Foreign matter identification and supervision system in coal mine area

Abstract

The invention provides a coal mine area foreign matter identification and supervision system, relates to the technical field of foreign matter identification, and aims to realize a foreign matter identification and supervision system with higher cost performance, and the system comprises an image acquisition module which is used for periodically carrying out image acquisition on a monitoring area to obtain an initial image; the image processing module is used for processing the initial image to obtain a feature image group, and the feature image group comprises a plurality of feature images; the image comparison module is used for comparing the N feature image groups at the continuous time points and judging whether the image difference at the current time point exceeds a preset feature difference threshold value or not; and the foreign matter judgment module is used for performing foreign matter identification when the image difference reaches a threshold value. The method has the advantages that foreign matter recognition is comprehensive and reliable, and computing power resources are saved.

Description

Technical Field

[0001] The present invention relates to the technical field of foreign body identification, and in particular to a foreign body identification and supervision system in a coal mine area. Background Art

[0002] As the main place for energy development, coal mining areas have received great attention for their safe production and resource management. The coal mine environment is highly risky, including hidden dangers such as foreign body intrusion, which seriously threatens personnel safety and production efficiency.

[0003] When identifying foreign intrusion, a series of image processing is required, including preprocessing, calling models for object recognition, etc. Since foreign object recognition requires continuous monitoring, frequent complex model calls and recognition processing are required, which can easily lead to large consumption of computing resources.

[0004] Therefore, it is necessary to improve the foreign body identification and supervision in coal mining areas to achieve a more cost-effective foreign body identification and supervision system. Summary of the invention

[0005] The purpose of the present invention is to provide a foreign body identification and supervision system in a coal mine area, which can realize a more cost-effective foreign body identification and supervision system.

[0006] The present invention is achieved through the following technical solutions:

[0007] A foreign body identification and supervision system in a coal mine area, comprising:

[0008] An image acquisition module is used to periodically acquire images of the monitoring area to obtain an initial image;

[0009] An image processing module, used for processing the initial image to obtain a set of feature image groups, wherein the feature image group includes a plurality of feature images;

[0010] An image comparison module is used to compare the N groups of feature image groups at consecutive time points to determine whether the image difference at the current time point exceeds a preset feature difference threshold;

[0011] The foreign body judgment module is used to identify foreign bodies when the image difference reaches a threshold.

[0012] Preferably, the method for processing the initial image is:

[0013] Adjusting the resolution of the initial image to a preset standard resolution;

[0014] Converting the initial image into a grayscale image;

[0015] Perform Gaussian filtering and normalization;

[0016] The filtering process is performed by Gabor filter.

[0017] Preferably, the method of performing filtering processing by using a Gabor filter is:

[0018] Set Gabor filters in multiple filtering directions separately:

[0019] G(x,y;λ,θ,ψ,σ,γ),θ=0°,45°,90°,135°;

[0020] Wherein, λ, ψ, σ, and γ represent the scale, phase offset, standard deviation of the Gaussian envelope, and aspect ratio of the Gabor filter, respectively, and θ represents the filtering direction of the Gabor filter;

[0021] Obtain a group of the characteristic image groups after filtering respectively:

[0022] I1(x,y)=I(x,y)*G(x,y;λ,0°,ψ,σ,γ);

[0023] I2(x,y)=I(x,y)*G(x,y;λ,45°,ψ,σ,γ);

[0024] I3(x,y)=I(x,y)*G(x,y;λ,90°,ψ,σ,γ);

[0025] I4(x,y)=I(x,y)*G(x,y;λ,135°,ψ,σ,γ);

[0026] Wherein, * is a two-dimensional convolution operation, I(x,y) represents the value of the pixel at the xth row and yth column in the image input to the Gabor filter, and I j (x, y) represents the feature value of the pixel at the xth row and yth column in the jth feature image obtained by processing the Gabor filter, j=1, 2, 3, 4.

[0027] Preferably, the method for comparing N groups of feature image groups at consecutive time points is:

[0028] While acquiring the characteristic image group other than the current time point, acquiring the foreign body judgment result at the corresponding time point, if there is no foreign body, no operation is performed, and if there is a foreign body, the foreign body is removed;

[0029] Respectively obtaining sub-difference reference values ​​of the feature image group at the current time point and the feature image groups at other time points among the N groups of feature image groups at consecutive time points;

[0030] Acquire a comprehensive difference parameter of the feature image group at a current time point according to all the sub-difference reference values;

[0031] If the comprehensive difference parameter is greater than the preset feature difference threshold, it is determined that there is a foreign object and the feature image group at the current time point is sent to the foreign object judgment module; otherwise, it is determined that there is no foreign object.

[0032] Preferably, the method for obtaining the sub-difference reference value of the feature image group at the current time point and the feature image group at other time points is:

[0033]

[0034] Among them, d N2i represents the sub-difference reference value between the feature image group at the current time point and the feature image group at the i-th time point, I N,j (x,y) represents the feature value of the pixel at the xth row and yth column in the jth feature image at the previous time point, I i,j (x, y) represents the feature value of the pixel at the xth row and yth column in the jth feature image at the ith time point. The resolution of the image is X×Y, and M is the total number of types of the feature images.

[0035] Preferably, the method for obtaining the comprehensive difference parameter of the feature image group at the current time point is:

[0036]

[0037] Among them, D all is the comprehensive difference parameter of the feature image group at the current time point, α i is the weight of the ith time point.

[0038] Preferably, the weight α of the i-th time point i The method to obtain is:

[0039]

[0040] Among them, e is a natural constant.

[0041] Preferably, the method for identifying foreign objects is:

[0042] Establishing a security database, storing contour feature data of a variety of security items in the security database;

[0043] Performing contour recognition on the feature image groups at the current time point respectively to obtain a plurality of recognized contour feature data;

[0044] Comparing the identified contour feature data with contour feature data of a plurality of security articles respectively;

[0045] If the difference between any identified contour feature data and the contour feature data of all safety items is greater than a preset contour difference threshold, the identified object corresponding to the contour feature data is determined to be a foreign object and an alarm is issued.

[0046] Preferably, the contour feature data includes area, perimeter, aspect ratio and roundness.

[0047] The technical solution of the present invention has at least the following advantages and beneficial effects:

[0048] The present invention can obtain image features covering a relatively long period of time through an image acquisition module and an image processing module, which is helpful to realize image foreign body recognition with strong continuity and reliability;

[0049] The present invention uses Gabor filters to perform multi-directional image processing, which helps to extract feature information such as direction, texture and edge, and facilitates obtaining more accurate image information comparison results;

[0050] The present invention compares the image changes in continuous time, and uses the comparison as the result of preliminary judgment to decide whether it is necessary to call the model for accurate foreign body recognition. When performing difference comparison on the pre-processed images, non-target changes such as illumination and background can be effectively filtered out, and the false alarm and missed alarm rates can be significantly reduced, which helps save computing resources.

[0051] The present invention assigns different weights to images at different time points when comparing the image at the current time point with the historical image. The closer the time is, the higher the importance of the comparison result of the images. On the basis of improving the comprehensiveness of the comparison, the accuracy of the difference comparison of continuous time is also improved.

[0052] The present invention constructs a coal mine monitoring system with comprehensive data coverage and high cost performance through the combination of image acquisition, feature extraction, continuous time difference comparison and foreign body recognition, and has strong adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 This is a schematic diagram of the principles of the coal mine foreign body identification and monitoring system provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0055] Example 1

[0056] This embodiment provides a foreign body identification and supervision system in a coal mine area. Figure 1 ,include:

[0057] An image acquisition module is used to periodically acquire images of the monitoring area to obtain an initial image;

[0058] An image processing module, used for processing the initial image to obtain a set of feature image groups, wherein the feature image group includes a plurality of feature images;

[0059] An image comparison module is used to compare the N groups of feature image groups at consecutive time points to determine whether the image difference at the current time point exceeds a preset feature difference threshold;

[0060] The foreign body judgment module is used to identify foreign bodies when the image difference reaches a threshold.

[0061] This embodiment effectively improves the reliability of foreign body monitoring in coal mines by acquiring image features at multiple time points. Pre-judgment through image feature differences at multiple time points helps reduce computing power consumption, especially during rest time and other situations where the scene does not change much. If there is no intruding foreign body, the scene difference at different time points will be very small, and there is no need to constantly call the model for foreign body identification.

[0062] In this embodiment, the method for processing the initial image is:

[0063] Adjusting the resolution of the initial image to a preset standard resolution;

[0064] Converting the initial image into a grayscale image;

[0065] Perform Gaussian filtering and normalization;

[0066] The filtering process is performed by Gabor filter.

[0067] Furthermore, the method of performing filtering processing by using a Gabor filter is:

[0068] Set Gabor filters in multiple filtering directions separately:

[0069] G(x,y;λ,θ,ψ,σ,γ),θ=0°,45°,90°,135°;

[0070] Wherein, λ, ψ, σ, and γ represent the scale, phase offset, standard deviation of the Gaussian envelope, and aspect ratio of the Gabor filter, respectively, and θ represents the filtering direction of the Gabor filter;

[0071] Obtain a group of the characteristic image groups after filtering respectively:

[0072] I1(x,y)=I(x,y)*G(x,y;λ,0°,ψ,σ,γ);

[0073] I2(x,y)=I(x,y)*G(x,y;λ,45°,ψ,σ,γ);

[0074] I3(x,y)=I(x,y)*G(x,y;λ,90°,ψ,σ,γ);

[0075] I4(x,y)=I(x,y)*G(x,y;λ,135°,ψ,σ,γ);

[0076] Wherein, * is a two-dimensional convolution operation, I(x,y) represents the value of the pixel at the xth row and yth column in the image input to the Gabor filter, and I j (x, y) represents the feature value of the pixel at the xth row and yth column in the jth feature image obtained by processing the Gabor filter, j=1, 2, 3, 4.

[0077] This embodiment uses Gabor filter to process the image in multiple directions, extracting rich feature information such as direction, texture and edge, and then can fully capture the complex structure and subtle changes of the monitoring area, providing high-quality feature data support for image comparison and difference analysis. The filtering direction θ of the Gabor filter can extract the image features in the corresponding direction; the scale λ of the Gabor filter, which is also the wavelength, can usually take a value greater than 2, and the phase offset ψ of the Gabor filter is generally in radians, with a value of [-π,π]; the standard deviation σ of the Gabor filter's Gaussian envelope can be 2π; the aspect ratio γ of the Gabor filter determines the shape of the Gabor filter's elliptical filter, which can be 0.5.

[0078] Next, the method for comparing the N groups of feature image groups at consecutive time points is:

[0079] While acquiring the characteristic image group other than the current time point, acquiring the foreign body judgment result at the corresponding time point, if there is no foreign body, no operation is performed, and if there is a foreign body, the foreign body is removed;

[0080] Respectively obtaining sub-difference reference values ​​of the feature image group at the current time point and the feature image groups at other time points among the N groups of feature image groups at consecutive time points;

[0081] Acquire a comprehensive difference parameter of the feature image group at a current time point according to all the sub-difference reference values;

[0082] If the comprehensive difference parameter is greater than the preset feature difference threshold, it is determined that there is a foreign object and the feature image group at the current time point is sent to the foreign object judgment module; otherwise, it is determined that there is no foreign object.

[0083] Specifically, the method for obtaining the sub-difference reference value of the feature image group at the current time point and the feature image group at other time points is:

[0084]

[0085] Among them, d N2i represents the sub-difference reference value between the feature image group at the current time point and the feature image group at the i-th time point, I N,j (x,y) represents the feature value of the pixel at the xth row and yth column in the jth feature image at the previous time point, I i,j (x, y) represents the feature value of the pixel at the xth row and yth column in the jth feature image at the ith time point. The resolution of the image is X×Y, and M is the total number of types of the feature images.

[0086] Furthermore, the method for obtaining the comprehensive difference parameter of the feature image group at the current time point is:

[0087]

[0088] Among them, D all is the comprehensive difference parameter of the feature image group at the current time point, α i is the weight of the ith time point.

[0089] As a preferred solution of this embodiment, the weight α of the i-th time point i The method to obtain is:

[0090]

[0091] Among them, e is a natural constant.

[0092] Finally, the method for identifying foreign objects is:

[0093] Establishing a security database, storing contour feature data of a variety of security items in the security database;

[0094] Performing contour recognition on the feature image group at the current time point respectively to obtain a plurality of recognized contour feature data; the contour recognition may specifically adopt a method such as Canny edge detection;

[0095] Comparing the identified contour feature data with contour feature data of a plurality of security articles respectively;

[0096] If the difference between any identified contour feature data and the contour feature data of all safety items is greater than a preset contour difference threshold, the identified object corresponding to the contour feature data is determined to be a foreign object and an alarm is issued.

[0097] As a preferred solution, the contour feature data includes area, perimeter, aspect ratio (width / height) and roundness (4π×area / perimeter). 2 ).

[0098] This embodiment compares the image changes at consecutive time points and uses the difference analysis as the basis for preliminary judgment to determine whether accurate foreign body recognition is required. When comparing image differences, the closer the time is to the current image, the more important the comparison result is, so a higher weight is given to it, making the system perform better when processing dynamic changes.

[0099] The contour feature data selected in this embodiment can extract the shape features of the object. When comparing, the average feature difference rate between the detected contour feature data and the contour feature data of the safety object can be calculated to determine whether they are consistent.

[0100] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A foreign body identification and supervision system in coal mining areas, characterized in that: include: An image acquisition module is used to periodically acquire images of the monitoring area to obtain an initial image; An image processing module, used for processing the initial image to obtain a set of feature image groups, wherein the feature image group includes a plurality of feature images; An image comparison module is used to compare the N groups of feature image groups at consecutive time points to determine whether the image difference at the current time point exceeds a preset feature difference threshold; The foreign body judgment module is used to identify foreign bodies when the image difference reaches a threshold.

2. A coal mine foreign body identification and supervision system according to claim 1, characterized in that: The method for processing the initial image is: Adjusting the resolution of the initial image to a preset standard resolution; Converting the initial image into a grayscale image; Perform Gaussian filtering and normalization; The filtering process is performed by Gabor filter.

3. A foreign body identification and monitoring system for coal mines according to claim 2, characterized in that: The method of filtering by Gabor filter is: Set Gabor filters in multiple filtering directions separately: G(x,y;λ,θ,ψ,σ,γ),θ=0°,45°,90°,135°; Wherein, λ, ψ, σ, and γ represent the scale, phase shift, standard deviation of the Gaussian envelope, and aspect ratio of the Gabor filter, respectively, and θ represents the filtering direction of the Gabor filter; Obtain a group of the characteristic image groups after filtering respectively: I1(x,y)=I(x,y)*G(x,y;λ,0°,ψ,σ,γ); I2(x,y)=I(x,y)*G(x,y;λ,45°,ψ,σ,γ); I3(x,y)=I(x,y)*G(x,y;λ,90°,ψ,σ,γ); I4(x,y)=I(x,y)*G(x,y;λ,135°,ψ,σ,γ); Wherein, * is a two-dimensional convolution operation, I(x,y) represents the value of the pixel at the xth row and yth column in the image input to the Gabor filter, and I j (x, y) represents the feature value of the pixel at the xth row and yth column in the jth feature image obtained by processing the Gabor filter, j=1, 2, 3, 4.

4. A coal mine foreign body identification and supervision system according to claim 1, characterized in that: The method for comparing N groups of feature image groups at consecutive time points is: While acquiring the characteristic image group other than the current time point, acquiring the foreign body judgment result at the corresponding time point, if there is no foreign body, no operation is performed, and if there is a foreign body, the foreign body is removed; Respectively obtaining sub-difference reference values ​​of the feature image group at the current time point and the feature image groups at other time points among the N groups of feature image groups at consecutive time points; Acquire a comprehensive difference parameter of the feature image group at a current time point according to all the sub-difference reference values; If the comprehensive difference parameter is greater than the preset feature difference threshold, it is determined that there is a foreign object and the feature image group at the current time point is sent to the foreign object judgment module; otherwise, it is determined that there is no foreign object.

5. A coal mine foreign body identification and supervision system according to claim 4, characterized in that: The method for obtaining the sub-difference reference value of the feature image group at the current time point and the feature image group at other time points is: Among them, d N2i represents the sub-difference reference value between the feature image group at the current time point and the feature image group at the i-th time point, I N,j (x,y) represents the feature value of the pixel at the xth row and yth column in the jth feature image at the previous time point, I i,j (x, y) represents the feature value of the pixel at the xth row and yth column in the jth feature image at the ith time point. The resolution of the image is X×Y, and M is the total number of types of the feature images.

6. A coal mine foreign body identification and supervision system according to claim 5, characterized in that: The method for obtaining the comprehensive difference parameter of the feature image group at the current time point is: Among them, D all is the comprehensive difference parameter of the feature image group at the current time point, α i is the weight of the ith time point.

7. A coal mine foreign body identification and supervision system according to claim 6, characterized in that: The weight α of the i-th time point i The method to obtain is: Among them, e is a natural constant.

8. A coal mine foreign body identification and supervision system according to claim 1, characterized in that: The method for identifying foreign objects is as follows: Establishing a security database, storing contour feature data of a variety of security items in the security database; Performing contour recognition on the feature image groups at the current time point respectively to obtain a plurality of recognized contour feature data; Comparing the identified contour feature data with contour feature data of a plurality of security articles respectively; If the difference between any identified contour feature data and the contour feature data of all safety items is greater than a preset contour difference threshold, the identified object corresponding to the contour feature data is determined to be a foreign object and an alarm is issued.

9. A foreign body identification and monitoring system for coal mine areas according to claim 8, characterized in that: The contour feature data includes area, perimeter, aspect ratio and roundness.

Images