Underground mine image detail enhancement algorithm inspired by Kepler's law

Abstract

The invention relates to the technical field of image detail enhancement, in particular to an under-mine image detail enhancement algorithm inspired by the Kepler's law. The method comprises the following steps: initializing the positions of a current optimal matching image block and candidate image blocks on a track; selecting the candidate image block with the minimum pixel difference as the current global optimal matching according to the pixel difference between each candidate image block and the original image block; calculating the solar gravitation and the current speed of the candidate image block by using the Kepler planetary motion law, comprehensively scanning the space to jump out of local optimum by changing the mark of the search direction, and updating the position of the candidate image block; and iteratively updating to obtain a globally best matched image block, and calculating to obtain an enhanced residual feature to obtain a detail enhanced image, thereby completing detail enhancement. According to the invention, based on the idea of Kepler planetary motion law, image search is carried out based on a pixel difference value minimization problem, and a Kepler optimization search strategy is utilized to search a local optimal solution, so that the effect of image detail enhancement is improved.

Description

Technical Field

[0001] The present application relates to the technical field of image detail enhancement, and in particular to an algorithm for enhancing image details in mines inspired by Kepler's law. Background Art

[0002] With the continuous advancement of computer vision technology, the acquisition of underground mine image data has become increasingly common. However, due to numerous factors, such as equipment limitations, insufficient lighting, and dust interference, the quality of images collected in many mines fails to meet the monitoring requirements required for safe production. This issue has led to the widespread use of image detail enhancement algorithms, driving the development of related technologies such as tone mapping, compression artifact removal, and texture classification.

[0003] Existing image enhancement algorithms, based on greedy block search and matching techniques, are prone to falling into local optimal solutions when searching for the most similar image blocks, resulting in incomplete detail restoration and difficulty achieving optimal results. This is particularly evident in complex environments such as uneven lighting and dust in mines. Existing methods generally suffer from blurred edges, artifacts, high complexity, and poor generalization. Especially in harsh conditions such as humidity and low illumination in mines, these issues can severely impact monitoring equipment's ability to accurately identify hazardous areas, hindering safety in production. Summary of the Invention

[0004] The present invention aims to solve one of the problems existing in the background technology.

[0005] To this end, the present invention provides a mine image detail enhancement algorithm inspired by Kepler's law.

[0006] The technical solution adopted by the present invention to solve its technical problem is:

[0007] A Kepler-inspired mine image detail enhancement algorithm, including:

[0008] Step 1: Initialize the positions of the current best matching image block (regarded as the "sun") and each candidate image block (regarded as the "satellite") on the orbit;

[0009] Step 2: Based on the pixel difference between each candidate image block and the original image block, the candidate image block with the smallest pixel difference is selected as the current global best match, and the Euclidean distance between the candidate image block and the best match image block is calculated;

[0010] Step 3: Use Kepler's laws of planetary motion to calculate the sun's gravity and the current velocity of the candidate image block. Then, by changing the search direction flag, we can fully scan the space to escape the local optimum and update the position of the candidate image block.

[0011] Step 4: Through repeated iterative updates, the globally best matching image block is obtained, and the enhanced residual features are obtained by calculation. After amplification, the features are added to the input image to obtain the detail enhanced image, thus completing the detail enhancement.

[0012] Furthermore, in step 2, the pixel difference between the candidate image block and the original image block is obtained by calculating the loss value between the candidate image block and the original image block, the current global best match is obtained by calculating the minimum pixel difference, and the Euclidean distance is determined.

[0013] Furthermore, based on the loss value between the candidate image block and the original image block, the mass of the sun at time t is calculated as:

[0014]

[0015] The mass of planet i at time t is:

[0016] in, Represents the loss value of the kth planet (candidate image block) and the original image block.

[0017] Furthermore, the Euclidean distance R between the Sun and the i-th planet at time t is i (t): R i (t)=||X s (t)-X i (t)||2, where X s (t) represents the position of the sun at time t, X i (t) represents the position of the i-th planet at time t.

[0018] Furthermore, in step 3, the velocity of the planet is calculated. The velocity of candidate planet i at time t is expressed as:

[0019]

[0020] Among them, X i represents the position of planet i, τ and is a random number in the interval [0,1], β and φ are population weights, Δs i (t) is the step size factor used to control the planet position update amplitude, X j represents the position of planet j other than planet i; κ represents the sign of changing the search direction; a i (t) represents the major axis of the elliptical orbit of planet i at time t, which is defined by Kepler's third law, L = [x min ,y min ] is the lower bound of the solution space, U=[x max ,y max] is the upper bound of the solution space.

[0021] Furthermore, in step three, a flag κ for changing the search direction is introduced to simulate the rotation of the planet and the counterclockwise and clockwise revolution of the planet around the sun.

[0022] Furthermore, in step 3, when a planet approaches the Sun, its velocity increases dramatically, allowing it to escape the Sun's gravitational pull. In this case, if the best solution so far (called the Sun) is a local minimum, the planet's velocity escapes the Sun's gravitational pull to find a better solution. The planet's position is updated using the following formula: i (t+1)=X i (t)+λ|F i (t)|(X s (t)-X i (t))+κV i (t), where X i (t+1) is the new position of planet i at time t+1, X s (t) represents the position of the sun, κ is a sign to change the search direction, F i (t) is the gravitational force on planet i, V i (t) is the velocity of planet i.

[0023] Furthermore, in step 4, the loss value between the current candidate image block (planet i) and the original image block is calculated. If the loss value is smaller than the loss value before the movement, the position of the current planet i after the movement is taken as the new position X′ of the planet i. i The new position of the planet at (t+1) can be expressed as:

[0024]

[0025] Furthermore, the image detail enhancement mechanism based on residual learning is:

[0026]

[0027] Res=argmin∑|I(p+Δp)-I(p)|,

[0028] I enhanced =I+α×f new (Res),

[0029] Where α is the detail layer magnification factor, f new (Res) is the detail feature obtained by Kepler optimization search, Δp is the offset of the coordinate, p is the position of the coordinate, I(p) is the image block centered at p in image I, and the initial residual feature Res is obtained by calculating all image blocks, and finally the enhanced image I is obtained enhanced .

[0030] Furthermore, we focus on edge clarity in the process of obtaining the initial residual features. and texture smoothness

[0031]

[0032] in and are the first-order and second-order Hamiltonian operators of the image patch I(p), is the feasible domain of p, and the loss function is: ζ and ξ represent the regularization constants that control the prior information.

[0033] The beneficial effects of the present invention are:

[0034] (1) By drawing on the idea of ​​Kepler's laws of planetary motion, the present invention transforms the image search and matching problem into a pixel difference minimization problem, and uses Kepler's optimization search strategy to avoid falling into the local optimal solution. By simulating the mechanism of planetary motion around the sun, the algorithm can find the optimal image block matching in a global range, thereby improving the effect of image detail enhancement;

[0035] (2) The present invention blurs the original image through bilinear interpolation to obtain a smooth layer, and subtracts the smooth layer from the original image to obtain a detail layer. The detail layer is then amplified and added to the original image. This process can effectively avoid edge blurring and artifacts while maintaining image clarity and detail.

[0036] (3) The present invention uses the Kepler optimization search strategy combined with the residual learning mechanism to achieve efficient image detail enhancement at a lower computational complexity. This method does not rely on deep learning models, avoids high training costs and data dependency problems, and is suitable for real-time monitoring needs in complex environments underground mines. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The present invention will be further described below with reference to the accompanying drawings and examples.

[0038] Figure 1 This is an algorithm flow chart of the mine image detail enhancement algorithm inspired by Kepler's law in the present invention.

[0039] Figure 2 This is an algorithm block diagram of the mine image detail enhancement algorithm inspired by Kepler's law in the present invention.

[0040] Figure 3 It is a conceptual diagram of Kepler's law in the present invention.

[0041] Figure 4This is a schematic diagram comparing the processing results of the image detail enhancement algorithm of the present invention and other existing algorithms on the four-fold detail magnification image of image 0001 in the CUMT-CUMID dataset.

[0042] Figure 5 This is a schematic diagram comparing the processing results of the image detail enhancement algorithm in the present invention and other existing algorithms on the four-fold detail magnification image of image 0445 in the CUMT-CUMID dataset.

[0043] Figure 6 This is a schematic diagram comparing the processing results of the image detail enhancement algorithm in the present invention and other existing algorithms on the four-fold detail magnification image of image 00458 in the CUMT-CUMID dataset. DETAILED DESCRIPTION

[0044] The present invention will now be described in further detail with reference to the accompanying drawings, which are simplified schematic diagrams that illustrate the basic structure of the present invention in a schematic manner.

[0045] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, features defined as "first" or "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise specified, "multiple" means two or more.

[0046] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed, detachable, or integral connections; mechanical or electrical connections; direct or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0047] A Kepler-inspired mine image detail enhancement algorithm, including:

[0048] In the first step, the image search and matching is converted into a pixel difference minimization problem, and the positions of the current best matching image block (regarded as the "sun") and each candidate image block (regarded as the "satellite") on the orbit are initialized.

[0049] First, the original image is blurred by bilinear interpolation to obtain a smooth layer, and then the smooth layer is subtracted from the original image to obtain a detail layer. Subsequently, in the detail-enhanced similar image block matching, the image search and matching is transformed into a pixel difference minimization problem, and the positions of the current best matching image block (considered as the "sun") and each candidate image block (considered as the "satellite") on the orbit are initialized, including:

[0050] Each planet is randomly placed in its respective orbit, the formula is:

[0051] Xi=L+λ×(UL)(1)

[0052] where X i =[x i ,y i ] represents the position of the i-th planet (candidate solution); L = [x min ,y min ] is the lower bound of the solution space, U=[x max ,y max ] is the upper bound of the solution space, and λ is a random number between 0 and 1. The eccentricity of the elliptical orbit of each planet is e i , is a random number between 0 and 1. The orbital period is T i , is a set of random numbers related to the length of the semi-major axis of the elliptical orbit and conforming to the normal distribution.

[0053] In the second step, the pixel difference between each candidate image block and the original image block is calculated, and the candidate image block with the smallest pixel difference is selected as the current global best match, and the Euclidean distance between the candidate image block and the best matching image block is determined.

[0054] The pixel difference between the candidate image block and the original image block. Following the law of universal gravitation, the sun is simulated to control the planets to move around it in an elliptical form through gravity. The definition of gravity is as follows:

[0055]

[0056] Among them, M s and m i Represent the mass of the sun and the i-th planet respectively. It should be noted that in image processing, the maximum and minimum normalization method is used to calculate the "mass" of the image block; In addition, γ is a constant, t and T max are the current number of iterations and the maximum number of iterations respectively.

[0057] The pixel difference between the candidate image patch and the original image patch is expressed as an overall loss value. The mass of the sun and planet i at time t is obtained by calculating the loss value with the original image patch:

[0058]

[0059] in represents the loss value of the kth planet (candidate image block) and the original image block. At the same time, R i (t) represents the Euclidean distance between the Sun and the i-th planet at time t, defined as follows:

[0060] R i (t)=||X s (t)-X i (t)||2 (5)

[0061] where X s (t) represents the position of the sun at time t, X i (t) represents the position of the i-th planet at time t.

[0062] In the third step, Kepler's laws of planetary motion are used to calculate the solar gravity and the current velocity of the candidate image block, and the position of the candidate image block is updated by fully scanning the space by changing the sign of the search direction.

[0063] S3.1 Calculate planetary velocity. The velocity of a planet is affected by the gravitational pull of the sun. If a planet is close to the sun, the sun's gravitational pull is relatively strong, and the planet will try to increase its speed to avoid being pulled toward the sun. However, if a planet is far from the sun, the sun's gravitational pull is relatively weak, and the planet's speed will slow down. This relationship can be expressed by the following equation:

[0064]

[0065] Where V i (t) represents the velocity of candidate planet i at time t, X i represents the position of planet i, τ and is a random number in the interval [0,1]. β and φ are population weights, Δs i (t) is the step size factor used to control the planet position update amplitude. j represents the positions of other planets except planet i; κ represents the sign of changing the search direction; a i (t) represents the major axis of the elliptical orbit of planet i at time t, which is defined by Kepler's third law:

[0066]

[0067] If R iIf (t)≤0.5, it means that the planet is close to the sun and will increase its speed to prevent it from drifting towards the sun due to the huge gravitational force of the sun. Otherwise, the planet's speed will slow down.

[0068] S3.2 Escaping Local Optima. In the solar system, most planets orbit the sun counterclockwise, rotating on their own axes; however, some orbit the sun clockwise. The proposed Kepler Optimization Based Detail Enhancement (KODE) algorithm for underground mine image detail enhancement, inspired by Kepler's law, simulates this behavior by using a flag κ that changes the search direction, allowing the planet to fully scan the search space and avoid falling into local optima.

[0069] S3.3 Update the positions of the planets. As mentioned earlier, the planets revolve around the sun in their own elliptical orbits. During the rotation, the planets will approach the sun for a certain period of time and then move away from the sun. When the planet moves away from the sun, the planet's speed slows down while moving away from the sun to seek a new solution. When a planet approaches the sun, its speed increases dramatically, allowing it to escape the sun's gravitational pull. In this case, if the best solution so far (called the sun) is a local minimum, the planet's speed will escape the sun's gravitational pull to find a better solution. The position of the planet is updated using the following formula:

[0070] X i (t+1)=X i (t)+λ|F i (t)|(X s (t)-X i (t))+kV i (t) (10)

[0071] where X i (t+1) is the new position of planet i at time t+1, X s (t) represents the position of the sun, and k is a sign to change the search direction. k is a random number. Since the direction of planetary motion may be counterclockwise or clockwise, this κ is used in the planetary motion formula (10) to randomly expand the search range to better find the optimal solution.

[0072] In the fourth step, the globally best matching image block is obtained through repeated iterative updates, and the enhanced residual features are obtained by calculation. After amplification, they are added to the input image to obtain a detail-enhanced image, thereby completing the detail enhancement.

[0073] This step is equivalent to a greedy strategy, which calculates the loss value between the current candidate image block (planet i) and the original image block. If the loss value is smaller than the loss value before the movement, the position of the current planet i after the movement is taken as the new position X′ of the planet i. i (t+1). Otherwise, the current position of planet i remains unchanged and is still X i (t).

[0074] The new position of the planet can be expressed as:

[0075]

[0076] Obtain f through Kepler optimization search new (Res) detail features are then added to the input image to obtain the enhanced image I enhanced , complete detail enhancement. It is worth noting that the relevant formula of the image detail enhancement mechanism based on residual learning is:

[0077]

[0078] Res=argmin∑|I(p+Δp)-I(p)| (13)

[0079] I enhanced =I+α×f new (Res) (14)

[0080] Where α is the detail layer magnification factor, f new (Res) is the detail layer, Δp is the offset of the coordinate, p is the position of the coordinate, and I(p) is the image block centered at p in image I. By calculating all image blocks, the initial residual feature Res is obtained, and finally the enhanced image I is obtained enhanced .

[0081] In addition, the process of initializing residual features in this invention is different from that of previous search and matching methods. This difference is reflected in two aspects. First, IPRH only uses local features, but the features used in our method are almost all features from non-local areas. Second, the loss function is different. IPRH only considers pixel loss. As shown in Equation (16), the loss function in our proposed method is redesigned, and two new regularization terms make the matching process focus not only on pixel differences, but also on edge clarity and texture smoothness, which are defined in Equations (17) and (18): and

[0082]

[0083] Loss Function This is a new block matching criterion designed by the present invention, which is optimized to obtain Δp, namely and are the first-order and second-order Hamiltonian operators of the image patch I(p), is the feasible region of p, ζ and ξ represent the regularization constants that control the prior information.

[0084] In summary, the advantages and positive effects of the present invention are as follows: the present invention draws on the idea of ​​Kepler's law of planetary motion, regards each candidate image block as a "planet", and randomly updates it around the current best image block (i.e., the "sun"), thereby more effectively exploring and utilizing the search space, and ultimately finding the most similar image block in the world, thereby improving the matching accuracy and avoiding falling into the local optimal solution. At the same time, a large number of experiments have proved that our algorithm can achieve better results in quantitative indicator testing and visual effect evaluation.

[0085] The technical solution of the present invention is further described below with reference to the accompanying drawings. First, a bilinear difference is performed on the input image to obtain the smooth layer and detail layer of the image, and the residual feature Res is obtained, which is an estimate of the detail layer signal. Second, the image search and matching problem is transformed into a pixel difference minimization problem, and the Kepler optimization search is used to avoid falling into the local optimum, and the best matching image block is obtained. The enhanced residual feature f is then calculated. new (Res) and adds it to the input image to obtain an enhanced image, thereby completing detail enhancement. This invention uses the Kepler optimization algorithm for iterative solution, aiming to minimize the error function and find the optimal image block match, thereby effectively avoiding falling into local optimality and more efficiently searching and utilizing space. The overall process of searching and matching based on Kepler's law to obtain the best matching image block is shown in pseudo code:

[0086] Input: original image block u, population number N, maximum number of cycles T max , initial gravitational constant G0, gravitational decay rate constant γ, eccentricity e i , orbital period T i

[0087] Output: The best matching image patch v of the original image patch u

[0088] Initialize the planet (candidate image block) position X by formula (1) i

[0089] Calculate the loss value of each image block and the original image block, and find the current best matching image block (ie the sun) X s

[0090] Initialize t=0

[0091] repeat

[0092] For each i∈{1, 2, ...N}, execute:

[0093] 1) Calculate the Euclidean distance R between the current best matching image block and the candidate image block using formula (5): i (t)

[0094] 2) Calculate the solar gravity F of the candidate image block using formula (2): i (t)

[0095] 3) Calculate the velocity V of the candidate image block using formula (6) i (t)

[0096] 4) Update the position X of the image block using formula (10) i (t+1)

[0097] 5) Update the candidate image block X using the greedy strategy using formula (11) i (t+1) and the position X of the best matching image block s

[0098] End the loop

[0099] t=t+1

[0100] Until t = T max

[0101] Return the best matching image patch v

[0102] The following is a detailed description of the technical effects of the present invention in combination with performance testing and experimental analysis. Existing effective image detail enhancement algorithms include the MGP algorithm, the CSGIS algorithm, the FSPIS algorithm, and the Easy2Hard algorithm. The CUMT-CUMID dataset is a professional image dataset collected in multiple coal mines using intrinsically safe mine alarm cameras. It contains a large number of high-definition images of typical mine scenes, covering image samples under different lighting conditions and environmental conditions, and truly reflects the impact of the complex underground environment on image quality. This makes the dataset particularly suitable for developing and verifying various image processing algorithms in mine environments.

[0103] In order to illustrate the technical effect of the present invention, we give Figure 4 、 Figure 5 and Figure 6 The experimental results. Figure 4In the figure, Figure a is the original 0001 image in the CUMT-CUMID dataset, small figure a1 is the original image after extracting four times the details, small figure a2 is the image after detail enhancement using the Easy2Hard algorithm, small figure a3 is the image after detail enhancement using the algorithm of this application, and small figure a4 is the image after detail enhancement using the FSPIS algorithm. Figure 4 In the a2, a3, and a4 panels, the Easy2Hard algorithm produces a large amount of white noise, which seriously affects the clarity of the original texture, while the FSPIS algorithm mistakenly changes the background color from gray to light green.

[0104] exist Figure 5 In the figure, Figure b is the original 0445 image in the CUMT-CUMID dataset, the small figure b1 is the original image after extracting four times the details, the small figure b2 is the image after detail enhancement using the CSGIS algorithm, the small figure b3 is the image after detail enhancement using the algorithm of this application, and the small figure b4 is the image after detail enhancement using the MGP algorithm. Figure 5 In the sub-images b2, b3, and b4, the CSGIS algorithm has a serious halo effect and the noise is further enhanced, while the MGP algorithm has dense areas of color estimation errors.

[0105] exist Figure 6 In the figure, Figure c is the original 0458 image in the CUMT-CUMID dataset, the small figure c1 is the original image with four times the details extracted, the small figure c2 is the image after the details are enhanced using the MGP algorithm, the small figure c3 is the image after the details are enhanced using the algorithm of this application, and the small figure c4 is the image after the details are enhanced using the Easy2Hard algorithm. Figure 5 In panels c2, c3, and c4, the MGP algorithm estimates the flesh color as light blue, while the Easy2Hard algorithm also loses large areas of texture, resulting in poor estimation results. In contrast, the KODE algorithm proposed in this paper performs better in capturing the feature representation of the image detail layer. Without changing the original detail structure and color, this algorithm effectively enhances detail, avoids color deviation, and achieves a superior display effect. This confirms that our algorithm has the strongest ability to enhance detail while preserving image structural information.

[0106] In summary, the above experiments demonstrate the effectiveness of the present invention, which utilizes the Kepler optimization concept to match more similar image blocks in a more global scope, avoiding the problem of local optimality and improving the ability to enhance image details.

[0107] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0108] With the above-described preferred embodiments of the present invention as a guide, and with reference to the above description, relevant personnel are fully capable of making various changes and modifications without departing from the technical spirit of this invention. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A mine image detail enhancement algorithm inspired by Kepler's law, characterized by: include, Step 1: Initialize the positions of the current best matching image block (regarded as the "sun") and each candidate image block (regarded as the "satellite") on the orbit; Step 2: Based on the pixel difference between each candidate image block and the original image block, the candidate image block with the smallest pixel difference is selected as the current global best match, and the Euclidean distance between the candidate image block and the best match image block is calculated; Step 3: Use Kepler's laws of planetary motion to calculate the sun's gravity and the current velocity of the candidate image block. Then, by changing the search direction flag, we can fully scan the space to escape the local optimum and update the position of the candidate image block. Step 4: Through repeated iterative updates, the globally best matching image block is obtained, and the enhanced residual features are obtained by calculation. After amplification, the features are added to the input image to obtain the detail enhanced image, thus completing the detail enhancement.

2. The mine image detail enhancement algorithm inspired by Kepler's law according to claim 1, characterized in that: In step 2, the pixel difference between the candidate image block and the original image block is obtained by calculating the loss value between the candidate image block and the original image block, the current global best match is obtained by calculating the minimum pixel difference, and the Euclidean distance is determined.

3. The mine image detail enhancement algorithm inspired by Kepler's law according to claim 2, characterized in that: Based on the loss value of the candidate image block and the original image block, the mass of the sun at time t is calculated as: The mass of planet i at time t is: in, Represents the loss value of the kth planet (candidate image block) and the original image block.

4. The mine image detail enhancement algorithm inspired by Kepler's law according to claim 2, characterized in that: The Euclidean distance R between the sun and the i-th planet at time t i (t): R i (t)=||X s (t)-X i (t)||2, where X s (t) represents the position of the sun at time t, X i (t) represents the position of planet i at time t.

5. The mine image detail enhancement algorithm inspired by Kepler's law according to claim 1, characterized in that: In step 3, the velocity of the planet is calculated. The velocity of candidate planet i at time t is expressed as: Among them, X i represents the position of planet i, τ and is a random number in the interval [0,1], β and φ are population weights, Δs i (t) is the step size factor used to control the planet position update amplitude, X j represents the position of planet j other than planet i; κ represents the sign of changing the search direction; a i (t) represents the major axis of the elliptical orbit of planet i at time t, which is defined by Kepler's third law, L = [x min ,y min ] is the lower bound of the solution space, U=[x max ,y max ] is the upper bound of the solution space.

6. The mine image detail enhancement algorithm inspired by Kepler's law according to claim 5, characterized in that: In step three, a flag κ is introduced to change the search direction to simulate the rotation of the planet and the counterclockwise and clockwise revolution of the planet around the sun.

7. The mine image detail enhancement algorithm inspired by Kepler's law according to claim 6, characterized in that: In step 3, when a planet approaches the Sun, its velocity increases dramatically, allowing it to escape the Sun's gravitational pull. In this case, if the best solution so far (called the Sun) is a local minimum, the planet's velocity escapes the Sun's gravitational pull to find a better solution. The planet's position is updated using the following formula: i (t+1)=X i (t)+λ|F i (t)|(X s (t)-X i (t))+kV i (t), where X i (t+1) is the new position of planet i at time t+1, X s (t) represents the position of the sun, κ is the sign of changing the search direction, F i (t) is the gravitational force on planet i, V i (t) is the velocity of planet i.

8. The mine image detail enhancement algorithm inspired by Kepler's law according to claim 1, characterized in that: In step 4, the loss value between the current candidate image block (planet i) and the original image block is calculated. If the loss value is smaller than the loss value before the movement, the position of the current planet i after the movement is taken as the new position X′ of the planet i. i The new position of the planet at (t+1) can be expressed as:

9. The mine image detail enhancement algorithm inspired by Kepler's law according to claim 8, characterized in that: The image detail enhancement mechanism based on residual learning is: Res=argminΣ|I(p+Δp)-I(p)|, I enhanced =Ⅰ+α×f new (Res), Where α is the detail layer magnification factor, f new (Res) is the detail feature obtained by Kepler optimization search, △p is the offset of the coordinate, p is the position of the coordinate, I(p) is the image block centered on p in image I, and the initial residual feature Res is obtained by calculating all image blocks, and finally the enhanced image I is obtained enhanced .

10. The mine image detail enhancement algorithm inspired by Kepler's law according to claim 9, characterized in that: Focus on edge clarity in the process of obtaining initial residual features and texture smoothness in and are the first-order and second-order Hamiltonian operators of the image patch I(p), is the feasible domain of p, and the loss function is: ζ and ξ represent the regularization constants that control the prior information.

Images