Image recognition system and method for automatic loading of coal mines

Abstract

The present invention discloses an image recognition system and method for automatic loading of coal mines, which relates to the field of coal mine image processing technology. The steps of the method include: installing a high-resolution camera at a key position of the loading system to collect images of ore distribution in the carriage in real time, and installing a quality sensor at the bottom of the carriage to monitor the quality of the coal in the carriage in real time; dividing the carriage loading area into a unit grid of size N×N; setting a coal mine target recognition strategy to identify the offset of the coal pile in the partition; setting a loading path dynamic programming algorithm, dynamically adjusting the partition weight, and gradually finding the optimal loading path. The present invention solves the problem of low recognition accuracy caused by dust and light changes in automatic loading of coal mines, and reduces the impact of the center of gravity offset of the carriage on loading safety.

Description

Technical Field

[0001] The present invention relates to the technical field of coal mine image processing, and in particular to an image recognition system and method for automatic loading of coal mines. Background Art

[0002] The automatic loading system for coal mines is an important part of modern coal mine production. Its main task is to load the mined coal into the transport vehicles efficiently and accurately. With the expansion of coal mining scale and the increase in transportation demand, traditional loading methods have gradually exposed many problems, such as uneven loading, deviation of the center of gravity of the carriage, and low loading efficiency. These problems not only affect the production efficiency of coal mines, but also increase safety hazards.

[0003] As an advanced automation technology, image recognition technology has been widely used in industrial production in recent years. Through high-resolution cameras and advanced image processing algorithms, image recognition technology can achieve real-time monitoring and intelligent control of the coal mine loading process. However, during the coal mine loading process, environmental factors such as dust, light changes, etc. affect the image quality, as well as the complex distribution of coal mines, which puts higher requirements on the image recognition system.

[0004] The automatic loading system based on image recognition not only improves the accuracy and efficiency of the coal mine loading process, but also effectively copes with the challenges brought by complex environmental conditions, and has significant research value and practical application significance. These technological innovations will promote the intelligent upgrading of coal mining and transportation processes and improve the safety and economy of coal mine production.

[0005] For example, the existing Chinese patent application with publication number CN117775780A discloses a coal mine railway automatic quantitative loading system. By obtaining the carriage information, the carriage position of the train carriage is measured by the positioning unit to provide accurate basic data for subsequent loading control; the loading control timing is determined based on the carriage position by the chute control unit, and the opening and closing timing of the gate, the control duration and the vertical displacement of the chute are output based on the carriage information and the loading control timing by the loading control model, so as to realize the automatic coal dropping control of the coal mine in the quantitative bin, thereby improving the accuracy and efficiency of loading; the coal pile in the train carriage is scanned by a laser scanning device to obtain the coal pile point cloud data; the coal pile volume and coal pile height information are calculated based on the coal pile point cloud data; the carriage eccentricity mass is calculated based on the coal pile volume and coal pile density, and the coal pile height information and the carriage eccentricity mass can determine whether it meets the loading requirements, further improving the accuracy of loading.

[0006] However, this design relies on sensor performance, and the environment at the coal mine site is changeable, such as dust, rainy and snowy weather, temperature changes, etc., and the night environment will also have a great impact on the recognition accuracy of the coal mine. These factors will interfere with image acquisition and recognition. The performance of existing systems in different environments may be unstable, resulting in insufficient adaptability and robustness of the system. In addition, during the automatic loading process, the vehicle may move or vibrate slightly. The existing system often finds it difficult to track the specific position and posture of the vehicle in real time, resulting in the inability of the loading equipment to accurately align with the loading port, affecting the loading effect, resulting in excessive or insufficient loading, and further affecting the vehicle's load balance, car utilization and transportation safety. To this end, the present invention provides an image recognition system and method for automatic loading of coal mines. Summary of the invention

[0007] The purpose of the present invention is to provide an image recognition system for automatic loading of coal mines to solve the existing problems raised in the above-mentioned background technology.

[0008] To achieve the above-mentioned purpose, the present invention provides the following technical solution: an image recognition system for automatic loading of coal mines, comprising:

[0009] An image recognition system for automatic loading of coal mines, characterized by:

[0010] Data acquisition module, which is used to install high-resolution cameras at key locations of the loading system to collect images of ore distribution in the carriage in real time, and to install quality sensors at the bottom of the carriage to monitor the quality of the coal in the carriage in real time;

[0011] A loading partition module is used to divide the carriage loading area into a unit grid of size N×N;

[0012] Image recognition and analysis module, used to set the coal mine target recognition strategy to identify the offset of coal piles in the partition;

[0013] The path planning module is used to set the dynamic planning algorithm of the loading path, dynamically adjust the partition weights, and gradually find the optimal loading path.

[0014] The present invention is further improved in that the coal mine target recognition strategy includes a coal mine edge detection sub-strategy, which obtains the actual coal pile image in the carriage through the high-resolution camera, and is matched with a dust compensation algorithm, a coal mine light compensation algorithm and a deep coal mine monomer extraction algorithm; the dust compensation algorithm is realized through a dark channel prior, extracts all pixels of the actual coal pile image in the carriage, calculates the value of the minimum color channel of all pixels, obtains the coal mine dark channel image, selects the pixel point with the largest value from the dark channel image to obtain the atmospheric light value A, introduces an adaptive dust channel window, and determines the size of the dark channel adaptive dust channel window of each pixel according to the texture complexity, thereby obtaining the dust compensation coefficient:

[0015]

[0016] in, represents the adjustment parameter, represents the RGB channel image of the actual coal pile image in the carriage, l represents the actual coal pile image in the carriage, C represents the color channel, represents the color channel mean of the RGB channel image of the actual coal pile in the carriage, Indicates the adaptive dust channel window size.

[0017] A further improvement of the present invention is that the calculation formula for the adaptive dust channel window size is:

[0018]

[0019] in, Indicates the maximum dust channel window size set. and represents the adjustment parameter; represents the gradient strength, represents the texture complexity, Represents the pixel coordinates of the actual coal pile image in the carriage with the center point as the origin.

[0020] The present invention is further improved in that the coal mine light compensation algorithm includes extracting the dust compensation coefficient to obtain a dust compensated coal mine image. , A is the atmospheric light value, Representing an input image, cyclically calculating the average brightness of the dust-compensated coal mine image and the brightness difference of each pixel to obtain all pixel illumination values, and arranging them in order from low to high to obtain a dust-compensated coal mine image pixel illumination value sequence; extracting the sum of the first 10% of the pixel illumination values ​​of the dust-compensated coal mine image pixel illumination value sequence as a reference illumination compensation value cgb, and extracting the corresponding pixel number cnu, and obtaining the coal mine light compensation coefficient by calculating the ratio of the product of cnu and the pixel illumination value to the reference illumination compensation value, and dynamically compensating the coal mine image based on the lookup table of the brightness histogram and the product of the coal mine light compensation coefficient to obtain a dust-light compensated coal mine image.

[0021] The present invention is further improved in that the deep coal mine monomer extraction algorithm introduces the active contour model method, combines the contour CNN model trained by historical coal mine images, and deploys it into the actual coal mine identification system to obtain the coal mine individuals in the carriage. The specific steps include: obtaining the initial coal mine contour in the dust-light compensated coal mine image by the canny edge detection algorithm, and representing the initial coal mine contour as a set of discrete points , s represents the arc length parameter of the initial coal mine contour, Represents the coordinates of the contour points of the initial coal mine contour; generates a coal mine feature map F(x) through a convolutional neural network, and applies the gradient of the coal mine feature map to the external energy of the energy functional, expressed as , represents the gradient of the coal mine characteristic map, and the total energy is expressed as:

[0022]

[0023] in, represents the slope of the coal mine profile, represents the curvature of the coal mine contour, and represents the weight parameter; minimized by solving the Euler-Lagrange equation , obtain the equilibrium position of the contour, and iterate through the gradient descent method to solve it. When the change of the contour is less than the set change threshold or reaches the maximum number of iterations, the iteration stops. The contour at this time is the final identified coal mine contour.

[0024] The present invention is further improved in that the coal mine target identification strategy also includes a coal mine offset quantum strategy, which includes collecting the coal mine contour in the area once after each layer of coal mine is loaded, identifying the front ore contour image and the rear ore contour image in the area, and removing the coal mine contour in the rear ore contour image that completely overlaps with the coal mine contour in the front ore contour image to obtain a second rear ore contour image; defining the center of gravity of each ore contour as its geometric center, taking the center of the carriage as the origin, calculating the average center of gravity of the ore contour in the front ore contour image to obtain the front center of gravity coordinates, and then calculating the average center of gravity of the coal mine contour in the second rear ore contour image to obtain the rear center of gravity coordinates; merging the mass and center of gravity coordinates of the two layers of ore to obtain the overall center of gravity coordinates in the new area:

[0025]

[0026] in, Indicates the ore quality corresponding to the previous ore contour image in the area, Indicates the ore quality corresponding to the second post-position ore contour image in the area, represents the front barycentric coordinates, Represents the post-center-of-gravity coordinates, and the coal mine offset in the area is obtained by the distance between the new overall center-of-gravity coordinates in the area and the overall center-of-gravity coordinates in the area calculated when loading the previous layer of coal mine.

[0027] A further improvement of the present invention is that the path planning module includes setting an initial weight W for each partition, calculating the ratio of the set maximum offset and the difference between the coal mine offset in the area and the set maximum offset to obtain the weight change coefficient of each partition, and updating the weight of each partition by multiplying the weight change coefficient and the initial weight; using the loading path dynamic planning algorithm, the loading path is replanned according to the adjusted weight.

[0028] The present invention is further improved in that the specific steps of the dynamic programming algorithm for the loading path include: defining the state of each loading during the loading process as the center of gravity position of the partition in the current carriage; defining the maximum value of the mass difference between the partition with the maximum ore mass and the partition with the minimum ore mass as the objective function; using the center of gravity offset of each area as the cost function; when the ore is newly loaded, calculating the cumulative sum of the costs of each area as the total cost of the path; selecting the path with the smallest total cost as the loading path for the next loading; evaluating the objective function after loading, if the objective function is less than the set quality balance value, the system controls the loading equipment to perform the loading task, and if the objective function is greater than or equal to the set quality balance value, the area corresponding to the maximum center of gravity offset in the path is eliminated until the objective function is less than the set quality balance value.

[0029] On the other hand, the present invention provides an image recognition method for automatic loading of coal mines, comprising the following steps:

[0030] S1. Install high-resolution cameras at key locations of the loading system to collect images of ore distribution in the carriage in real time, and install quality sensors at the bottom of the carriage to monitor the quality of the coal in the carriage in real time;

[0031] S2, dividing the carriage loading area into unit grids of size N×N;

[0032] S3, setting the coal mine target identification strategy to identify the offset of the coal pile in the partition;

[0033] S4. Set the dynamic programming algorithm for the loading path, dynamically adjust the partition weights, and gradually find the optimal loading path.

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

[0035] 1. The present invention firstly effectively eliminates the image quality degradation caused by environmental factors (such as dust and light changes) by using a dark channel priori algorithm, a dust compensation coefficient and a light compensation algorithm, thereby ensuring accurate identification of the ore contour;

[0036] 2. The carriage is divided into multiple loading areas, and the loading weight of each area is dynamically adjusted according to the real-time monitored offset, which improves the balance and efficiency of loading. Through the dynamic path planning algorithm, the loading path can be optimized according to the real-time status of each area, avoiding excessive concentration of ore in a certain area, ensuring the uniformity of loading and the balance of the center of gravity of the carriage;

[0037] 3. By accurately calculating and dynamically adjusting the center of gravity offset, the impact of the center of gravity offset on loading safety is reduced, and the safety of loading operations is improved; by real-time monitoring and optimization of the loading path, loading time and resource waste are reduced, and the overall loading efficiency is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a framework diagram of the image recognition system used for automatic loading of coal mines according to the present invention;

[0039] Figure 2 This is a flow chart of the coal mine light compensation algorithm of the image recognition system for automatic loading of coal mines in the present invention;

[0040] Figure 3 This is a flow chart of the deep coal mine monomer extraction algorithm of the image recognition system for automatic loading of coal mines in the present invention;

[0041] Figure 4 The present invention is a flow chart of the image recognition system method used for automatic loading of coal mines. DETAILED DESCRIPTION

[0042] The technical solution of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations on the technical solution of the present invention. The embodiments of the present invention and the technical features in the embodiments may be combined with each other unless there is a conflict.

[0043] The term "and / or" is only a description of the association relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " generally indicates that the related objects are in an "or" relationship.

[0044] Example 1

[0045] Figure 1 The framework diagram of an image recognition system for automatic loading of coal mines disclosed in this embodiment is shown, including:

[0046] Data acquisition module, which is used to install high-resolution cameras at key locations of the loading system (such as the loading port, the top of the carriage, and both sides) to collect images of the ore distribution in the carriage in real time, and to install quality sensors at the bottom of the carriage to monitor the quality of the coal in the carriage in real time;

[0047] A loading partition module is used to divide the carriage loading area into a unit grid of size N×N;

[0048] Image recognition and analysis module, used to set the coal mine target recognition strategy to identify the offset of coal piles in the partition;

[0049] The path planning module is used to set the dynamic planning algorithm for the loading path, dynamically adjust the partition weights, and gradually find the optimal loading path. The algorithm encourages balanced loading and center of gravity optimization through a reward mechanism.

[0050] The coal mine target recognition strategy includes a coal mine edge detection sub-strategy, which obtains the actual coal pile image in the carriage through the high-resolution camera, and is matched with a dust compensation algorithm, a coal mine light compensation algorithm and a deep coal mine monomer extraction algorithm;

[0051] In a fog-free image, the pixel values ​​in at least one color channel of most non-sky areas are close to zero. Therefore, the dark channel prior can effectively remove the effects of fog and dust in the image. The dust compensation algorithm is implemented through the dark channel prior. All pixels of the actual coal pile image in the carriage are extracted, and the value of the minimum color channel of all pixels is calculated to obtain the dark channel image of the coal mine. The pixel point with the largest value is selected from the dark channel image to obtain the atmospheric light value A. Since the details of the coal mine image are relatively similar, the traditional dark channel prior uses a fixed window size for minimum value calculation, which may cause the details of the coal mine image to be lost or over-processed; therefore, an adaptive dust channel window is introduced to determine the size of the dark channel adaptive dust channel window of each pixel according to the texture complexity, and then the dust compensation coefficient is obtained:

[0052]

[0053] in, represents the adjustment parameter, represents the RGB channel image of the actual coal pile image in the carriage, l represents the actual coal pile image in the carriage, C represents the color channel, represents the color channel mean of the RGB channel image of the actual coal pile in the carriage, Indicates the adaptive dust channel window size:

[0054] ;

[0055] in, Indicates the maximum dust channel window size that is set; and Represents a tuning parameter used to balance the effects of gradient strength and texture complexity on window size; represents the gradient strength, represents the texture complexity, Represents the pixel coordinates of the actual coal pile image in the carriage with the center point as the origin.

[0056] Since the texture of coal mine is more obvious, the dark channel can be compensated by texture;

[0057] The formula for calculating texture complexity is: , Represents pixels Neighborhood of represents the average gray value in the neighborhood, Represents pixels Any pixel in the neighborhood, express Input.

[0058] Figure 2The present embodiment shows the coal mine light compensation algorithm flow chart of the image recognition system for automatic loading of coal mines; the coal mine light compensation algorithm includes extracting the dust compensation coefficient to obtain the dust compensation coal mine image , A is the atmospheric light value, Representing an input image, cyclically calculating the average brightness of the dust-compensated coal mine image and the brightness difference of each pixel to obtain all pixel illumination values, and arranging them in order from low to high to obtain a dust-compensated coal mine image pixel illumination value sequence; extracting the sum of the first 10% of the pixel illumination values ​​of the dust-compensated coal mine image pixel illumination value sequence as a reference illumination compensation value cgb, and extracting the corresponding pixel number cnu, and obtaining the coal mine light compensation coefficient by calculating the ratio of the product of cnu and the pixel illumination value to the reference illumination compensation value, and dynamically compensating the coal mine image based on the lookup table of the brightness histogram and the product of the coal mine light compensation coefficient to obtain a dust-light compensated coal mine image.

[0059] By estimating the global and local illumination changes of the scene, the brightness and contrast of the image are dynamically adjusted to compensate for the image quality degradation caused by illumination changes.

[0060] Figure 3 The flowchart of the deep coal mine monomer extraction algorithm of the image recognition system for automatic loading of coal mines in this embodiment is shown. The deep coal mine monomer extraction algorithm introduces the active contour model method, combines the contour CNN model trained by historical coal mine images, and is deployed in the actual coal mine identification system to obtain the coal mine individuals in the carriage. The specific steps include: obtaining the initial coal mine contour in the dust-light compensated coal mine image through the canny edge detection algorithm, and representing the initial coal mine contour as a set of discrete points , s represents the arc length parameter of the initial coal mine contour, Represents the coordinates of the contour points of the initial coal mine contour; generates a coal mine feature map F(x) through a convolutional neural network, and applies the gradient of the coal mine feature map to the external energy of the energy functional, expressed as , represents the gradient of the coal mine characteristic map, and the total energy is expressed as:

[0061]

[0062] in, represents the slope of the coal mine profile, represents the curvature of the coal mine contour, and represents the weight parameter; minimized by solving the Euler-Lagrange equation , obtain the equilibrium position of the contour, and iterate through the gradient descent method to solve it. When the change of the contour is less than the set change threshold or reaches the maximum number of iterations, the iteration stops. The contour at this time is the final identified coal mine contour.

[0063] Internal energy is used to maintain the smoothness of the contour and prevent it from being overly distorted, while external energy is used to guide the contour to approach the edge of the target and is usually determined by the gradient information of the image.

[0064] The coal mine target identification strategy also includes a coal mine offset quantum strategy, which includes collecting the coal mine contour in the area once after each layer of coal mine is loaded, identifying the front ore contour image and the rear ore contour image in the area, and removing the coal mine contour in the rear ore contour image that completely overlaps with the coal mine contour in the front ore contour image to obtain a second rear ore contour image; defining the center of gravity of each ore contour as its geometric center, taking the center of the carriage as the origin, calculating the average center of gravity of the ore contour in the front ore contour image to obtain the front center of gravity coordinates, and then calculating the average center of gravity of the coal mine contour in the second rear ore contour image to obtain the rear center of gravity coordinates; merging the mass and center of gravity coordinates of the two layers of ore to obtain the overall center of gravity coordinates in the new area:

[0065]

[0066] in, Indicates the ore quality corresponding to the previous ore contour image in the area, Indicates the ore quality corresponding to the second post-position ore contour image in the area, that is, the ore quality is recorded once for each layer filled, and the ore quality of the layer is obtained after and before filling. represents the front barycentric coordinates, Represents the post-center-of-gravity coordinates, and the coal mine offset in the area is obtained by the distance between the new overall center-of-gravity coordinates in the area and the overall center-of-gravity coordinates in the area calculated when loading the previous layer of coal mine.

[0067] The path planning module includes setting an initial weight W for each partition. By setting the target weight, the system can give priority to filling key areas, reduce the center of gravity offset of the carriage, optimize the loading balance, calculate the ratio of the set maximum offset and the difference between the coal mine offset in the area and the set maximum offset to obtain the weight change coefficient of each partition, and update the weight of each partition by multiplying the weight change coefficient and the initial weight; use the loading path dynamic planning algorithm to re-plan the loading path according to the adjusted weight.

[0068] The specific steps of the loading path dynamic programming algorithm include: defining the state of each loading process as the center of gravity position of the partition in the current carriage; defining the maximum value of the mass difference between the partition with the maximum ore mass and the partition with the minimum ore mass as the objective function; using the center of gravity offset of each area as the cost function; when the ore is newly loaded, calculating the accumulated sum of the costs of each area as the total cost of the path; selecting the path with the minimum total cost as the loading path for the next loading; evaluating the objective function after loading, if the objective function is less than the set quality balance value, the system controls the loading equipment to perform the loading task, if the objective function is greater than or equal to the set quality balance value, then eliminating the area corresponding to the maximum center of gravity offset in the path until the objective function is less than the set quality balance value.

[0069] The threshold and weight settings can be set by the operator.

[0070] Example 2

[0071] Figure 4 The flowchart of the image recognition method for automatic loading of coal mines disclosed in this embodiment is shown. Based on the same inventive concept as that of embodiment 1, the present invention provides an image recognition method for automatic loading of coal mines, including:

[0072] S1. Install high-resolution cameras at key locations of the loading system to collect images of ore distribution in the carriage in real time, and install quality sensors at the bottom of the carriage to monitor the quality of the coal in the carriage in real time;

[0073] S2, dividing the carriage loading area into unit grids of size N×N;

[0074] S3, setting the coal mine target identification strategy to identify the offset of the coal pile in the partition;

[0075] S4. Set the dynamic programming algorithm for the loading path, dynamically adjust the partition weights, and gradually find the optimal loading path.

[0076] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0077] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0078] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0079] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0080] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the enlightenment of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the purpose of the present invention and the claims, which all fall within the protection of the present invention.

Claims

1. Image recognition system for automatic loading of coal mines, characterized by: include: Data acquisition module, which is used to install high-resolution cameras at key locations of the loading system to collect images of ore distribution in the carriage in real time, and to install quality sensors at the bottom of the carriage to monitor the quality of the coal in the carriage in real time; A loading partition module is used to divide the carriage loading area into a unit grid of size N×N; Image recognition and analysis module, used to set the coal mine target recognition strategy to identify the offset of coal piles in the partition; The path planning module is used to set the dynamic planning algorithm of the loading path, dynamically adjust the partition weights, and gradually find the optimal loading path; The coal mine target identification strategy also includes a coal mine offset quantum strategy, which includes collecting the coal mine contour in the area once after each layer of coal mine is loaded, identifying the front ore contour image and the rear ore contour image in the area, and removing the coal mine contour in the rear ore contour image that completely overlaps with the coal mine contour in the front ore contour image to obtain a second rear ore contour image; defining the center of gravity of each ore contour as its geometric center, taking the center of the carriage as the origin, calculating the average center of gravity of the ore contour in the front ore contour image to obtain the front center of gravity coordinates, and then calculating the average center of gravity of the coal mine contour in the second rear ore contour image to obtain the rear center of gravity coordinates; merging the mass and center of gravity coordinates of the two layers of ore to obtain the overall center of gravity coordinates in the new area: ； in, Indicates the ore quality corresponding to the previous ore contour image in the area, Indicates the ore quality corresponding to the second post-position ore contour image in the area, represents the front barycentric coordinates, Represents the post-center-of-gravity coordinates, then the coal mine offset in the region is obtained by the distance between the new overall center-of-gravity coordinates in the region and the overall center-of-gravity coordinates in the region calculated when loading the previous layer of coal mines; The path planning module includes setting an initial weight W for each partition, calculating the ratio of the set maximum offset and the difference between the coal mine offset in the area and the set maximum offset to obtain the weight change coefficient of each partition, and updating the weight of each partition by multiplying the weight change coefficient and the initial weight; using the loading path dynamic planning algorithm to re-plan the loading path according to the adjusted weight; The specific steps of the loading path dynamic programming algorithm include: defining the state of each loading process as the center of gravity position of the partition in the current carriage; defining the maximum value of the mass difference between the partition with the maximum ore mass and the partition with the minimum ore mass as the objective function; using the center of gravity offset of each area as the cost function; when the ore is newly loaded, calculating the accumulated sum of the costs of each area as the total cost of the path; selecting the path with the minimum total cost as the loading path for the next loading; evaluating the objective function after loading, if the objective function is less than the set quality balance value, the system controls the loading equipment to perform the loading task, if the objective function is greater than or equal to the set quality balance value, then eliminating the area corresponding to the maximum center of gravity offset in the path until the objective function is less than the set quality balance value.

2. The image recognition system for automatic loading of coal mines according to claim 1 is characterized in that: The coal mine target recognition strategy includes a coal mine edge detection sub-strategy, which obtains the actual coal pile image in the carriage through the high-resolution camera, and is matched with a dust compensation algorithm, a coal mine light compensation algorithm and a deep coal mine monomer extraction algorithm; the dust compensation algorithm is implemented through a dark channel prior, extracts all pixels of the actual coal pile image in the carriage, calculates the value of the minimum color channel of all pixels, obtains the coal mine dark channel image, selects the pixel point with the largest value from the dark channel image to obtain the atmospheric light value A, introduces an adaptive dust channel window, and determines the size of the dark channel adaptive dust channel window of each pixel according to the texture complexity, thereby obtaining the dust compensation coefficient: ； in, represents the adjustment parameter, represents the RGB channel image of the actual coal pile image in the carriage, l represents the actual coal pile image in the carriage, C represents the color channel, represents the color channel mean of the RGB channel image of the actual coal pile in the carriage, Indicates the adaptive dust channel window size.

3. The image recognition system for automatic loading of coal mines according to claim 2 is characterized in that: The calculation formula for the adaptive dust channel window size is: ； in, Indicates the maximum dust channel window size set. and represents the adjustment parameter; represents the gradient strength, Represents the texture complexity, Represents the pixel coordinates of the actual coal pile image in the carriage with the center point as the origin.

4. The image recognition system for automatic loading of coal mines according to claim 3 is characterized in that: The coal mine light compensation algorithm includes extracting the dust compensation coefficient to obtain a dust compensated coal mine image , A is the atmospheric light value, Representing an input image, cyclically calculating the average brightness of the dust-compensated coal mine image and the brightness difference of each pixel to obtain all pixel illumination values, and arranging them in order from low to high to obtain a dust-compensated coal mine image pixel illumination value sequence; extracting the sum of the first 10% of the pixel illumination values ​​of the dust-compensated coal mine image pixel illumination value sequence as a reference illumination compensation value cgb, and extracting the corresponding pixel number cnu, and obtaining the coal mine light compensation coefficient by calculating the ratio of the product of cnu and the pixel illumination value to the reference illumination compensation value, and dynamically compensating the coal mine image based on the lookup table of the brightness histogram and the product of the coal mine light compensation coefficient to obtain a dust-light compensated coal mine image.

5. The image recognition system for automatic loading of coal mines according to claim 4 is characterized in that: The deep coal mine monomer extraction algorithm introduces the active contour model method, combines the contour CNN model trained by historical coal mine images, and deploys it into the actual coal mine identification system to obtain the coal mine individuals in the carriage. The specific steps include: obtaining the initial coal mine contour in the dust-light compensated coal mine image through the canny edge detection algorithm, and representing the initial coal mine contour as a set of discrete points , s represents the arc length parameter of the initial coal mine contour, Represents the coordinates of the contour points of the initial coal mine contour; generates a coal mine feature map F(x) through a convolutional neural network, and applies the gradient of the coal mine feature map to the external energy of the energy functional, expressed as , represents the gradient of the coal mine characteristic map, and the total energy is expressed as: ； in, represents the slope of the coal mine profile, represents the curvature of the coal mine contour, and represents the weight parameter; minimized by solving the Euler-Lagrange equation , obtain the equilibrium position of the contour, and iterate through the gradient descent method to solve it. When the change of the contour is less than the set change threshold or reaches the maximum number of iterations, the iteration stops. The contour at this time is the final identified coal mine contour.

6. An image recognition method for automatic loading of coal mines, based on the image recognition system for automatic loading of coal mines as claimed in any one of claims 1 to 5, characterized in that: include: S1. Install high-resolution cameras at key locations of the loading system to collect images of ore distribution in the carriage in real time, and install quality sensors at the bottom of the carriage to monitor the quality of the coal in the carriage in real time; S2, dividing the carriage loading area into unit grids of size N×N; S3, setting the coal mine target identification strategy to identify the offset of the coal pile in the partition; S4. Set the dynamic programming algorithm for the loading path, dynamically adjust the partition weights, and gradually find the optimal loading path.

Images