methods based on the TV model and the Zhang Hongying model.

[0136] In practical applications, refer to Figure 5 ,Will Figures 2-4 The PSNR values ​​of each noise reduction map are plotted as a line graph. (This is achieved through...) Figure 5 It can be seen that the image denoising method based on the adaptive total variation image denoising model of this application produces denoised images with relatively large PSNR values ​​at each peak signal-to-noise ratio.

[0137] In summary, this application has good image noise reduction effect.

[0138] Those skilled in the art will appreciate that embodiments of the application can be readily used as software, hardware, or a combination of software and hardware. In one

[0139] The computer program instructions can 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 such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can 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 such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can 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 such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks.

[0140] The computer program instructions can 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 such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can 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 such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can 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 such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks.

[0141] The computer program instructions can 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 such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can 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 such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can 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 such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks.

[0142] The embodiments of the application described above are intended to be merely exemplary and those skilled in the art will recognize that many changes can be made to the specific embodiments described which will fall within the scope of the present application as defined by the claims below.

Claims

1. An image denoising method based on an adaptive total variation image denoising model, characterized in that, The method comprises the following steps: obtaining an image to be de-noised; preprocessing the obtained image to be de-noised to obtain an original image matrix; based on preset model parameters, importing the original image matrix into an adaptive total variation image de-noising model, performing multiple iterations of de-noising to obtain multiple de-noised images, and selecting a de-noised image with the best de-noising effect from the multiple de-noised images as the de-noised image of the image to be de-noised; wherein the adaptive function of the adaptive total variation image de-noising model is used to control the de-noising intensity in the de-noising process; the step of importing the original image matrix into the adaptive total variation image de-noising model based on the preset model parameters and performing multiple iterations of de-noising to obtain multiple de-noised images comprises: performing the following iteration steps: get n, , , and the initial values or the updated values corresponding to the last iteration step and the model parameters; updating the de-noised image matrix of the current iteration number by the following formula: ； wherein the superscript n represents an initial state or an updated state of a previous iteration step, the superscript n+1 represents a current updated state, n is an iteration number, and an initial value of n is 0, is a denoised image matrix, and an initial value of , is an original image matrix, is a second-order derivative of the denoised image matrix in an edge direction, and an initial value of is a second-order derivative of the original image matrix in the edge direction, is a second-order derivative of the denoised image matrix in a gradient direction, and an initial value of is a second-order derivative of the original image matrix in the gradient direction, is a gradient modulus of the denoised image matrix, and an initial value of is a gradient modulus of the original image matrix, and P is an adaptive function, and are model parameters. determining whether the iteration number meets the criterion: if yes, stopping the iteration step and outputting the current updated state of the denoising image matrix to obtain the corresponding denoising image, otherwise, outputting the current updated state of the denoising image matrix to obtain the corresponding denoising image, and updating n by the following formula: , , and : ； ； ； ； n = n + 1; In the formula, For denoising image matrix pairs The partial derivative, For denoising image matrix pairs The partial derivative, For denoising image matrix pairs The second-order partial derivative, For denoising image matrix pairs The second-order partial derivative, For denoising image matrix pairs The second-order partial derivative; the adaptive function comprises the following formula: ； ； wherein, is a texture feature of the denoised image, is a second derivative level set curvature, is a normalized difference curvature, is a normalized local variance; wherein the normalized differential curvature comprises the following formula: ； wherein is the differential curvature, is the minimum of the function, is the maximum of the function; the normalized local variance comprises the following formula: ； In the formula, is the variance; the variance comprises the following formula: ； wherein is the average value of the texture feature of the local region of the noise-reduced image, N is a pixel range parameter of the local region, and N is an odd positive integer smaller than m1 and m2, if <0 and / or m1 and / or <0 and / or m2, then the element ( , ) is 0.

2. The image denoising method based on the adaptive total variation image denoising model according to claim 1, characterized in that, the original image matrix is a two-dimensional matrix of m1*m2, and each element in the two-dimensional matrix corresponds to a pixel point of the image to be de-noised; wherein m1 is the number of rows of the two-dimensional matrix, and m2 is the number of columns of the two-dimensional matrix.

3. The image denoising method based on the adaptive total variation image denoising model according to claim 1, characterized in that, is updated by the following formula , , , and : ； ； ； ； ； In the formula, (,) are elements, where x is an integer belonging to [0, m1] and y is an integer belonging to [0, m2]. If x = 0 and / or y = 0, then let the element ( , If y=0 and / or x=m1, then let the element ( , If y = m2 and / or x = m1, then let the element ( , If x=0 and / or y=m2, then let the element ( , If x = m1, then let the element () be 0; , If x = 0, then set the element () to 0; if x = 0, then set the element () to 0. , If y = m2, then let the element () be 0; , If y=0, then set the element () to 0; if y=0, then set the element () to 0. , The value is 0.

4. The image denoising method based on the adaptive total variation image denoising model according to claim 1, characterized in that, the step of determining whether the iteration number meets the standard comprises: when n is greater than or equal to 300, the iteration number meets the standard; when n is less than 300, the iteration number does not meet the standard.

5. The image denoising method based on the adaptive total variation image denoising model according to claim 1, characterized in that, the second-order derivative level set curvature comprises the following formula: 。

Images