Image processing-based dust detection methods, devices, and related equipment

Abstract

This invention provides a dust detection method and apparatus based on image processing, comprising: acquiring dust data within a current area, the dust data including dust image data and dust concentration data; processing the dust image data and dust concentration data using a trained dust feature analysis model to obtain dust feature data within the current area; and integrating and outputting the dust feature data through a visualization module to generate a visualized dust feature report. This method achieves a solution for combining image data and concentration data to obtain accurate dust feature data. By comprehensively analyzing the current dust concentration and image within a certain area, the real-time performance, accuracy, and stability of dust detection within that area are improved.

Description

Technical Field

[0001] This invention relates to the field of environmental monitoring, and in particular to a dust detection method, apparatus, electronic device, and storage medium based on image processing. Background Technology

[0002] In the fields of environmental monitoring and air quality assessment, dust detection and analysis are crucial. Dust not only affects air quality but may also have a significant impact on human health and the industrial production environment. Therefore, accurately monitoring and assessing the dust content in the air has become an important task for environmental protection and health supervision.

[0003] Current dust detection methods cannot combine image data with corresponding dust concentration data to determine the current state of dust, resulting in low accuracy and stability in dust detection. Summary of the Invention

[0004] This invention provides an image processing-based dust detection method to address the problem that existing dust detection methods cannot combine image data and corresponding dust concentration data to determine the current state of dust, resulting in low detection accuracy and instability.

[0005] In a first aspect, embodiments of the present invention provide a dust detection method based on image processing, the method comprising the following steps:

[0006] Obtain dust data within the current area, including dust image data and dust concentration data;

[0007] By using a trained dust feature analysis model, the dust image data and dust concentration data are processed to obtain dust feature data for the current area.

[0008] The dust feature data is integrated and output through the visualization module to generate a visualized dust feature report.

[0009] Optionally, acquiring dust data in the current area includes:

[0010] The dust concentration data in the current area is collected by a preset concentration sensor within a certain period of time.

[0011] The dust in the current area is captured by a preset optical camera within a certain period of time to obtain corresponding dust image data.

[0012] Optionally, the step of capturing images of dust in the current area using a preset optical camera within a certain time period to obtain corresponding dust image data includes:

[0013] The first image data obtained by the capture is subjected to image enhancement preprocessing to obtain the second image data;

[0014] The second image data is preprocessed with feature labeling to obtain the corresponding dust image data, which contains the corresponding dust features.

[0015] Optionally, the step of processing the dust image data and dust concentration data using a trained dust feature analysis model to obtain dust feature data in the current area includes:

[0016] The similarity between the dust features corresponding to the dust image data and the dust concentration data is calculated to determine the overlapping part of the dust image data and the dust concentration data;

[0017] Based on the overlapping portion, dust characteristic data in the current area is determined, including the current concentration, concentration trend, and corresponding early warning scheme.

[0018] Optionally, before processing the dust image data and dust concentration data using the trained dust feature analysis model to obtain the dust feature data of the current area, the method further includes:

[0019] Obtain the dust feature analysis model to be trained and the training dust sample set, wherein the training dust sample set includes dust concentration, dust image corresponding to the dust concentration, and dust concentration label corresponding to the dust image;

[0020] Based on dust concentration, dust images corresponding to the dust concentration, and dust concentration labels corresponding to the dust images, the dust feature analysis model to be trained is iteratively trained, and a trained dust feature analysis model is obtained after the iterative training is completed.

[0021] Optionally, the method further includes iteratively training the dust feature analysis model to be trained based on dust concentration, dust images corresponding to the dust concentration, and dust concentration labels corresponding to the dust images, and obtaining a trained dust feature analysis model after the iterative training is completed.

[0022] The dust image corresponding to the latest dust concentration is used as the latest training sample to replace the current training set;

[0023] The dust feature analysis model to be trained is iteratively trained using the latest training samples to obtain a trained dust feature analysis model.

[0024] Optionally, the step of integrating and outputting the dust feature data through the visualization module to generate a visualized dust feature report includes:

[0025] Based on the dust feature data, curves of each feature data are determined in the dust feature data;

[0026] Based on the aforementioned curve, trend charts for each feature data are obtained;

[0027] Based on the trend chart and curve, a corresponding early warning strategy is generated, and a visual report is produced for display.

[0028] Secondly, embodiments of the present invention also provide an image processing-based dust detection device, the image processing-based dust detection device comprising:

[0029] The acquisition module is used to acquire dust data in the current area, including dust image data and dust concentration data;

[0030] The processing module is used to process the dust image data and dust concentration data through a trained dust feature analysis model to obtain dust feature data in the current area.

[0031] The generation module is used to integrate and output the dust feature data through the visualization module to generate a visualized dust feature report.

[0032] Thirdly, embodiments of the present invention provide an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps in the dust detection method based on image processing provided in the embodiments of the present invention.

[0033] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the dust detection method based on image processing provided in the embodiments of the invention.

[0034] In this embodiment of the invention, dust data within the current area is acquired, including dust image data and dust concentration data. A trained dust feature analysis model is used to process the dust image data and dust concentration data to obtain dust feature data within the current area. A visualization module integrates and outputs the dust feature data to generate a visualized dust feature report. This method achieves a solution that combines image data and concentration data to obtain accurate dust feature data. By comprehensively analyzing the current dust concentration and image within a certain area, the real-time performance, accuracy, and stability of dust detection within that area are improved. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 This is a flowchart of a dust detection method based on image processing provided in an embodiment of the present invention;

[0037] Figure 2 This is a schematic diagram of the structure of a dust detection method and apparatus based on image processing provided in an embodiment of the present invention;

[0038] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0039] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0040] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0041] like Figure 1 As shown, Figure 1 This is a flowchart of a dust detection method based on image processing provided by an embodiment of the present invention. The dust detection method based on image processing includes the following steps:

[0042] S101. Obtain dust data for the current area.

[0043] In this embodiment of the invention, the above-mentioned dust detection method based on image processing can be applied to a dust detection device based on image processing. The dust detection device based on image processing has functions such as dust data processing, dust data transmission and reception, and dust data storage. It can be built based on a server or server cluster. The server or server cluster can be an electronic device with dust data processing capabilities.

[0044] The aforementioned dust data may include, but is not limited to, dust image data and dust concentration data. Specifically, the aforementioned dust image data can be obtained by real-time imaging of the dust portion within the current area using a high-resolution camera. Generally, the aforementioned area can be determined based on the camera's maximum dust capture capability. This maximum dust capture capability can be determined based on the clarity of the dust image captured within the largest area; that is, the clearer the captured dust image is over a larger area, the stronger the maximum dust capture capability.

[0045] It should be noted that the aforementioned high-resolution camera is also equipped with a light source system. This system uses a light source with wavelengths and intensities that match the ambient light in the current environment, enabling the camera to clearly capture images of dust particles in the air under various ambient lighting conditions.

[0046] The dust concentration data mentioned above can be a specific value obtained by acquiring and calculating the number of dust particles in a certain area through a preset concentration sensor. This dust concentration data can indicate the degree of dust particle distribution in the current area.

[0047] S102. Using the trained dust feature analysis model, process the dust image data and dust concentration data to obtain the dust feature data of the current area.

[0048] In this embodiment of the invention, the trained dust feature analysis model can include, but is not limited to, any deep learning model that can combine dust image data and dust concentration data for comprehensive analysis and processing, and output dust feature data in the current area, such as BERT (Bidirectional Encoder Representations from Transformers), CLIP (Contrastive Language–Image Pre-training), and MMBT (Multimodal Bitransformer).

[0049] The aforementioned dust feature data can be obtained by processing dust image data and dust concentration data using the trained dust feature analysis model. This comprehensive information can be used to describe the detailed characteristics of dust, including but not limited to the size, shape, distribution of dust particles, and other features related to the current environmental concentration.

[0050] In one possible embodiment, the dust image data and dust concentration data are input into the trained dust feature analysis model. The model calculates and outputs accurate dust feature data by analyzing the size, shape, and distribution of dust particles in the image and combining the dust concentration data. It is understood that the output data may include, but is not limited to, dust concentration, distribution trend, and potential risk level. This data is integrated through a visualization module to generate an easy-to-understand dust feature report, providing a basis for environmental monitoring and health risk assessment.

[0051] S103. The dust feature data is integrated and output through the visualization module to generate a visualized dust feature report.

[0052] In this embodiment of the invention, the visualization module can display the output dust feature data on a smart terminal for users to view, and can generate a feature report that integrates the corresponding data.

[0053] In one possible embodiment, the aforementioned visualization module transforms the concentration, form, distribution, and trend data of dust into intuitive graphics and charts, and displays the dust status in the area in real time via a smart terminal. It is understood that corresponding early warnings and protective measures for the dust status can also be generated so that users can quickly understand the current air quality status and respond accordingly.

[0054] In this embodiment of the invention, dust data within the current area is acquired, including dust image data and dust concentration data. A trained dust feature analysis model is used to process the dust image data and dust concentration data to obtain dust feature data for the current area. A visualization module integrates and outputs the dust feature data to generate a visualized dust feature report. This method achieves a solution that combines image data and concentration data to obtain accurate dust feature data. By comprehensively analyzing the current dust concentration and image within a certain area, the real-time performance, accuracy, and stability of dust detection within that area are improved.

[0055] Optionally, the step of acquiring dust data in the current area may further include: collecting dust data in the current area within a certain period of time using a preset concentration sensor to obtain corresponding dust concentration data; and capturing images of dust in the current area within a certain period of time using a preset optical camera to obtain corresponding dust image data.

[0056] In this embodiment of the invention, the aforementioned preset concentration sensor may include, but is not limited to, any sensor capable of setting a region and collecting dust within that region. Specifically, after collecting dust in the current region, the preset concentration sensor can calculate the dust concentration data within the current region based on the set region and the number of dust particles collected.

[0057] The aforementioned preset optical camera can be a high-resolution camera device specifically designed for real-time image capture of airborne dust particles. The camera's position, shooting frequency, and image quality can be determined based on the maximum dust capture capability mentioned above, ensuring stable acquisition of clear dust image data under varying lighting and environmental conditions. Generally, the aforementioned preset optical camera is also equipped with suitable filters and supplementary lighting to reduce ambient light interference, improve shooting results, and make dust particles easier for subsequent image processing and analysis algorithms to identify and process.

[0058] Optionally, the step of taking pictures of dust in the current area within a certain period of time using a preset optical camera to obtain corresponding dust image data further includes performing image enhancement preprocessing on the first image data to obtain second image data; performing feature labeling preprocessing on the second image data to obtain corresponding dust image data, wherein the dust image data contains corresponding dust features.

[0059] In this embodiment of the invention, the above-mentioned image enhancement preprocessing may include, but is not limited to, denoising, contrast enhancement, edge detection, etc. Specifically, filtering algorithms (such as Gaussian filtering) can be used to remove noise from the image, histogram equalization can be used to enhance the contrast of the image, and the Canny edge detection algorithm can be used to extract the edge information of dust particles in the image.

[0060] More specifically, the above-mentioned denoising process can use filters (such as Gaussian filters, median filters, etc.) to denoise the image, so as to reduce the impact of noise in the image on subsequent processing; the above-mentioned image enhancement can use histogram equalization, contrast enhancement and other techniques to enhance the contrast and clarity of the image, so as to better observe and analyze dust particles in the image; the above-mentioned edge detection can use edge detection algorithms (such as Sobel, Canny, etc.) to detect edge information in the image, thereby locating and extracting the contour and shape features of dust particles.

[0061] The aforementioned feature labeling preprocessing can extract features such as the shape, size, and distribution of dust particles through morphological processing and contour extraction techniques, which can then be used for subsequent dust concentration analysis and identification.

[0062] Dust particles in an image can also be separated from the background using segmentation algorithms (such as thresholding, region growing, etc.) for subsequent feature extraction and analysis.

[0063] In one possible embodiment, key features such as the shape, size, and distribution of dust particles are accurately identified and extracted from the acquired dust images through morphological processing and contour extraction techniques. Segmentation processing, such as threshold segmentation and region growing algorithms, is then used to further separate the dust particles in the image from the background, making the features of the dust particles more obvious and easier to analyze.

[0064] This processing method significantly improves the usability of the data, making the analysis and identification of dust concentration more accurate and effective.

[0065] Optionally, in the step of processing dust image data and dust concentration data through a trained dust feature analysis model to obtain dust feature data in the current area, the method further includes calculating the similarity between the dust features corresponding to the dust image data and the dust concentration data to determine the overlapping part of the dust image data and the dust concentration data; and determining the dust feature data in the current area based on the overlapping part.

[0066] In this embodiment of the invention, the above similarity calculation can be performed based on the dust concentration data corresponding to the dust features and the collected dust concentration data. Specifically, the similarity can be calculated between the concentration data obtained by the sensor and these image features, such as using cosine similarity or correlation coefficient, to determine the overlapping part between the image data and the concentration data.

[0067] The above methods enable precise monitoring and analysis of the actual state of dust, improving the accuracy and reliability of dust concentration measurement. This data fusion technology not only enhances the effectiveness of monitoring data but also provides a more scientific and real-time basis for dust management, thereby improving the overall performance and responsiveness of the environmental monitoring system.

[0068] The aforementioned dust characteristic data may include, but is not limited to, parameters such as current concentration, concentration trend, and corresponding early warning schemes. Specifically, since the morphology and characteristics of different dusts vary in different regions, for example, due to the large size of dust particles, the calculation method for concentration calculation can also vary. Therefore, different types of dust can be determined based on the detected different dust characteristic data.

[0069] In one possible embodiment, by calculating the similarity between the dust features in the image data and the collected dust features, the overlapping part in the image data is identified, and the features of the overlapping part of the image data are extracted to determine the dust feature data of the corresponding type of dust.

[0070] Optionally, before processing the dust image data and dust concentration data using the trained dust feature analysis model to obtain the dust feature data of the current area, the method further includes acquiring the dust feature analysis model to be trained and the training dust sample set; iteratively training the dust feature analysis model to be trained based on the dust concentration, the dust image corresponding to the dust concentration, and the dust concentration label corresponding to the dust image, and obtaining the trained dust feature analysis model after the iterative training is completed.

[0071] In this embodiment of the invention, the dust feature analysis model to be trained may include, but is not limited to, any deep learning model that cannot yet combine dust image data and dust concentration data for comprehensive analysis and processing to output dust feature data in the current area, such as BERT (Bidirectional Encoder Representations from Transformers), CLIP (Contrastive Language–Image Pre-training), and MMBT (Multimodal Bitransformer). Specifically, since different types of dust and dust image data have similarities, before the deep learning model is fully trained, outputting dust image data and dust concentration data to the dust feature analysis model to be trained may result in outputting dust feature data that deviates from the actual results.

[0072] The aforementioned training dust sample set may include, but is not limited to, dust concentration, dust images corresponding to dust concentration, and dust concentration labels corresponding to dust images.

[0073] The dust images corresponding to the dust concentrations mentioned above can be image information of different types of dust at different concentrations.

[0074] The dust concentration labels mentioned above can be specific and accurate concentration data of different types of dust stored in the database under different image information.

[0075] In one possible embodiment, the dust feature analysis model to be trained is iteratively trained using the aforementioned dust concentration, the dust image corresponding to the dust concentration, and the dust concentration label corresponding to the dust image. Specifically, the aforementioned dust concentration and the dust image corresponding to the dust concentration are input into the dust feature analysis model to be trained, and the parameters of the model are adjusted with the goal of infinitely approaching the dust concentration label corresponding to the dust image. When the similarity between the output dust feature data and the dust concentration label corresponding to the dust image is within a preset similarity threshold, the model training is completed, and the trained dust feature analysis model is obtained.

[0076] Optionally, the step of iteratively training the dust feature analysis model to be trained based on dust concentration, dust images corresponding to dust concentration, and dust concentration labels corresponding to dust images, and obtaining the trained dust feature analysis model after the iterative training is completed, further includes: using the latest acquired dust images corresponding to dust concentration as the latest training samples to replace the current training set; and iteratively training the dust feature analysis model to be trained using the latest training samples to obtain the trained dust feature analysis model.

[0077] In this embodiment of the invention, a training threshold detection can be performed on the training samples, and the current training samples can be judged according to the preset training threshold to determine whether they conform to the training content of the dust feature analysis model to be trained.

[0078] In one possible embodiment, if the detection error exceeds a preset training threshold or the coefficient of determination does not reach the threshold, updated training samples and corresponding optimized image preprocessing algorithms, such as denoising, contrast enhancement, and edge detection, will be obtained from the database to ensure that the quality and clarity of the input image and the corresponding training samples meet the requirements of clarity and quality. An adaptive filtering algorithm will be used to dynamically adjust the preprocessing parameters according to different environmental conditions to improve the preprocessing effect, and then the deep learning model will be retrained.

[0079] Optionally, the step of integrating and outputting dust feature data through the visualization module to generate a visualized dust feature report includes: determining the curves of each feature data in the dust feature data based on the dust feature data; obtaining the trend charts of each feature data based on the curves; generating corresponding early warning strategies based on the trend charts and curves, and generating a visualized report for display.

[0080] In this embodiment of the invention, dust characteristic data, such as dust size, shape, and concentration, are collected and analyzed. This data is then processed using a trained dust characteristic analysis model to plot curves of various characteristics. These curves are used to generate trend charts of dust characteristics, which are then displayed on a smart terminal to show the dust behavior and patterns changing over time. Based on this data and charts, early warning strategies are automatically formulated. For example, when the dust concentration is detected to exceed a safety threshold, management personnel are immediately notified to take action.

[0081] The above methods not only help monitoring teams track historical changes in dust levels but also predict future trends. All data and charts are presented in an intuitive visualization report, which improves decision-making efficiency and response speed, ensures environmental safety and the protection of human health, and enhances the efficiency of dust detection.

[0082] In one possible embodiment, after a user enters the dust detection system via ID detection, they can manage alarm operations through a computer and set alarm thresholds for different detection areas according to specific needs. Specifically, the alarm thresholds can be dynamically adjusted to ensure that the dust detection system can flexibly respond to changes in dust concentration under different environments.

[0083] More specifically, when the dust concentration exceeds a preset threshold, the system will issue an alarm to the user through pop-up windows, sound alerts, and red warning signs. The user can click on the alarm notification to view detailed information, such as the alarm time, detection area, and concentration value. The user can also view all historical alarm records in the alarm log, handle alarms, and add notes (such as handling measures and results).

[0084] In another possible embodiment, the user can also adjust the parameters of the image processing module (such as filter parameters, edge detection parameters, etc.) to optimize the image processing effect, and support saving and restoring the default parameter configuration, so that the user can quickly adjust the settings in different environments.

[0085] More specifically, the dust detection system can also provide a management interface for deep learning models, allowing users to upload, update, and switch model files, and view detailed model information (such as training parameters, version number, etc.).

[0086] like Figure 2 As shown, this embodiment of the invention also provides a dust detection device 200 based on image processing, which includes:

[0087] The acquisition module 201 is used to acquire dust data in the current area, the dust data including dust image data and dust concentration data;

[0088] Processing module 202 is used to process the dust image data and dust concentration data through a trained dust feature analysis model to obtain dust feature data in the current area;

[0089] The generation module 203 is used to integrate and output the dust feature data through the visualization module to generate a visualized dust feature report.

[0090] Optionally, the acquisition module 201 mentioned above includes:

[0091] The first acquisition submodule is used to collect dust in the current area within a certain period of time through a preset concentration sensor to obtain the corresponding dust concentration data.

[0092] The second acquisition submodule is used to capture images of dust in the current area within a certain period of time using a preset optical camera to obtain corresponding dust image data.

[0093] Optionally, the second acquisition submodule mentioned above also includes:

[0094] The first preprocessing unit is used to perform image enhancement preprocessing on the first image data obtained by the capture to obtain the second image data;

[0095] The second preprocessing unit is used to perform feature labeling preprocessing on the second image data to obtain the corresponding dust image data, wherein the dust image data contains the corresponding dust features.

[0096] Optionally, the above processing module 202 includes:

[0097] The first calculation submodule is used to perform similarity calculation between the dust features corresponding to the dust image data and the dust concentration data to determine the overlapping part of the dust image data and the dust concentration data;

[0098] The second calculation submodule is used to determine the dust characteristic data in the current area based on the overlapping part. The dust characteristic data includes the current concentration, concentration trend and corresponding early warning scheme.

[0099] Optionally, the above-mentioned device further includes:

[0100] The first training module is used to acquire the dust feature analysis model to be trained and the training dust sample set. The training dust sample set includes dust concentration, dust images corresponding to the dust concentration, and dust concentration labels corresponding to the dust images.

[0101] The second training module is used to iteratively train the dust feature analysis model to be trained based on dust concentration, dust images corresponding to the dust concentration, and dust concentration labels corresponding to the dust images, and to obtain the trained dust feature analysis model after the iterative training is completed.

[0102] Optionally, the second training module mentioned above also includes:

[0103] The first training submodule is used to replace the current training set with the latest training sample, which is the dust image corresponding to the latest dust concentration.

[0104] The second training submodule is used to iteratively train the dust feature analysis model to be trained using the latest training samples to obtain a trained dust feature analysis model.

[0105] Optionally, the above-mentioned generation module 203 includes:

[0106] The first determining submodule is used to determine the curve of each feature data in the dust feature data based on the dust feature data;

[0107] The second determining submodule is used to obtain a trend chart of each feature data based on the curve chart;

[0108] The third determination submodule is used to generate corresponding early warning strategies based on the trend chart and curve chart, and to generate a visual report for display.

[0109] like Figure 3 As shown, this embodiment of the invention also provides an electronic device 300, including a processor, which can execute any of the above-described image processing-based dust detection methods.

[0110] Specifically, it includes a processor 301 and a memory 302, as well as a computer program stored in the memory 302 and capable of running on the processor 301, which executes an image processing-based dust detection method, wherein:

[0111] The processor 301 executes the calculator program for the dust detection method based on image processing, stored in the memory 302, and performs the following steps:

[0112] Obtain dust data within the current area, including dust image data and dust concentration data;

[0113] By using a trained dust feature analysis model, the dust image data and dust concentration data are processed to obtain dust feature data for the current area.

[0114] The dust feature data is integrated and output through the visualization module to generate a visualized dust feature report.

[0115] Optionally, the processor 301 performs the process of acquiring dust data in the current area, including:

[0116] The dust concentration data in the current area is collected by a preset concentration sensor within a certain period of time.

[0117] The dust in the current area is captured by a preset optical camera within a certain period of time to obtain corresponding dust image data.

[0118] Optionally, the processor 301 executes the step of taking pictures of the dust in the current area with a preset optical camera within a certain period of time to obtain corresponding dust image data, including:

[0119] The first image data obtained by the capture is subjected to image enhancement preprocessing to obtain the second image data;

[0120] The second image data is preprocessed with feature labeling to obtain the corresponding dust image data, which contains the corresponding dust features.

[0121] Optionally, the processor 301 executes the trained dust feature analysis model to process the dust image data and dust concentration data to obtain dust feature data in the current area, including:

[0122] The similarity between the dust features corresponding to the dust image data and the dust concentration data is calculated to determine the overlapping part of the dust image data and the dust concentration data;

[0123] Based on the overlapping portion, dust characteristic data in the current area is determined, including the current concentration, concentration trend, and corresponding early warning scheme.

[0124] Optionally, before the processor 301 executes the trained dust feature analysis model to process the dust image data and dust concentration data to obtain dust feature data in the current area, the method further includes:

[0125] Obtain the dust feature analysis model to be trained and the training dust sample set, wherein the training dust sample set includes dust concentration, dust image corresponding to the dust concentration, and dust concentration label corresponding to the dust image;

[0126] Based on dust concentration, dust images corresponding to the dust concentration, and dust concentration labels corresponding to the dust images, the dust feature analysis model to be trained is iteratively trained, and a trained dust feature analysis model is obtained after the iterative training is completed.

[0127] Optionally, the processor 301 further executes the iterative training of the dust feature analysis model to be trained based on dust concentration, dust images corresponding to the dust concentration, and dust concentration labels corresponding to the dust images, and obtains the trained dust feature analysis model after the iterative training is completed. The method further includes:

[0128] The dust image corresponding to the latest dust concentration is used as the latest training sample to replace the current training set;

[0129] The dust feature analysis model to be trained is iteratively trained using the latest training samples to obtain a trained dust feature analysis model.

[0130] Optionally, the processor 301 further performs the function of integrating and outputting the dust feature data through the visualization module to generate a visualized dust feature report, including:

[0131] Based on the dust feature data, curves of each feature data are determined in the dust feature data;

[0132] Based on the aforementioned curve, trend charts for each feature data are obtained;

[0133] Based on the trend chart and curve, a corresponding early warning strategy is generated, and a visual report is produced for display.

[0134] This invention also provides a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the various processes of the image processing-based dust detection method or the application-side image processing-based dust detection method provided in this invention, and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0135] Those skilled in the art will understand that implementing all or part of the processes in the above embodiments can be accomplished by a computer program instructing related hardware, and can be stored in a computer-readable storage medium. When executed, the program can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0136] The above description discloses only preferred embodiments of the present invention and should not be construed as limiting the scope of the present invention. Therefore, equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1. A dust detection method based on image processing, characterized in that, include: Acquire dust data within the current area, including dust image data and dust concentration data; By using a trained dust feature analysis model, the dust image data and dust concentration data are processed to obtain dust feature data for the current area. The dust feature data is integrated and output through the visualization module to generate a visualized dust feature report; The acquisition of dust data in the current area includes: The dust concentration data in the current area is collected by a preset concentration sensor within a certain period of time. The dust in the current area is captured by a preset optical camera within a certain period of time to obtain corresponding dust image data. The step of capturing images of dust in the current area using a preset optical camera within a certain time period to obtain corresponding dust image data includes: The first image data obtained by the capture is subjected to image enhancement preprocessing to obtain the second image data; The second image data is preprocessed with feature labeling to obtain the corresponding dust image data, which contains the corresponding dust features; The process involves using a trained dust feature analysis model to process the dust image data and dust concentration data to obtain dust feature data for the current area, including: The similarity between the dust features corresponding to the dust image data and the dust concentration data is calculated to determine the overlapping part of the dust image data and the dust concentration data; Based on the overlapping portion, dust characteristic data in the current area is determined, including the current concentration, concentration trend, and corresponding early warning scheme.

2. The dust detection method based on image processing as described in claim 1, characterized in that, Before processing the dust image data and dust concentration data using the trained dust feature analysis model to obtain the dust feature data for the current area, the method further includes: Obtain the dust feature analysis model to be trained and the training dust sample set, wherein the training dust sample set includes dust concentration, dust image corresponding to the dust concentration, and dust concentration label corresponding to the dust image; Based on dust concentration, dust images corresponding to the dust concentration, and dust concentration labels corresponding to the dust images, the dust feature analysis model to be trained is iteratively trained, and a trained dust feature analysis model is obtained after the iterative training is completed.

3. The dust detection method based on image processing as described in claim 2, characterized in that, The method further includes iteratively training the dust feature analysis model to be trained based on dust concentration, dust images corresponding to the dust concentration, and dust concentration labels corresponding to the dust images, and obtaining a trained dust feature analysis model after the iterative training is completed. The dust image corresponding to the latest dust concentration is used as the latest training sample to replace the current training set; The dust feature analysis model to be trained is iteratively trained using the latest training samples to obtain a trained dust feature analysis model.

4. The dust detection method based on image processing as described in claim 1, characterized in that, The process of integrating and outputting the dust feature data through a visualization module to generate a visualized dust feature report includes: Based on the dust feature data, curves of each feature data are determined in the dust feature data; Based on the aforementioned curve, trend charts for each feature data are obtained; Based on the trend chart and curve, a corresponding early warning strategy is generated, and a visual report is produced for display.

5. A dust detection device based on image processing, characterized in that, include: The acquisition module is used to acquire dust data in the current area, including dust image data and dust concentration data; The processing module is used to process the dust image data and dust concentration data through a trained dust feature analysis model to obtain dust feature data in the current area. The generation module is used to integrate and output the dust feature data through the visualization module to generate a visualized dust feature report; The acquisition module is also used to collect dust in the current area within a certain period of time using a preset concentration sensor to obtain corresponding dust concentration data; and to capture images of dust in the current area within a certain period of time using a preset optical camera to obtain corresponding dust image data. The first image data obtained by the capture is subjected to image enhancement preprocessing to obtain the second image data; the second image data is subjected to feature labeling preprocessing to obtain the corresponding dust image data, wherein the dust image data contains the corresponding dust features; The processing module is further configured to perform similarity calculation between the dust features corresponding to the dust image data and the dust concentration data to determine the overlapping part of the dust image data and the dust concentration data; based on the overlapping part, determine the dust feature data in the current area, wherein the dust feature data includes the current concentration, concentration trend and corresponding early warning scheme.

6. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the steps of the image processing-based dust detection method as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the image processing-based dust detection method as described in any one of claims 1 to 4.

Images