Remote mine site image processing method and system

Abstract

The present disclosure provides a remote mine image processing method and system, belonging to the technical field of image processing, which comprises: inputting target environment parameters into a target model to determine noise reduction parameters; determining a target noise reduction algorithm based on the noise reduction parameters; processing target data based on the target noise reduction algorithm to obtain mine data, and sending the mine data to a display device of a mine monitoring center; wherein the target environment parameters are environment parameters of a target region corresponding to the target data, and the target data includes monitoring image data of the target region. The remote mine image processing method and system provided by the present disclosure can improve the effect of mine image processing.

Description

TECHNICAL FIELD

[0001] The present disclosure belongs to the technical field of image processing, and more particularly, to a remote mine image processing method and system. BACKGROUND

[0002] In the modern mining field, a remote mine monitoring system has become an important means to ensure production safety and improve operational efficiency. Through monitoring devices deployed at various locations in the mine, a large amount of monitoring image and video data can be collected in real time, providing key basis for the management and decision-making of the mine. However, due to the characteristics of the mine itself, the environment of different areas in the mine differs greatly, and in addition, the progress and operation types of different areas are different, further leading to the changeable environment of each area in the mine.

[0003] To this end, the traditional image processing algorithm uses fixed parameter settings when processing images. However, in the complex and variable mine environment, the image noise characteristics at different times and different locations differ greatly. The fixed parameter denoising algorithm cannot automatically adjust the parameters according to the environmental changes, resulting in poor effect of mine monitoring image processing.

[0004] Therefore, there is a need for a mine image processing method with good effect and high reliability. SUMMARY

[0005] The purpose of the present disclosure is to provide a remote mine image processing method and system to improve the effect of mine image processing by adjusting the image processing parameters according to the environmental parameters.

[0006] The first aspect of the embodiment of the present disclosure provides a remote mine image processing method, comprising:

[0007] inputting a target environmental parameter into a target model to determine a denoising parameter;

[0008] determining a target denoising algorithm based on the denoising parameter;

[0009] processing target data based on the target denoising algorithm to obtain mine data, and sending the mine data to a display device of a mine monitoring center; wherein the target environmental parameter is an environmental parameter of a target area corresponding to the target data, and the target data includes monitoring image data of the target area.

[0010] The second aspect of the embodiment of the present disclosure provides a remote mine image processing system, comprising:

[0011] a parameter determination module configured to input a target environmental parameter into a target model to determine a denoising parameter;

[0012] an algorithm determination module configured to determine a target denoising algorithm based on the denoising parameter;

[0013] The data processing module is configured to process the target data based on a target noise reduction algorithm to obtain mine area data, and send the mine area data to a display device of the mine area monitoring center; wherein the target environmental parameter is an environmental parameter of a target region corresponding to the target data, and the target data includes monitoring image data of the target region.

[0014] In a third aspect, an electronic device is provided, which includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the steps of the remote mine area image processing method described above are implemented.

[0015] In a fourth aspect, a computer readable storage medium is provided, which stores a computer program. When the computer program is executed by a processor, the steps of the remote mine area image processing method described above are implemented.

[0016] The remote mine area image processing method and system provided by the embodiments of the present disclosure have the following advantages:

[0017] The present disclosure introduces a target environmental parameter to determine a noise reduction parameter, which can be adaptively adjusted according to different regions and different time of mine area environment, thereby improving the accuracy and practicability of image processing. The present disclosure dynamically selects a noise reduction algorithm most suitable for the current environment, which can reduce image noise and improve the effect and reliability of remote mine area image processing. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present disclosure, and other drawings can be obtained by those skilled in the art without creative labor.

[0019] Figure 1 The flowchart of the remote mine area image processing method provided by an embodiment of the present disclosure is shown in the figure.

[0020] Figure 2 The structural block diagram of the remote mine area image processing system provided by an embodiment of the present disclosure is shown in the figure.

[0021] Figure 3 The schematic block diagram of the electronic device provided by an embodiment of the present disclosure is shown in the figure. DETAILED DESCRIPTION

[0022] In the following description, for purposes of explanation and not limitation, specific details are set forth such as particular architectures, technologies, techniques, etc. in order to provide a thorough understanding of the embodiments of the present disclosure. However, it will be apparent to those skilled in the art that the present disclosure can be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present disclosure with unnecessary detail.

[0023] For the purpose of the present disclosure, the technical solutions and advantages will be clearer, specific embodiments will be described below with reference to the accompanying drawings.

[0024] Reference is made to Figure 1 , Figure 1 The flowchart of the remote mine image processing method provided by an embodiment of the present disclosure, the method comprises:

[0025] S101: input the target environment parameter into the target model to determine the noise reduction parameter.

[0026] In this embodiment, the target environment parameter is the environment parameter of the target region corresponding to the target data, the target data is the monitoring image data of the target region, and the target region is a region in the mine. Due to the large difference in environment of each region in the mine, and secondly due to the difference in construction operation environment and operation type of different mine regions, the difference in environment of each region in each mine is further aggravated. For example, in some mine regions, due to the relative closedness of the underground operation space, the dust concentration is prone to rise sharply during mining, and the humidity will also be different due to the difference in underground water source and ventilation condition; the light intensity will also be different due to the difference in windowing condition of each mine region, the higher the dust concentration, the more data noise collected, and the poorer the data quality.

[0027] The target environment parameter can be a parameter such as dust concentration, humidity, or light intensity that will affect image acquisition and processing, the target data can be an image or video collected by monitoring, and the target region can be a coal storage area, a gangue accumulation area, a mine, a beneficiation area, a production area, etc.

[0028] The target model can be a machine learning model trained, and in this embodiment, the base model of the target model can be a support vector machine model. The target model is trained by a data set composed of a preset number of environment parameters and their corresponding standard noise reduction parameters. The preset number refers to the number that can meet the learning and verification of the support vector machine model, which can be set according to experience or determined according to the number of data in the training data set when solving similar problems. The standard noise reduction parameter can be understood as the relatively optimal or optimal noise reduction parameter under the influence of this environment parameter. The noise reduction parameter is a parameter in the noise reduction algorithm or noise reduction model, and the noise reduction algorithm can be a bilateral filtering algorithm.

[0029] It should be noted that in the present embodiment, the target environment parameter is the environment parameter of the target region in the process of collecting target data, and the target data can be collected through monitoring installed in each region of the mining area.

[0030] S102: determining a target noise reduction algorithm based on the noise reduction parameter.

[0031] In the present embodiment, the noise reduction parameter can be part or all of the parameters in the bilateral filtering algorithm. In the present embodiment, the noise reduction parameter is selected from the spatial domain kernel radius, the gray value domain standard deviation, and the spatial domain standard deviation in the bilateral filtering algorithm parameters. The target noise reduction algorithm is the bilateral filtering algorithm.

[0032] In the present scenario, the reasons for selecting the above three parameters are as follows:

[0033] The spatial domain kernel radius determines the range of pixel neighborhood considered during filtering. In mining area monitoring images, noise can be distributed in different size regions. By adjusting the spatial domain kernel radius, the ability of the filtering algorithm to suppress noise of different scales can be controlled. A larger radius can effectively remove large-area noise, but may blur image details. A smaller radius can better preserve image edges and detail information, and is suitable for cases where noise is more dispersed.

[0034] The gray value domain standard deviation reflects the degree of variation of pixel gray values in the image. The gray values of mining area monitoring images may vary due to factors such as lighting conditions and device characteristics. When the gray value domain standard deviation is large, the algorithm will pay more attention to processing pixels with large gray differences to balance the overall gray distribution of the image and reduce the impact of uneven lighting and other factors; when the standard deviation is small, the algorithm will tend to preserve the original gray details of the image to avoid excessive smoothing that can cause image information loss.

[0035] The spatial domain standard deviation describes the spatial distribution of pixels. In the mining environment, different regions of the image may have different spatial characteristics, such as equipment, terrain, etc. A larger spatial domain standard deviation means that the algorithm will pay more attention to the spatial relationship between pixels, and will have a better suppression effect on noise that is widely distributed in space, and can repair some irregular noise regions in the image to some extent; a smaller spatial domain standard deviation makes the algorithm focus more on local pixel processing, which helps to preserve fine structures and texture information in the image.

[0036] Therefore, in the present embodiment, the spatial domain kernel radius, the gray value domain standard deviation, and the spatial domain standard deviation in the bilateral filtering algorithm parameters are selected as the noise reduction parameter. After obtaining the noise reduction parameter, it is brought into the bilateral filtering algorithm, and the remaining parameters can be set according to the reference values of the bilateral filtering algorithm, and finally the target noise reduction algorithm is obtained.

[0037] S103: processing the target data based on the target denoising algorithm to obtain mine area data, and sending the mine area data to a display device of the mine area monitoring center; wherein the target environmental parameter is an environmental parameter of a target region corresponding to the target data, and the target data is monitoring image data of the target region.

[0038] In the present embodiment, it can be known from the foregoing description that the denoising parameter of the target denoising algorithm is determined based on the target environmental parameter, and the target environmental parameter is an environmental parameter of a target region corresponding to the target data, so the target denoising algorithm is a processing algorithm suitable for the target data. The target data is monitoring image data of the target region, which can be a monitoring image or a monitoring video, for example. The mine area data is the target data after denoising processing, and therefore, the mine area data can also be an image or a video. After obtaining the mine area data, the mine area data is sent to the display device of the mine area monitoring center, so that relevant personnel can understand the real-time image data of each region in the mine area.

[0039] From the above, it can be concluded that the present disclosure can adaptively adjust to different regions and different time points of the mine area environment by introducing the target environmental parameter to determine the denoising parameter, thereby improving the accuracy and practicability of image processing. The present disclosure can reduce image noise and improve the effect and reliability of remote mine area image processing by dynamically selecting the denoising algorithm most suitable for the current environment.

[0040] As can be known from S101, the target model is obtained by training a data set composed of a preset number of environmental parameters and corresponding standard denoising parameters, which can be specifically explained as follows: the target model is a support vector machine model.

[0041] Before inputting the target environmental parameter into the target model to determine the denoising parameter, the following steps are further included:

[0042] training the support vector machine model based on a first data set to obtain the target model; wherein the data in the first data set is historical environmental parameters and corresponding standard denoising parameters, and the standard denoising parameters are determined by optimizing the initial denoising parameters based on a particle swarm optimization algorithm.

[0043] In the present embodiment, the historical environmental parameters refer to environmental parameters in historical mine area regions, such as dust concentration, humidity, and light intensity. The standard denoising parameters are part or all of the parameters in the relatively optimal or optimal denoising algorithm corresponding to the environmental parameters. The plurality of historical environmental parameters and the corresponding standard denoising parameters form the first data set, and the number and quality of the data in the first data set should meet the training and verification of the support vector machine model.

[0044] Considering that the process of corresponding a standard denoising parameter to each environmental parameter combination is time-consuming and laborious, in the embodiment, the initial denoising parameter is optimized by the particle swarm optimization algorithm to obtain the standard denoising parameter. The essence of the initial denoising parameter is the spatial domain kernel radius, the gray value domain standard deviation and the spatial domain standard deviation in the bilateral filtering algorithm. The initial parameter value can be the default value of the bilateral filtering algorithm.

[0045] Specifically, the process of determining the standard denoising parameter based on the particle swarm optimization algorithm includes:

[0046] Determining the particle dimension of the particle swarm optimization algorithm based on the parameters of the denoising algorithm;

[0047] Determining the fitness function of the particle swarm optimization algorithm based on the environmental parameters;

[0048] Iteratively calculating the initial denoising parameter based on the particle dimension and the fitness function until the difference between the values of the continuous multiple fitness functions is less than the first threshold value or the iteration number reaches the first number, to obtain the standard denoising parameter.

[0049] In the embodiment, the particle dimension in the particle swarm optimization algorithm can be understood as the position coordinate dimension of the particle. The position of each particle corresponds to a solution. Therefore, the particle dimension should be consistent with the parameters of the denoising algorithm. The parameters of the denoising algorithm selected in the foregoing are the spatial domain kernel radius, the gray value domain standard deviation and the spatial domain standard deviation. Therefore, in the embodiment, the particle dimension in the particle swarm optimization algorithm is three-dimensional, which can be represented as [spatial domain kernel radius, gray value domain standard deviation, spatial domain standard deviation].

[0050] Considering the influence of environmental factors (dust concentration, humidity, and light intensity) on image quality, the influence of environmental factors should be added to the fitness function, for example, the fitness function is:

[0051] , wherein, represents the value of the fitness function, is the weight coefficient corresponding to the peak signal-to-noise ratio, is the weight coefficient corresponding to the structural similarity index, , is the peak signal-to-noise ratio, is the structural similarity index, is the weight coefficient corresponding to the th environmental parameter, is the normalized value of the th environmental parameter, for example, represents the normalized value corresponding to the dust concentration, represents the normalized value corresponding to the humidity, represents the normalized value corresponding to the light intensity. represents a weight corresponding to the dust concentration, represents a weight corresponding to the humidity, represents a weight corresponding to the illumination intensity, and The values of the weights can be determined empirically or based on the degree of influence of each environmental parameter on image quality.

[0052] Specifically, the scores of the environmental parameters can be calculated based on a random forest algorithm, and the scores of the environmental parameters are normalized to obtain the weights corresponding to the environmental parameters. Alternatively, the correlation can be determined by calculating the Pearson correlation coefficient, and normalized to obtain the weights corresponding to the environmental parameters. The values of the weights can be determined empirically or based on the degree of influence of each environmental parameter on image quality.

[0053] The logic of the fitness function is as follows: The basic image quality score is calculated. The PSNR is used to measure the distortion of the image, and the higher the value, the better the denoising or enhancement effect. The SSIM takes into account the brightness, contrast and structural information of the image, and the closer to 1, the higher the image quality and the more consistent with the human eye's perception of image quality.

[0054] The comprehensive influence of environmental factors on image quality is calculated. The actual value of the environmental factor is converted to the interval [0, 1] to facilitate unified measurement and comparison. is used to adjust the basic image quality score, i.e., if the environmental factor has a greater influence on the image quality, the value of will be close to 1, will be smaller, thereby reducing the basic image quality score. Conversely, if the environmental factor is ideal and has a smaller influence on the image quality, is close to 1, and the basic image quality score is less affected. Thus, the pros and cons of the image preprocessing algorithm parameters can be reasonably evaluated according to different environmental conditions, and the particle swarm optimization algorithm can find the most suitable parameter combination under the current environment, making the fitness function more consistent with the needs of the application scenario in the mining area.

[0055] The iteration termination condition of the particle swarm optimization algorithm can be that the difference between consecutive multiple fitness function values is less than a first threshold value or the number of iterations reaches a first number of times. The first number of times and the specific number of consecutive times can be determined empirically, and generally the first number of times can be set to 150-250 times and the number of times can be set to 3-6 times. When the iteration calculation is completed, the denoising parameter corresponding to the optimal position of the population is the standard denoising parameter.

[0056] ​​In the iterative optimization process of the particle swarm optimization algorithm, its inertia weight and learning factor are parameters that control the iterative calculation direction of each particle. In this embodiment, the inertia weight of the particle swarm optimization algorithm can be determined based on the number of iterations. That is, the inertia weight of the particle swarm optimization algorithm is determined based on the number of iterations; the iterative calculation direction of the particle swarm optimization algorithm is controlled based on the inertia weight and the learning factor.

[0057] Specifically, the inertia weights of the particle swarm optimization algorithm are determined based on the first formula and the number of iterations. The first formula can be: ,in, For the first Inertia weights in the next iteration. This is the maximum value of the inertia weight, which can generally be set to 0.9, giving the particles greater global search capability in the early stages of iteration. This is the minimum value of the inertia weight, usually set to 0.4, which makes the particles focus on local search in the later stages of iteration. For the first count, This represents the current iteration number. Represents the natural constant.

[0058] The logic of the first formula is: A cosine function is used to control the basic trend of change in inertia weights at the start of the iteration. , At this point, the inertial weight is close to The particles possess strong global search capabilities. As the number of iterations increases, the value of the cosine function gradually decreases, and the inertial weight also decreases accordingly. When the iteration reaches halfway point... , The inertia weight is at an intermediate value. In the later stages of iteration, the inertia weight continues to decrease, approaching... Particles tend to search locally.

[0059] It is an exponential decay function used to fine-tune the basic trend described above. In the early stages of iteration, the value of the exponential function is close to 0, which makes the decrease rate of the inertia weight relatively slow, ensuring that the particles have enough time to perform a global search. As the number of iterations increases, the value of the exponential function gradually increases, and the decrease rate of the inertia weight accelerates, thus enabling a faster shift to local search in the later stages of iteration.

[0060] The inertia weight can be determined based on the number of iterations using the first formula. The learning factor includes the group learning factor and the individual learning factor. The group learning factor and the individual learning factor can be set according to the reference values ​​of the particle swarm optimization algorithm, for example, both can be set to 2.

[0061] From the above, the present disclosure can determine the noise reduction parameters suitable for different mine environment by introducing the support vector machine model and training based on the historical environment parameters and the corresponding standard noise reduction parameters. The standard noise reduction parameters are obtained by optimizing the particle swarm optimization algorithm, which ensures the accuracy of the noise reduction algorithm in different environments and improves the efficiency and accuracy of remote mine image processing. The present disclosure further optimizes the search performance of the particle swarm optimization algorithm by adjusting the inertia weight, making the optimization process more reliable, and thus obtaining more accurate standard noise reduction parameters, thereby improving the training effect of the support vector machine model and improving the effect and reliability of mine image processing.

[0062] The purpose of the aforementioned particle swarm optimization algorithm is to determine the standard noise reduction parameters corresponding to the historical environment parameters, i.e., the optimal or optimal noise reduction parameters corresponding to each environment parameter combination. Training the support vector machine model based on the data set (first data set) composed of historical environment parameters and corresponding standard noise reduction parameters can obtain the target model. The present embodiment describes the process of training the support vector machine model:

[0063] In an embodiment of the present disclosure, the kernel function of the support vector machine model is a Gaussian kernel function;

[0064] Training the support vector machine model based on the first data set includes:

[0065] In response to the concentration degree value of the standard noise reduction parameters being less than the first concentration degree value, the reference value of the penalty parameter is reduced based on the first penalty step, and the target penalty parameter of the Gaussian kernel function is obtained;

[0066] In response to the concentration degree value of the standard noise reduction parameters being greater than the second concentration degree value, the reference value of the penalty parameter is increased based on the second penalty step, and the target penalty parameter of the Gaussian kernel function is obtained;

[0067] Training the support vector machine model based on the Gaussian kernel function with the determined target penalty parameter and the first data set;

[0068] Wherein, the first concentration degree value is less than the second concentration degree value.

[0069] In the present embodiment, the support vector machine model uses a Gaussian kernel function as the kernel function, which is used to map the input data to a high-dimensional space for better handling of nonlinear problems.

[0070] The concentration degree value can be determined by calculating the variance or information entropy of the data. Taking variance as an example, variance can measure the dispersion degree of data. The smaller the value, the more concentrated the data. Therefore, the size of variance and concentration degree value should be related but opposite, so the concentration degree value can be obtained through a simple mapping relationship or linear formula, which will not be described here.

[0071] When the concentration degree value of the standard denoising parameter is less than the first concentration degree value, it means that the distribution of the standard denoising parameter in the data space is relatively dispersed. In this case, in order not to make the model overemphasize the accurate classification of each sample (because there are some outliers or noise points in the data dispersion), the reference value of the penalty parameter can be reduced based on the first penalty step to obtain the target penalty parameter of the Gaussian kernel function. A smaller penalty parameter will make the model have greater tolerance for errors in the training data and pay more attention to the generalization ability, avoiding overfitting.

[0072] On the contrary, when the concentration degree value of the standard denoising parameter is greater than the second concentration degree value, it means that the distribution of the standard denoising parameter in the data space is relatively concentrated. At this time, the reference value of the penalty parameter is increased based on the second penalty step to obtain the target penalty parameter. A larger penalty parameter will make the model more strictly require accurate classification of each sample, try to reduce the training error, and improve the accuracy of model training. The first penalty step and the second penalty step can be determined based on experience or based on the concentration degree value. Specifically, the first penalty step and the second penalty step can be determined based on the second formula, the second formula: wherein, represents the first penalty step or the second penalty step, when , represents the first penalty step, when , represents the second penalty step, represents the concentration degree value, is the first concentration degree value, is the second concentration degree value, is a preset minimum step, is a preset maximum step, and the first concentration degree value and the second concentration degree value can be preset according to experience.

[0073] When is less than , the second formula tends to produce a smaller step, close to , when is greater than , the second formula tends to produce a larger step, close to . The first penalty step or the second penalty step can be obtained through the above description, and if the concentration degree value of the standard denoising parameter is between the first concentration degree value and the second concentration degree value, the reference value of the penalty parameter can not be adjusted, that is, the reference value of the penalty parameter is taken as the penalty parameter of the Gaussian kernel function in response to the concentration degree value of the standard denoising parameter being greater than or equal to the second concentration degree value and the concentration degree value of the standard denoising parameter being less than or equal to the first concentration degree value. The reference value of the penalty parameter can be the default value of the support vector machine model.

[0074] From the above, the present disclosure can evaluate the distribution state of the standard denoising parameter by introducing the concentration degree value, and dynamically adjust the penalty parameter of the Gaussian kernel function in the support vector machine model according to the distribution state. Make the support vector machine model better adapt to the characteristics of different data sets, improve the performance and stability of the model, and then improve the effect and reliability of the mine image processing.

[0075] In an embodiment of the present disclosure, the kernel function of the support vector machine model is a Gaussian kernel function;

[0076] Training the support vector machine model based on the first data set, comprising:

[0077] Identifying the first scatter plot based on the first neural network model;

[0078] In response to the distribution of the standard denoising parameter in the data space satisfying the first condition, reducing the reference value of the kernel coefficient based on the first kernel coefficient step, to obtain the target kernel coefficient of the Gaussian kernel function;

[0079] In response to the distribution of the standard denoising parameter in the data space satisfying the second condition, increasing the reference value of the kernel coefficient based on the second kernel coefficient step, to obtain the target kernel coefficient of the Gaussian kernel function;

[0080] Training the support vector machine model based on the Gaussian kernel function with the determined target kernel coefficient and the first data set;

[0081] Wherein, the first neural network model is determined based on a certain number of scatter plots and their corresponding distribution results, the first scatter plot is a scatter plot of the distribution of the standard denoising parameter in the data space, and the first condition and the second condition are different.

[0082] In this embodiment, in addition to the aforementioned penalty parameter, the support vector machine model also has a kernel function, and the kernel coefficient determines the distribution of data in high-dimensional space. If the data distribution of the denoising algorithm parameter presents a complex nonlinear structure in the low-dimensional space, a larger kernel coefficient may be needed to map the data to a higher-dimensional space, so that the support vector machine model can better classify or regress; if the data distribution is relatively simple and the linear separability is good, a smaller kernel coefficient may make the support vector machine model achieve better results.

[0083] Therefore, in this embodiment, the first neural network model can be used to identify and process the scatter plot of the distribution of the standard denoising parameter in the data space, to determine whether the distribution of the standard denoising parameter in the data space satisfies the first condition, and whether the distribution of the standard denoising parameter in the data space satisfies the second condition; the first condition is to appear complex nonlinear characteristics, and the second condition is to appear simple linear or simple nonlinear characteristics.

[0084] The first neural network model is determined based on a certain number of scatter plots and corresponding distribution results, the certain number being a number of data sufficient for training of the first neural network model, which can be a preset value. The first kernel coefficient step size and the second kernel coefficient step size can be determined according to experience or in an experimental process. The reference value of the kernel coefficient is the default value of the support vector machine model.

[0085] If the first scatter plot does not satisfy the first condition and the second condition, the reference value of the kernel coefficient can not be adjusted, that is, the reference value of the kernel coefficient is taken as the target kernel coefficient of the Gaussian kernel function in response to the distribution of the standard denoising parameter in the data space not satisfying the first condition and the second condition.

[0086] From the above, it can be concluded that the present disclosure improves the classification and regression performance of the support vector machine model by introducing the first neural network model to identify the scatter plot of the distribution of the standard denoising parameter in the data space and dynamically adjusting the kernel coefficient of the Gaussian kernel function in the support vector machine model, thereby improving the effect and reliability of the mine image processing.

[0087] In an embodiment of the present disclosure, the monitoring image data includes a monitoring image and a monitoring video.

[0088] The remote mine image processing method further includes:

[0089] In response to the target data being a monitoring image of the target region, the environmental parameter of the target region when the target data is collected is taken as the target environmental parameter;

[0090] In response to the target data being a monitoring video of the target region and the change rate of the environmental parameter of the target region in the target period being less than a first threshold, the environmental parameter at any time in the target period is taken as the target environmental parameter;

[0091] In response to the target data being a monitoring video of the target region and the change rate of the environmental parameter of the target region in the target period being greater than or equal to the first threshold, the environmental parameter at each time in the target period is taken as the target environmental parameter;

[0092] The starting time of the target period is the starting time of collecting the target data, and the ending time of the target period is the ending time of collecting the target data.

[0093] In this embodiment, the change rate of the environmental parameter can be determined by calculating the ratio of the change amount of each environmental parameter in the target period to the length of the target period. The change rate of each environmental parameter can be calculated and weighted, and the weight can be evenly distributed or set according to preference.

[0094] In practical application scenarios in the mining area, the collected target data exists in two forms, namely a single monitoring image and a continuous monitoring video. Due to the importance of safety in the mining area (for example, clearly monitoring the running status of equipment in the mining area, personnel activities, etc. to prevent accidents) and the high requirement for data visibility (to ensure clear images or videos for staff to accurately determine the situation), different processing methods need to be set according to different situations of the target data to determine the target environment parameters, so as to better perform image processing.

[0095] When the target data is a monitoring image of the target area, the processing method is relatively simple. The environment parameters of the target area when the monitoring image is collected, such as the dust concentration, humidity, and light intensity at that time, are directly used as the target environment parameters. Because a single image only corresponds to a specific collection time, the environment parameters at that time can reflect the environmental conditions affecting the image quality at that time, and the parameters required for subsequent image processing are determined based on this.

[0096] If the target data is a monitoring video of the target area, and the change rate of the environment parameters of the target area in the target period (i.e. from the start time to the end time of collecting the monitoring video) is less than a pre-set first threshold. That is, during the video collection period, the environment is relatively stable and does not change much. Therefore, at this time, the environment parameters at any time within the target period can be used as the target environment parameters. This is because in a stable environment, the environment parameters at any time can represent the environmental conditions of the entire period, and selecting the parameters at any time can meet the requirements of image processing, simplify the processing process, and reduce the amount of calculation.

[0097] When the target data is a monitoring video, and the change rate of the environment parameters of the target area in the target period is greater than or equal to the first threshold, it indicates that the environment changes significantly during the video collection process. In this case, in order to more accurately process each frame of image in the video (because the environmental changes at different times will have different effects on the image quality), the environment parameters at each time within the target period need to be used as the target environment parameters. This can perform more accurate image processing (such as determining appropriate noise reduction parameters) according to the specific environmental conditions of each frame of image, to ensure the overall quality and visibility of the video, and meet the high requirements of the mining area for safety and data visibility. The target period can be pre-set according to the actual needs of the mining area, and the first threshold can be determined in the experimental process or set according to experience.

[0098] From the above, it can be seen that the present disclosure introduces the change rate of the environment parameters and the determination method of the target environment parameters, enhances the adaptability and accuracy of image processing, optimizes the processing process, improves the calculation efficiency, meets the high requirements of the mining area for safety and data visibility, and improves the effect of remote mining area image processing.

[0099] The remote mine image processing method corresponding to the above embodiment, Figure 2 A structural block diagram of a remote mine image processing system is provided for an embodiment of the present disclosure. For ease of illustration, only parts related to the embodiments of the present disclosure are shown. For reference Figure 2 The remote mine image processing system 20 includes a parameter determination module 21, an algorithm determination module 22, and a data processing module 23.

[0100] The parameter determination module 21 is configured to input a target environment parameter into a target model to determine a noise reduction parameter.

[0101] The algorithm determination module 22 is configured to determine a target noise reduction algorithm based on the noise reduction parameter.

[0102] The data processing module 23 is configured to process target data based on the target noise reduction algorithm to obtain mine data, and send the mine data to a display device of a mine monitoring center; wherein the target environment parameter is an environment parameter of a target region corresponding to the target data, and the target data includes monitoring image data of the target region.

[0103] In an embodiment of the present disclosure, the target model is a support vector machine model.

[0104] The remote mine image processing system 20 further includes a model training module configured to train the support vector machine model based on a first data set to obtain the target model before inputting the target environment parameter into the target model to determine the noise reduction parameter; wherein the data in the first data set is historical environment parameters and corresponding standard noise reduction parameters, and the standard noise reduction parameters are determined by optimizing the initial noise reduction parameters based on a particle swarm optimization algorithm.

[0105] In an embodiment of the present disclosure, the remote mine image processing system 20 further includes a standard noise reduction parameter determination module configured to determine the particle dimension of the particle swarm optimization algorithm based on the parameters of the noise reduction algorithm.

[0106] The fitness function of the particle swarm optimization algorithm is determined based on the environment parameters.

[0107] The initial noise reduction parameters are iteratively calculated based on the particle dimension and the fitness function until the difference between consecutive multiple fitness function values is less than a first threshold value or the number of iterations reaches a first number, to obtain the standard noise reduction parameters.

[0108] In an embodiment of the present disclosure, the remote mine image processing system 20 further includes an iterative calculation control module configured to determine the inertia weight of the particle swarm optimization algorithm based on the number of iterations.

[0109] The iterative calculation direction of the particle swarm optimization algorithm is controlled based on the inertia weight and a learning factor.

[0110] In an embodiment of the present disclosure, the kernel function of the support vector machine model is a Gaussian kernel function.

[0111] The model training module is specifically configured to, in response to the concentration degree value of the standard noise reduction parameter being less than a first concentration degree value, decrease the reference value of the penalty parameter based on a first penalty step, to obtain a target penalty parameter of the Gaussian kernel function.

[0112] In response to the concentration degree value of the standard noise reduction parameter being greater than a second concentration degree value, increase the reference value of the penalty parameter based on a second penalty step, to obtain a target penalty parameter of the Gaussian kernel function.

[0113] Train the support vector machine model based on the Gaussian kernel function with the determined target penalty parameter and the first data set.

[0114] The first concentration degree value is less than the second concentration degree value.

[0115] In an embodiment of the present disclosure, the kernel function of the support vector machine model is a Gaussian kernel function.

[0116] The model training module is specifically configured to further identify the first scatter plot based on a first neural network model.

[0117] In response to the distribution of the standard noise reduction parameter in the data space satisfying a first condition, decrease the reference value of the kernel coefficient based on a first kernel coefficient step, to obtain a target kernel coefficient of the Gaussian kernel function.

[0118] In response to the distribution of the standard noise reduction parameter in the data space satisfying a second condition, increase the reference value of the kernel coefficient based on a second kernel coefficient step, to obtain a target kernel coefficient of the Gaussian kernel function.

[0119] Train the support vector machine model based on the Gaussian kernel function with the determined target kernel coefficient and the first data set.

[0120] The first neural network model is determined based on a certain number of scatter plots and corresponding distribution results, the first scatter plot is a scatter plot of the distribution of the standard noise reduction parameter in the data space, and the first condition and the second condition are different.

[0121] In an embodiment of the present disclosure, the monitoring image data includes a monitoring image and a monitoring video.

[0122] The remote mine area image processing system 20 further includes a target environment parameter determination module configured to, in response to the target data being a monitoring image of a target region, take an environment parameter of the target region when the target data is collected as a target environment parameter.

[0123] In response to the fact that the target data is the monitoring video of the target area, and the rate of change of the environmental parameters of the target area within the target time period is less than the first threshold, the environmental parameters at any time within the target time period are used as the target environmental parameters.

[0124] In response to the fact that the target data is the monitoring video of the target area, and the rate of change of the environmental parameters of the target area within the target time period is greater than or equal to the first threshold, the environmental parameters at each moment within the target time period are used as the target environmental parameters.

[0125] The start time of the target time period is the start time of collecting target data, and the end time of the target time period is the end time of collecting target data.

[0126] See Figure 3 , Figure 3 This is a schematic block diagram of an electronic device provided according to an embodiment of the present disclosure. Figure 3 The electronic device 300 in this embodiment may include one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The processors 301, input devices 302, output devices 303, and memories 304 communicate with each other via a communication bus 305. The memories 304 store computer programs, including program instructions. The processors 301 execute the program instructions stored in the memories 304. Specifically, the processors 301 are configured to invoke the program instructions to perform the functions of each module / unit in the above-described device embodiments, for example... Figure 2 The functions of the parameter determination module 21, algorithm determination module 22, and data processing module 23 are shown.

[0127] It should be understood that, in the embodiments of this disclosure, the processor 301 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0128] Input device 302 may include a touchpad, a fingerprint sensor (for collecting the user's fingerprint information and fingerprint orientation information), a microphone, etc., and output device 303 may include a display (LCD, etc.), a speaker, etc.

[0129] The memory 304 can include read-only memory and random access memory, and provide instructions and data to the processor 301. A portion of the memory 304 can also include non-volatile random access memory. For example, the memory 304 can also store device type information.

[0130] In specific implementations, the processor 301, the input device 302, and the output device 303 described in the embodiments of the present disclosure can execute the implementation manners described in the first and second embodiments of the remote mine image processing method provided by the embodiments of the present disclosure, and can also execute the implementation manners of the electronic device described in the embodiments of the present disclosure, which will not be described here.

[0131] In another embodiment of the present disclosure, a computer readable storage medium is provided, and the computer readable storage medium stores a computer program. The computer program includes program instructions, and the program instructions are executed by a processor to implement all or part of the processes of the above-mentioned embodiments. The computer program can also be used to instruct related hardware to complete, and the computer program can be stored in a computer readable storage medium. When the computer program is executed by the processor, the steps of each method embodiment described above can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or some intermediate forms. The computer readable medium can include any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0132] The computer readable storage medium can be an internal storage unit of the electronic device of any of the preceding embodiments, such as a hard disk or a memory of the electronic device. The computer readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the computer readable storage medium can include both the internal storage unit and the external storage device of the electronic device. The computer readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer readable storage medium can also be used to temporarily store data that has been output or will be output.

[0133] Those skilled in the art can clearly understand that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized in electronic hardware, computer software or a combination of both. In order to clearly illustrate the interchangeability of hardware and software, the components and steps of each example have been described in the foregoing description in a general manner. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present disclosure.

[0134] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the electronic device and the units described above can refer to the corresponding processes in the foregoing method embodiments, which will not be repeated here.

[0135] In several embodiments provided in the present application, it should be understood that the disclosed electronic device and method can be implemented in other ways. For example, the device embodiments described above are only schematic, and the division of units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces or units, and can also be electrical, mechanical or other forms of connection.

[0136] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, i.e. they can be located in one place or distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiments of the present disclosure.

[0137] In addition, each functional unit in each embodiment of the present disclosure can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or software functional unit.

[0138] The above is only a specific implementation of the present disclosure, but the protection scope of the present disclosure is not limited thereto. Any skilled person in the art can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present disclosure, and these modifications or replacements should be covered in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be subject to the protection scope of the claims.

Claims

1. A method for processing images of remote mining areas, characterized in that, include: Input the target environment parameters into the target model to determine the noise reduction parameters; The target noise reduction algorithm is determined based on the aforementioned noise reduction parameters; The target data is processed based on the target noise reduction algorithm to obtain mining area data, and the mining area data is sent to the display device of the mining area monitoring center; wherein, the target environmental parameters are the environmental parameters of the target area corresponding to the target data, and the target data is the monitoring image data of the target area; The target model is a support vector machine model; Before inputting the target environment parameters into the target model and determining the noise reduction parameters, the method further includes: The support vector machine model is trained based on the first dataset to obtain the target model; wherein, the data in the first dataset are historical environmental parameters and corresponding standard denoising parameters, and the standard denoising parameters are determined by optimizing the initial denoising parameters based on the particle swarm optimization algorithm; The process of determining the standard noise reduction parameters based on the particle swarm optimization algorithm includes: The particle dimension of the particle swarm optimization algorithm is determined based on the parameters of the noise reduction algorithm. The fitness function of the particle swarm optimization algorithm is determined based on environmental parameters; The initial denoising parameters are iteratively calculated based on the particle dimension and the fitness function until the difference between multiple consecutive fitness function values ​​is less than the first threshold or the number of iterations reaches the first number, thus obtaining the standard denoising parameters.

2. The remote mining area image processing method as described in claim 1, characterized in that, Also includes: The inertia weights of the particle swarm optimization algorithm are determined based on the number of iterations. The inertia weight and learning factor control the iterative calculation direction of the particle swarm optimization algorithm.

3. The remote mining area image processing method as described in claim 1, characterized in that, The kernel function of the support vector machine model is a Gaussian kernel function; The training of the support vector machine model based on the first dataset includes: In response to the fact that the concentration value of the standard noise reduction parameter is less than the first concentration value, the reference value of the penalty parameter is reduced based on the first penalty step size to obtain the target penalty parameter of the Gaussian kernel function; In response to the concentration value of the standard noise reduction parameter being greater than the second concentration value, the reference value of the penalty parameter is increased based on the second penalty step size to obtain the target penalty parameter of the Gaussian kernel function; The support vector machine model is trained based on the Gaussian kernel function with determined target penalty parameters and the first dataset; The first concentration level value is less than the second concentration level value.

4. The remote mining area image processing method as described in claim 1, characterized in that, The kernel function of the support vector machine model is a Gaussian kernel function; The training of the support vector machine model based on the first dataset includes: The first scatter plot is identified based on the first neural network model; In response to the distribution of the standard noise reduction parameters in the data space satisfying the first condition, the reference value of the kernel coefficient is reduced based on the first kernel coefficient step size to obtain the target kernel coefficient of the Gaussian kernel function; In response to the distribution of the standard noise reduction parameters in the data space satisfying the second condition, the reference value of the kernel coefficient is increased based on the second kernel coefficient step size to obtain the target kernel coefficient of the Gaussian kernel function; The support vector machine model is trained based on the Gaussian kernel function with determined target kernel coefficients and the first dataset; The first neural network model is determined based on a certain number of scatter plots and their corresponding distribution results. The first scatter plot is a scatter plot of the distribution of the standard noise reduction parameters in the data space. The first condition and the second condition are different.

5. The remote mining area image processing method as described in claim 1, characterized in that, The surveillance image data includes: surveillance images and surveillance videos; Remote mining area image processing methods also include: In response to the target data being a monitoring image of the target area, the environmental parameters of the target area at the time the target data was collected are taken as the target environmental parameters; In response to the fact that the target data is a surveillance video of the target area, and the rate of change of the environmental parameters of the target area within the target time period is less than a first threshold, the environmental parameters at any time within the target time period are used as the target environmental parameters. In response to the fact that the target data is a surveillance video of the target area, and the rate of change of the environmental parameters of the target area within the target time period is greater than or equal to a first threshold, the environmental parameters at each moment within the target time period are used as the target environmental parameters. Wherein, the start time of the target time period is the start time of collecting the target data, and the end time of the target time period is the end time of collecting the target data.

6. A remote mining area image processing system, characterized in that, include: The parameter determination module is used to input the target environment parameters into the target model and determine the noise reduction parameters; The algorithm determination module is used to determine the target noise reduction algorithm based on the noise reduction parameters; The data processing module is used to process the target data based on the target noise reduction algorithm to obtain mining area data, and send the mining area data to the display device of the mining area monitoring center; wherein, the target environmental parameters are the environmental parameters of the target area corresponding to the target data, and the target data includes the monitoring image data of the target area; The target model is a support vector machine model; The model training module is used to train the support vector machine model based on the first dataset to obtain the target model before inputting the target environment parameters into the target model and determining the denoising parameters; wherein, the data in the first dataset are historical environment parameters and corresponding standard denoising parameters, and the standard denoising parameters are determined by optimizing the initial denoising parameters based on the particle swarm optimization algorithm; The standard noise reduction parameter determination module is used to determine the particle dimension of the particle swarm optimization algorithm based on the parameters of the noise reduction algorithm. The fitness function of the particle swarm optimization algorithm is determined based on environmental parameters; The initial denoising parameters are iteratively calculated based on the particle dimension and the fitness function until the difference between multiple consecutive fitness function values ​​is less than the first threshold or the number of iterations reaches the first number, thus obtaining the standard denoising parameters.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 5.

Images