Texture perception smoothing method and device based on local extreme difference

Abstract

The present invention discloses a texture-aware smoothing method and device based on local extreme difference, belonging to the field of image processing technology. The method of the present invention first obtains an image to be smoothed; then the image to be smoothed is input into a texture-aware smoothing model based on local extreme difference, and the texture-aware smoothing model is solved to obtain a first smoothing result corresponding to the image to be smoothed. In the texture-aware smoothing model based on local extreme difference, since the texture regularization term is determined based on the L1 norm of the texture perception operator, it has a strong discriminant effect on the texture and structure of the image to be smoothed, so that the texture features in the image to be smoothed can be accurately perceived, the amount of information obtained can be increased, and the quality and accuracy of the first smoothing result can be improved, and the universality is strong, the processing efficiency is high, and the subsequent image accurate recognition is guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to a texture perception smoothing method and device based on local extreme difference. Background Art

[0002] With the widespread application of image processing technology in various industries, how to improve the amount of information obtained and processing efficiency has become an urgent problem to be solved. Image smoothing, as an indispensable preprocessing step, also needs to conform to current needs.

[0003] However, existing image smoothing methods are usually implemented using convex optimization methods, which cannot take into account the problem of information acquisition and processing efficiency, and are not suitable for application in actual complex scenarios. Especially in high-tech fields such as military, medicine, and remote sensing that require real-time response, the low universality of image smoothing methods is not only time-consuming, but also unable to meet the accuracy requirements, which is not conducive to the realization of downstream tasks such as image recognition. Summary of the invention

[0004] The present invention provides a texture perception smoothing method and device based on local extreme difference, which are used to solve the defects in the prior art.

[0005] The present invention provides a texture perception smoothing method based on local extreme difference, comprising:

[0006] Obtaining the image to be smoothed;

[0007] Inputting the image to be smoothed into a texture-aware smoothing model based on local range, and solving the texture-aware smoothing model to obtain a first smoothing result corresponding to the image to be smoothed;

[0008] The texture perception smoothing model includes a data fidelity term and a texture regularization term, wherein the data fidelity term is determined based on an input image and an output image, and the texture regularization term is determined based on an L1 norm of a texture perception operator;

[0009] The texture perception operator is determined based on the extreme difference of pixel values ​​within a local window of the output image and the gradient of the output image.

[0010] According to a texture-aware smoothing method based on local range provided by the present invention, solving the texture-aware smoothing model to obtain a first smoothing result corresponding to the image to be smoothed includes:

[0011] Using the first smoothing result obtained in the current iteration as the output image and the first smoothing result obtained in the previous iteration as the input image, updating the texture-aware smoothing model to obtain a first updated model;

[0012] The first update model is iteratively solved to obtain a second smoothing result corresponding to the image to be smoothed.

[0013] According to a texture-aware smoothing method based on local range provided by the present invention, the first update model is iteratively solved to obtain a second smoothing result corresponding to the image to be smoothed, and then the method includes:

[0014] Perform Gaussian filtering on the second smoothing result to obtain a third smoothing result.

[0015] According to a texture-aware smoothing method based on local range provided by the present invention, the first update model is iteratively solved to obtain a second smoothing result corresponding to the image to be smoothed, and then the method further includes:

[0016] Perform joint bilateral filtering on the second smoothed result to obtain a fourth smoothed result.

[0017] According to a texture-aware smoothing method based on local range provided by the present invention, the first update model is iteratively solved to obtain a second smoothing result corresponding to the image to be smoothed, and then the method further includes:

[0018] Based on a globally optimized relative Gaussian filtering algorithm, the second smoothing result is optimized to obtain a fifth smoothing result.

[0019] According to a texture-aware smoothing method based on local range provided by the present invention, solving the texture-aware smoothing model to obtain a first smoothing result corresponding to the image to be smoothed includes:

[0020] The texture regularization term is adopted The equivalent substitution is performed in the norm form to obtain the substitution result;

[0021] Decomposing the replacement result, and substituting the decomposed result into the texture-aware smoothing model to obtain an alternative smoothing model;

[0022] The candidate smoothing model is converted into a matrix form, and the candidate smoothing model in the matrix form is solved to obtain the first smoothing result.

[0023] According to a texture perception smoothing method based on local extreme difference provided by the present invention, the texture perception operator is expressed based on the following formula:

[0024]

[0025] in, is the texture perception operator, S represents the output image, is the extreme value of the pixel value in the local window of the output image, represents the gradient of the output image, and r is the size of the local window.

[0026] The present invention also provides a texture perception smoothing device based on local extreme difference, comprising:

[0027] An image acquisition module, used for acquiring an image to be smoothed;

[0028] A model solving module, used for inputting the image to be smoothed into a texture-aware smoothing model based on local range, and solving the texture-aware smoothing model to obtain a first smoothing result corresponding to the image to be smoothed;

[0029] The texture perception smoothing model includes a data fidelity term and a texture regularization term, wherein the data fidelity term is determined based on an input image and an output image, and the texture regularization term is determined based on an L1 norm of a texture perception operator;

[0030] The texture perception operator is determined based on the extreme difference of pixel values ​​within a local window of the output image and the gradient of the output image.

[0031] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the texture-perceived smoothing method based on local extreme differences as described above is implemented.

[0032] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the texture-perceived smoothing method based on local extreme differences as described above is implemented.

[0033] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements any one of the above-mentioned texture-perceived smoothing methods based on local extreme differences.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] The texture perception smoothing method and device based on local extreme difference provided by the present invention first obtain an image to be smoothed; then input the image to be smoothed into a texture perception smoothing model based on local extreme difference, and solve the texture perception smoothing model to obtain a first smoothing result corresponding to the image to be smoothed. In the texture perception smoothing model based on local extreme difference, since the texture regularization term is determined based on the L1 norm of the texture perception operator, it has a strong discriminant effect on the texture and structure of the image to be smoothed, so that the texture features in the image to be smoothed can be accurately perceived, the amount of information obtained can be increased, and the quality and accuracy of the first smoothing result can be improved, and the universality is strong, the processing efficiency is high, and the subsequent image accurate recognition is guaranteed. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on the drawings in the following description without paying any creative work.

[0037] Figure 1 It is a flow chart of the texture perception smoothing method based on local extreme difference provided by the present invention;

[0038] Figure 2 It is a structural schematic diagram of a texture perception smoothing device based on local extreme difference provided by the present invention;

[0039] Figure 3 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] Figure 1 FIG. 1 is a flow chart of a texture perception smoothing method based on local extreme difference provided in an embodiment of the present invention, such as Figure 1 As shown, the method includes:

[0042] S1, obtaining the image to be smoothed;

[0043] S2, inputting the image to be smoothed into a texture-aware smoothing model based on local range, and solving the texture-aware smoothing model to obtain a first smoothing result corresponding to the image to be smoothed;

[0044] The texture perception smoothing model includes a data fidelity term and a texture regularization term, wherein the data fidelity term is determined based on an input image and an output image, and the texture regularization term is determined based on an L1 norm of a texture perception operator;

[0045] The texture perception operator is determined based on the extreme difference of pixel values ​​within a local window of the output image and the gradient of the output image.

[0046] Specifically, the texture-perceptual smoothing method based on local extreme differences provided in an embodiment of the present invention is executed by a texture-perceptual smoothing device based on local extreme differences, which can be configured in a computer, which can be a local computer or a cloud computer. The local computer can be a computer, a tablet, etc., which is not specifically limited here.

[0047] First, step S1 is performed to obtain an image to be smoothed. The image to be smoothed may be an image under uniform light that needs to be smoothed, for example, an image under uniform light conditions obtained in various application scenarios, so as to reduce the impact of light problems on the smoothing result.

[0048] Then, step S2 is executed to input the image to be smoothed into a texture-aware smoothing model based on local extreme differences, and the texture-aware smoothing model is solved to obtain a first smoothing result corresponding to the image to be smoothed.

[0049] The texture-aware smoothing model based on local extremes can be expressed by the following formula (1):

[0050] (1)

[0051] in, represents the output image, represents the input image, represents the data fidelity term, represents the texture regularization term, represents the texture-aware operator, Represents the smoothness parameter, which is used to control the degree of smoothness and is a constant. Represents the L1 norm, which has a certain sparsity effect.

[0052] Texture-aware operators It can be expressed by the following formula (2):

[0053] (2)

[0054] in, is a texture perception operator, S represents the output image, is the extreme value of the pixel value in the local window of the output image, represents the gradient of the output image, Represents the gradient operation, which is divided into x and y directions in actual operation.

[0055] is the pixel value extreme value in the local window of the output image, which refers to the difference between the maximum pixel value and the minimum pixel value in the local window, and r is the size of the local window. It is understood that the shape of the local window can be set as needed, for example, it can be circular or rectangular. If the shape of the local window is circular, r can be the radius of the local window, and if the shape of the local window is rectangular, r can be the side length of the local window.

[0056] In the embodiment of the present invention, because the oscillation frequency of texture is higher than that of structure, the difference between the two in local range is obvious. The introduction of texture perception operator can make the texture perception smoothing model based on local range give a larger penalty to the pixels that need to be filtered, playing a role of strong discrimination.

[0057] Therefore, compared with the operation of performing convolution weighted summation on the pixel values ​​at each pixel point in the local window, the extreme difference of the pixel values ​​in the local window in the embodiment of the present invention can perceive small oscillations, amplify the gap between the structure and texture areas, and realize the discrimination of texture. For weak structures and details that are difficult to discern, the size of the local window is set to be smaller than the preset size to avoid loss. The preset size can be set as needed to ensure that the size of the local window is small.

[0058] When solving the texture-aware smoothing model based on local extreme differences, the image to be smoothed can be used as the input image, and the texture-aware smoothing model based on local extreme differences can be solved once, and the output image obtained is the first smoothing result corresponding to the image to be smoothed. Since the smoothing strength of one solution is limited, multiple iterations can be introduced. At this time, the first smoothing result obtained in the previous iteration is used as the input image of the current iteration, and the output image of the current iteration is the first smoothing result obtained in the current iteration. It can be understood that the input image of the first iteration is the image to be smoothed.

[0059] The texture-aware smoothing method based on local extreme difference provided in the embodiment of the present invention first obtains the image to be smoothed; then the image to be smoothed is input into the texture-aware smoothing model based on local extreme difference, and the texture-aware smoothing model is solved to obtain the first smoothing result corresponding to the image to be smoothed. In the texture-aware smoothing model based on local extreme difference, since the texture regularization term is determined based on the L1 norm of the texture-aware operator, it has a strong discriminant effect on the texture and structure of the image to be smoothed, so that the texture features in the image to be smoothed can be accurately perceived, the amount of information obtained can be increased, and the quality and accuracy of the first smoothing result can be improved. It has strong universality and high processing efficiency, which provides a guarantee for subsequent accurate image recognition.

[0060] On the basis of the above embodiment, solving the texture-aware smoothing model to obtain a first smoothing result corresponding to the image to be smoothed includes:

[0061] The texture regularization term is adopted The equivalent substitution is performed in the norm form to obtain the substitution result;

[0062] Decomposing the replacement result, and substituting the decomposed result into the texture-aware smoothing model to obtain an alternative smoothing model;

[0063] The candidate smoothing model is converted into a matrix form, and the candidate smoothing model in the matrix form is solved to obtain the first smoothing result.

[0064] Specifically, when solving the texture perception smoothing model based on local extremes, since the L1 norm is difficult to calculate directly, it needs to be converted into an equivalent form to facilitate the subsequent sparse solution. Here, the absolute value form is used for equivalent substitution, and the texture perception operator has the conversion process of formula (3):

[0065] (3)

[0066] in, and It is the smallest positive value to ensure that the denominator is not 0. For absolute value calculation, represents the L2 norm, represents the square of the L2 norm, which is equivalent to . For alternative results.

[0067] Observe the substitution result. For the convenience of subsequent representation, it is decomposed into two parts. The substitution result can be decomposed into the decomposition result shown in formula (4):

[0068] ; (4)

[0069] The alternative result is decomposed and substituted into the texture-aware smoothing model based on local extreme difference. Since the sparse solution is required to calculate the two-dimensional image later, u and w need to be divided into the horizontal x direction and the vertical y direction. The texture-aware smoothing model of formula (1) is rewritten to obtain the alternative smoothing model:

[0070] (5)

[0071] in, is the x-component of u, is the y-component of u, is the x-component of w, is the y-direction component of .

[0072] To facilitate further derivation and calculation, the alternative smoothing model is converted into a matrix form, as shown in formula (6):

[0073] (6)

[0074] in, and represents the Toeplitz matrix of the discrete gradient operator in the horizontal x direction and the vertical y direction, , , and are all diagonal matrices, and their diagonal values ​​are , , and correspond.

[0075] Furthermore, the minimization iterative solution of formula (6) can be performed by setting formula (6) equal to 0 to obtain the linear form shown in the following formula (7):

[0076] (7)

[0077] Among them, 1 represents the identity matrix, and j represents the x and y directions.

[0078] In the embodiment of the present invention, a process of solving the texture-perceptual smoothing model based on local extreme differences is provided, which can realize a fast solution to the texture-perceptual smoothing model based on local extreme differences.

[0079] On the basis of the above embodiment, solving the texture-aware smoothing model to obtain a first smoothing result corresponding to the image to be smoothed includes:

[0080] Using the first smoothing result obtained in the current iteration as the output image and the first smoothing result obtained in the previous iteration as the input image, updating the texture-aware smoothing model to obtain a first updated model;

[0081] The first update model is iteratively solved to obtain a second smoothing result corresponding to the image to be smoothed.

[0082] Specifically, in order to avoid the situation where the smoothing strength of a single solution is weak and the first smoothing result obtained is of poor quality, when solving the texture-aware smoothing model based on local extreme differences, the first smoothing result obtained in the current iteration can be used as the output image, and the first smoothing result obtained in the previous iteration can be used as the input image to update the texture-aware smoothing model based on local extreme differences to obtain a first updated model.

[0083] The first update model can be expressed as:

[0084] (8)

[0085] in, is the first smoothing result obtained in the previous iteration, is the first smoothing result obtained in the current iteration, i is the number of the current iteration, and N is the preset number of iterations, that is, the maximum number of iterations.

[0086] Thereafter, the above-mentioned solving process is iterated until a preset number of iterations is reached, and the first smoothing result obtained in the current iteration is used as the second smoothing result corresponding to the image to be smoothed.

[0087] In the embodiment of the present invention, by iteratively solving the first update model, the quality of the obtained second smoothing result can be higher and the accuracy can be stronger. Moreover, the gradient in the first update model can guide the iterative process, thereby effectively improving the quality of the second smoothing result.

[0088] On the basis of the above embodiment, the first update model is iteratively solved to obtain a second smoothing result corresponding to the image to be smoothed, and then the method includes:

[0089] Perform Gaussian filtering on the second smoothing result to obtain a third smoothing result.

[0090] Specifically, in order to suppress high-frequency information in the second smoothing result, after the first updating model is iteratively solved to obtain the second smoothing result, Gaussian filtering may be performed on the second smoothing result to obtain the third smoothing result.

[0091] That is:

[0092] (9)

[0093] in, is the second smoothing result, is the Gaussian filter operator, is the third smoothing result. The Gaussian filter operator can be expressed as:

[0094] (10)

[0095] in, represents the size of the filter window, represents any point in the filter window, Indicates the center point of the filter window.

[0096] On the basis of the above embodiment, the first update model is iteratively solved to obtain a second smoothing result corresponding to the image to be smoothed, and then the following step further includes:

[0097] Perform joint bilateral filtering on the second smoothed result to obtain a fourth smoothed result.

[0098] Specifically, since the local extreme difference significantly distinguishes the areas on both sides of the edge, the texture of some complex images exists in a large-scale form on the edge. Given the weak edge-preserving ability of Gaussian filtering, the iteration of formula (8) cannot smooth the edge texture without loss. To solve this problem, a simple and fast local filter can be used to replace the Gaussian filter to achieve edge-preserving performance. Therefore, after iteratively solving the first update model to obtain the second smoothing result, the second smoothing result can be subjected to joint bilateral filtering to obtain the fourth smoothing result.

[0099] Here, the second smoothing result may be input into the joint bilateral filtering model, and the joint bilateral filtering model may be solved to obtain a fourth smoothing result.

[0100] The joint bilateral filtering model can be expressed as:

[0101] (11)

[0102] in, is the pixel value of pixel point p in the fourth smoothing result, represents the neighborhood of pixel point p, q represents any point in the neighborhood of pixel point p, is the filter weight at pixel p, represents the pixel value of pixel point p in the second smoothing result, represents the pixel value of pixel point q in the second smoothing result, and They are the spatial weight distribution function and the pixel local range weight distribution function respectively.

[0103] In the embodiment of the present invention, by performing joint bilateral filtering on the second smoothing result, the fourth smoothing result obtained can retain the edge features in the image to be smoothed.

[0104] On the basis of the above embodiment, the first update model is iteratively solved to obtain a second smoothing result corresponding to the image to be smoothed, and then the following step further includes:

[0105] Based on a globally optimized relative Gaussian filtering algorithm, the second smoothing result is optimized to obtain a fifth smoothing result.

[0106] Specifically, in view of the situation where some images have details and texture scales similar to each other, it is necessary to maintain the structural features in the image to be smoothed. Therefore, after iteratively solving the first update model to obtain the second smoothing result, the second smoothing result can be optimized based on the globally optimized relative Gaussian filtering algorithm to obtain the fifth smoothing result.

[0107] Here, the second smoothing result may be input into a relative Gaussian filtering model based on global optimization, and the fifth smoothing result may be obtained by solving the relative Gaussian filtering model based on global optimization.

[0108] The relative Gaussian filtering model based on global optimization can be expressed as:

[0109] (12)

[0110] in, and are the size of the first window and the size of the second window respectively, and . and Represent the gradient of the fifth smoothing result in the x and y directions respectively.

[0111] In the embodiment of the present invention, the second smoothing result is optimized based on the global optimized relative Gaussian filtering algorithm, so that the obtained fifth smoothing result can retain the structural features in the image to be smoothed.

[0112] like Figure 2 As shown, based on the above embodiment, an embodiment of the present invention provides a texture perception smoothing device based on local extreme difference, including:

[0113] An image acquisition module 21, used to acquire an image to be smoothed;

[0114] A model solving module 22, used for inputting the image to be smoothed into a texture-aware smoothing model based on local range, and solving the texture-aware smoothing model to obtain a first smoothing result corresponding to the image to be smoothed;

[0115] The texture perception smoothing model includes a data fidelity term and a texture regularization term, wherein the data fidelity term is determined based on an input image and an output image, and the texture regularization term is determined based on an L1 norm of a texture perception operator;

[0116] The texture perception operator is determined based on the extreme difference of pixel values ​​within a local window of the output image and the gradient of the output image.

[0117] Based on the above embodiment, the model solving module is specifically used for:

[0118] Using the first smoothing result obtained in the current iteration as the output image and the first smoothing result obtained in the previous iteration as the input image, updating the texture-aware smoothing model to obtain a first updated model;

[0119] The first update model is iteratively solved to obtain a second smoothing result corresponding to the image to be smoothed.

[0120] Based on the above embodiment, the model solving module is further specifically used for:

[0121] Perform Gaussian filtering on the second smoothing result to obtain a third smoothing result.

[0122] Based on the above embodiment, the model solving module is further specifically used for:

[0123] Perform joint bilateral filtering on the second smoothed result to obtain a fourth smoothed result.

[0124] Based on the above embodiment, the model solving module is further specifically used for:

[0125] Based on a globally optimized relative Gaussian filtering algorithm, the second smoothing result is optimized to obtain a fifth smoothing result.

[0126] Based on the above embodiment, the model solving module is specifically used for:

[0127] The texture regularization term is equivalently replaced by an L2 norm to obtain a replacement result;

[0128] Decomposing the replacement result, and substituting the decomposed result into the texture-aware smoothing model to obtain an alternative smoothing model;

[0129] The candidate smoothing model is converted into a matrix form, and the candidate smoothing model in the matrix form is solved to obtain the first smoothing result.

[0130] Based on the above embodiment, the texture perception operator is expressed based on formula (2).

[0131] Specifically, the functions of each module in the local range-based texture perception smoothing device provided in the embodiment of the present invention correspond one-to-one to the operating procedures of each step in the above-mentioned method embodiment, and the effects achieved are also consistent. Please refer to the above-mentioned embodiment for details, and no further details will be given in the embodiment of the present invention.

[0132] Figure 3 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 3 As shown, the electronic device may include: a processor (Processor) 310, a communication interface (Communications Interface) 320, a memory (Memory) 330 and a communication bus 340, wherein the processor 310, the communication interface 320, and the memory 330 communicate with each other through the communication bus 340. The processor 310 may call the logic instructions in the memory 330 to execute the texture perception smoothing method based on local extreme difference provided in the above embodiments.

[0133] In addition, the logic instructions in the above-mentioned memory 330 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0134] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the texture-perceived smoothing method based on local extremes provided in the above-mentioned embodiments.

[0135] On the other hand, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when the computer program is executed by a processor to execute the texture-perceived smoothing method based on local range provided in the above embodiments.

[0136] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0137] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0138] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A texture-aware smoothing method based on local range, characterized in that: include: Obtaining the image to be smoothed; Inputting the image to be smoothed into a texture-aware smoothing model based on local range, and solving the texture-aware smoothing model to obtain a first smoothing result corresponding to the image to be smoothed; The texture perception smoothing model includes a data fidelity term and a texture regularization term, wherein the data fidelity term is determined based on an input image and an output image, and the texture regularization term is determined based on an L1 norm of a texture perception operator; The texture perception operator is determined based on the extreme difference of pixel values ​​in a local window of the output image and the gradient of the output image; The texture perception operator is expressed based on the following formula: ; in, is the texture perception operator, S represents the output image, is the extreme value of the pixel value in the local window of the output image, represents the gradient of the output image, is the size of the local window; The texture-aware smoothing model based on local extremes can be expressed by the following formula: ; in, represents the input image, represents the data fidelity term, represents the texture regularization term, represents the texture-aware operator, Represents the smoothness parameter, which is used to control the degree of smoothness and is a constant. represents the L1 norm.

2. The texture perception smoothing method based on local range according to claim 1, characterized in that: The step of solving the texture-aware smoothing model to obtain a first smoothing result corresponding to the image to be smoothed includes: Using the first smoothing result obtained in the current iteration as the output image and the first smoothing result obtained in the previous iteration as the input image, updating the texture-aware smoothing model to obtain a first updated model; The first update model is iteratively solved to obtain a second smoothing result corresponding to the image to be smoothed.

3. The texture perception smoothing method based on local range according to claim 2, characterized in that: The iterative solution of the first update model to obtain a second smoothing result corresponding to the image to be smoothed comprises: Perform Gaussian filtering on the second smoothing result to obtain a third smoothing result.

4. The texture perception smoothing method based on local range according to claim 2, characterized in that: The iterative solution of the first update model to obtain a second smoothing result corresponding to the image to be smoothed further includes: Perform joint bilateral filtering on the second smoothed result to obtain a fourth smoothed result.

5. The texture perception smoothing method based on local range according to claim 2, characterized in that: The iterative solution of the first update model to obtain a second smoothing result corresponding to the image to be smoothed further includes: Based on a globally optimized relative Gaussian filtering algorithm, the second smoothing result is optimized to obtain a fifth smoothing result.

6. The texture perception smoothing method based on local range according to any one of claims 1 to 5, characterized in that: The step of solving the texture-aware smoothing model to obtain a first smoothing result corresponding to the image to be smoothed includes: The texture regularization term is equivalently replaced in a norm form to obtain a replacement result; Decomposing the replacement result, and substituting the decomposed result into the texture-aware smoothing model to obtain an alternative smoothing model; The candidate smoothing model is converted into a matrix form, and the candidate smoothing model in the matrix form is solved to obtain the first smoothing result.

7. A texture perception smoothing device based on local range, characterized in that: include: An image acquisition module, used for acquiring an image to be smoothed; A model solving module, used for inputting the image to be smoothed into a texture-aware smoothing model based on local range, and solving the texture-aware smoothing model to obtain a first smoothing result corresponding to the image to be smoothed; The texture perception smoothing model includes a data fidelity term and a texture regularization term, wherein the data fidelity term is determined based on an input image and an output image, and the texture regularization term is determined based on an L1 norm of a texture perception operator; The texture perception operator is determined based on the extreme difference of pixel values ​​in a local window of the output image and the gradient of the output image; The texture perception operator is expressed based on the following formula: ; in, is the texture perception operator, S represents the output image, is the extreme value of the pixel value in the local window of the output image, represents the gradient of the output image, is the size of the local window; The texture-aware smoothing model based on local extremes can be expressed by the following formula: ; in, represents the input image, represents the data fidelity term, represents the texture regularization term, represents the texture-aware operator, Represents the smoothness parameter, which is used to control the degree of smoothness and is a constant. represents the L1 norm.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the texture-aware smoothing method based on local extreme differences as described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the texture-aware smoothing method based on local range is implemented as described in any one of claims 1 to 6.

Images