Underground coal mine flame detection method and device based on visible light characteristics
By using flame detection method based on visible light characteristics in coal mines, the area image is converted into HSV color space and the characteristics are extracted, and the flame area is identified by combining color and size information, the problem of low accuracy of flame detection in coal mines is solved, and the flame recognition is improved Accuracy and security.
Patent Information
- Application Number
- CN202411037998.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-07-30
AI Technical Summary
The underground environment of coal mines is complex, with closed, narrow and uneven light, resulting in low accuracy of existing flame detection methods, making it difficult to effectively identify flame targets.
A flame detection method based on visible light characteristics is adopted, and the area image is acquired, and the area image is converted into HSV color space, and the tone, saturation and brightness characteristic images are extracted. Combined with color and size information, the characteristic value is calculated to identify the flame region.
It improves the accuracy of underground flame detection of coal mines, reduces false alarms and missed reports, and enhances the reliability of underground safety monitoring.
Smart Images

Figure CN118968023A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of image processing technology, and in particular to a method and device for detecting flames in underground coal mines based on visible light features. Background Art
[0002] The underground environment is relatively complex, and flammable and explosive gas is easily produced during the mining process, posing a serious threat to underground safety. Therefore, flame detection of the underground environment is an important measure to ensure underground safety.
[0003] Visible light cameras are usually deployed in mines to monitor unsafe factors in the underground environment. Using a large number of visible light cameras deployed in mines for flame detection is currently a common method. However, the underground environment of coal mines is closed, narrow, and has uneven lighting. The current methods for underground flame detection often detect lights or other objects similar to flames as flame targets, and the flame detection accuracy is low. Summary of the invention
[0004] The present application aims to solve one of the technical problems in the related art at least to some extent.
[0005] To this end, the first purpose of the present application is to propose a coal mine underground flame detection method based on visible light characteristics to achieve accurate flame detection.
[0006] The second purpose of the present application is to propose a coal mine underground flame detection device based on visible light characteristics.
[0007] The third objective of the present application is to provide an electronic device.
[0008] A fourth objective of the present application is to provide a computer-readable storage medium.
[0009] A fifth object of the present application is to provide a computer program product.
[0010] To achieve the above-mentioned purpose, the first embodiment of the present application proposes a method for detecting flames in coal mines based on visible light characteristics, comprising:
[0011] Acquire a current downhole regional image, and identify one or more target regional images of suspected flame regions in the regional image;
[0012] Acquire a hue H channel image, a saturation S channel image, and a lightness V channel image corresponding to the target area image, and determine a feature image based on the H channel image, the S channel image, and the V channel image;
[0013] Determining a first characteristic value of the target area image according to the characteristic image;
[0014] Determining a second eigenvalue and a third eigenvalue of the target area image according to color information and size information of the suspected flame area in the target area image;
[0015] A flame region in the region image is determined according to the first eigenvalue, the second eigenvalue, and the third eigenvalue.
[0016] To achieve the above-mentioned purpose, the second embodiment of the present application proposes a coal mine underground flame detection device based on visible light characteristics, comprising:
[0017] A first acquisition module is used to acquire a current downhole regional image and identify a target regional image of one or more suspected flame regions in the regional image;
[0018] A second acquisition module is used to acquire a hue H channel image, a saturation S channel image and a lightness V channel image corresponding to the target area image, and determine a feature image according to the H channel image, the S channel image and the V channel image;
[0019] A third acquisition module, used to determine a first characteristic value of the target area image according to the characteristic image;
[0020] A fourth acquisition module, used to determine the second eigenvalue and the third eigenvalue of the target area image according to the color information and size information of the suspected flame area in the target area image;
[0021] A detection module is used to determine the flame area in the area image according to the first eigenvalue, the second eigenvalue and the third eigenvalue.
[0022] To achieve the above-mentioned purpose, the third aspect of the present application provides an electronic device, including: a processor, and a memory communicatively connected to the processor;
[0023] The memory stores computer-executable instructions;
[0024] The processor executes the computer-executable instructions stored in the memory to implement the method described in the embodiment of the first aspect.
[0025] To achieve the above-mentioned purpose, the fourth aspect embodiment of the present application proposes a computer-readable storage medium, in which computer-readable storage medium is stored computer execution instructions, and when the computer execution instructions are executed by a processor, they are used to implement the method described in the first aspect embodiment.
[0026] To achieve the above-mentioned purpose, the fifth aspect of the present application proposes a computer program product, including a computer program, which implements the method described in the first aspect when executed by a processor.
[0027] The present application provides a method and device for detecting flames in underground coal mines based on visible light features. By extracting a target area image of a suspected flame area and obtaining a first eigenvalue, a second eigenvalue, and a third eigenvalue of the target area image, the method and device identify whether a flame area exists in the target area image from a color dimension and a size dimension based on the first eigenvalue, the second eigenvalue, and the third eigenvalue, thereby improving the accuracy of flame identification in underground coal mine environments.
[0028] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS
[0029] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:
[0030] Figure 1 A schematic diagram of a flow chart of a method for detecting flames in underground coal mines based on visible light characteristics provided in an embodiment of the present application;
[0031] Figure 2 A schematic flow chart of another method for detecting flames in underground coal mines based on visible light characteristics provided in an embodiment of the present application;
[0032] Figure 3 A schematic structural diagram of a coal mine underground flame detection device based on visible light characteristics provided in an embodiment of the present application. DETAILED DESCRIPTION
[0033] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.
[0034] The following describes a method and device for detecting flames in an underground coal mine based on visible light characteristics according to an embodiment of the present application with reference to the accompanying drawings.
[0035] Figure 1 The following is a flow chart of a method for detecting flames in coal mines based on visible light characteristics provided in an embodiment of the present application. Figure 1 As shown, the method comprises the following steps:
[0036] S101, obtaining a current regional image of the well, and identifying one or more target regional images of suspected flame regions in the regional image.
[0037] In some implementations, one or more monitoring cameras may be deployed in an underground area of a coal mine, and images of the underground area may be captured in real time by the monitoring cameras.
[0038] It is understandable that due to the dim environment in the mine, the flame area should be presented as a brighter area in the regional image. In this embodiment, the regional image can be preliminarily identified to determine that one or more bright areas in the regional image are suspected flame areas, and the image of the range where the suspected flame area is located is extracted as the target area image.
[0039] S102, obtaining a hue H channel image, a saturation S channel image and a lightness V channel image corresponding to the target area image, and determining a feature image according to the H channel image, the S channel image and the V channel image.
[0040] Considering that there may be false targets with similar flame hue, such as lights or reflected light caused by lights shining on objects, in the underground environment, the target area image in the RGB color space captured by the monitoring camera is converted to the HSV color space in this embodiment to obtain a hue H channel image, a saturation S channel image, and a lightness V channel image.
[0041] It can be understood that the color characteristics of the flame are red tones, yellow tones and high-brightness white. Therefore, pixels can be screened in the H channel image, the S channel image and the V channel image according to the color characteristics of the flame, and the pixels that meet the color characteristics of the flame are marked to obtain the marked H channel image, the marked S channel image and the marked V channel image.
[0042] Furthermore, the intersection area of the marked H channel image, the marked S channel image and the marked V channel image can be obtained. The pixel points in the intersection area are the pixels that are marked simultaneously in the marked H channel image, the marked S channel image and the marked V channel image. Therefore, the image corresponding to the intersection area is a more detailed image of the suspected flame area extracted based on the color features, and the image corresponding to the intersection area is used as the feature image.
[0043] S103: Determine a first characteristic value of the target area image according to the characteristic image.
[0044] It can be understood that compared with the fixed illumination area, the real flame area shakes, so the area image captured in the previous frame can be obtained, and the target area position in the area image of the previous frame and the feature image corresponding to the target area position can be extracted; based on the feature image of the previous frame and the current real-time feature image, it is determined whether the feature image has changed, thereby obtaining the first feature value.
[0045] In some implementations, the sizes of the current target area image and the corresponding previous frame target area image can be unified, for example, the scale sizes of the current target area image and the corresponding previous frame target area image can be adjusted to 64*64 to facilitate comparative analysis of the previous frame target area image and the current target area image.
[0046] In some implementations, the similarity between the feature image of the previous frame and the current real-time feature image can be obtained as the first feature value. The greater the similarity, the more likely the feature image is an area caused by fixed lighting. The smaller the similarity, the more likely the current feature image is an area where the flame is located.
[0047] S104, determining a second eigenvalue and a third eigenvalue of the target area image according to color information and size information of the suspected flame area in the target area image.
[0048] It can be understood that the target area image is an image of the area where the suspected flame area is located, so the pixel difference between the target area image and the corresponding previous frame target area image can be used as the second eigenvalue of the current target area image.
[0049] Since the characteristic of flames that they are easy to shake will cause pixel value fluctuations in the images of the previous and next frames, the smaller the second eigenvalue of the target area image is, the smaller the color difference between the previous and next frames is, and the more likely it is that the area is caused by fixed lighting. The larger the second eigenvalue of the target area image is, the greater the color difference between the previous and next frames is, and the more likely it is that the target area image is a flame area.
[0050] In some implementations, the size information actually extracted from the target area image can also be obtained to reflect the area size of the current suspected flame area; the size information of the corresponding previous frame target area image is further obtained, and the difference between the size information of the previous frame and the current size information is determined as the third eigenvalue. The smaller the third eigenvalue is, the smaller the change between the target area images of the previous and next frames is, and the more likely it is that the area is caused by fixed lighting. The larger the third eigenvalue is, the greater the change in the area of the target area images of the previous and next frames is, and the more likely it is that the flame is located.
[0051] S105, determining a flame region in the region image according to the first eigenvalue, the second eigenvalue, and the third eigenvalue.
[0052] Optionally, a weighted summation of the first eigenvalue, the second eigenvalue and the third eigenvalue can be performed by presetting a weight coefficient to obtain a comprehensive eigenvalue, and the comprehensive eigenvalue is compared with a preset threshold to determine whether the current target area image is a flame area. Optionally, the first eigenvalue, the second eigenvalue and the third eigenvalue can be composed into a feature vector, and the feature vector is input into a pre-trained classification model to determine whether the current target area image is a flame area.
[0053] The target area image of each suspected flame area in the area image is identified to determine whether there is a flame area in the area image, thereby reducing the possibility of accidents.
[0054] In this embodiment, the suspected flame area in the regional image is extracted, and the suspected flame area is analyzed in a targeted manner to reduce the analysis cost. A feature image in the HSV color space is obtained based on the target area image, and then a first eigenvalue is obtained according to the feature image. The second eigenvalue and the third eigenvalue are determined in combination with the color information and size information of the target area image. Then, the flame area is identified according to the first eigenvalue, the second eigenvalue and the third eigenvalue. The flame characteristics themselves and the color characteristics and size characteristics of the target area image are combined to perform flame identification, thereby improving the accuracy of flame identification.
[0055] Figure 2 The following is a flow chart of another method for detecting flames in coal mines based on visible light characteristics provided in an embodiment of the present application. Figure 2 As shown, the method comprises the following steps:
[0056] S201, obtaining a current downhole regional image, and identifying one or more target regional images of suspected flame regions in the regional image.
[0057] In some implementations, the regional image can be input into a pre-trained target recognition network to output one or more suspected flame regions in the regional image; and the minimum rectangular bounding box image corresponding to the suspected flame region is extracted as the target regional image of the suspected flame region.
[0058] Optionally, the target recognition network can be a conventional deep neural network. After the regional image is input into the pre-trained target recognition network, the target recognition network extracts features of the regional image, thereby outputting a suspected flame region. In this embodiment, the output result can be a four-tuple, and the four numerical values in the four-tuple can represent the positions of the four sides of the minimum rectangular bounding box of the suspected flame region, so as to determine the minimum rectangular bounding box of the suspected flame region according to the output four-tuple; for example, the four-tuple is {(1, t), (r, b)}, where (1, t) represents the upper left corner coordinates of the minimum rectangular bounding box, and (r, b) represents the lower right corner coordinates of the minimum rectangular bounding box; the image corresponding to the minimum rectangular bounding box is used as the target region image of the suspected flame region, and each target region image is analyzed to reduce the analysis cost while improving the recognition accuracy.
[0059] Furthermore, it is also possible to obtain the regional image of the well at the previous moment; match the suspected flame area in the regional image at the previous moment with the suspected flame area in the current regional image to obtain a regional matching set; and delete the target regional image corresponding to the suspected flame area that does not exist in the regional matching set.
[0060] Optionally, the Hungarian matching algorithm can be used to match the suspected flame area at the previous moment with the current suspected flame area to determine the current area matching set; in this embodiment, the intersection-and-union ratio between the minimum rectangular enclosing box of the suspected flame area and the minimum rectangular enclosing box of the current suspected flame area can be calculated, so as to determine whether it is a matching area based on the intersection-and-union ratio, and delete the target area image corresponding to the suspected flame area that does not exist in the area matching set to reduce interference from other factors.
[0061] In some implementations, in order to avoid the impact of short-term result fluctuations, a tracking threshold and an exit tracking threshold can also be set for screening. If the target area image does not meet the tracking threshold, it is marked for tracking. When the target area image meets the tracking threshold, it is tracked and analyzed. Accordingly, when the target area image meets the exit tracking threshold, it is deleted from the tracking list. As an example, assume that the tracking threshold and the exit tracking threshold are both set to 3. For the target area image of any suspected flame area at the current i-th moment, the corresponding target area image is matched at the i-1th moment and the i-2th moment, that is, the suspected flame area appears three times in a row and satisfies the tracking threshold. Then, it is determined to perform target tracking on the target area image of the current suspected flame area, that is, to perform subsequent analysis and identification on the target area image of the suspected flame area. If the target area image of a suspected flame area previously satisfies the tracking threshold, for example, it exists at the i-3th moment, the i-2nd moment and the i-1th moment, but does not exist in the area image at the current i-th moment, then the target area image of the suspected flame area is tracked and marked. If the target area image of the suspected flame area appears again at the i+1th moment, it is directly identified and analyzed without tracking judgment. If the target area image of the suspected flame area does not exist at both the i+1th moment and the i+2th moment, that is, it does not exist for three consecutive times, then it is determined to exit tracking of the target area image of the suspected flame area and delete it from the tracking mark list.
[0062] In the embodiment of the present application, the implementation method of step S201 can be implemented by any of the methods in the embodiments of the present disclosure, which is not limited here and will not be repeated.
[0063] S202 , binarizing the H channel image, the S channel image, and the V channel image according to a first preset condition to obtain a first H channel image, a first S channel image, and a first V channel image.
[0064] The first preset condition includes an H channel threshold, a first S channel threshold and a first V channel threshold.
[0065] In some implementations, a target pixel point whose pixel value is less than an H channel threshold value in an H channel image is obtained, the pixel value of the target pixel point is set to a maximum value, and other pixels are set to a minimum value, to obtain a binary first H channel image.
[0066] In some implementations, a target pixel point whose pixel value is greater than a first S channel threshold value is obtained in the S channel image, the pixel value of the target pixel point is set to the maximum value, and other pixels are set to the minimum value to obtain a binary first S channel image.
[0067] In some implementations, a target pixel point whose pixel value is greater than a first V channel threshold value in a V channel image is obtained, the pixel value of the target pixel point is set to a maximum value, and other pixels are set to a minimum value, to obtain a binary first V channel image.
[0068] S203 : Determine a first characteristic image according to the first H channel image, the first S channel image, and the first V channel image.
[0069] In this embodiment, the first characteristic image is a screened regional image with red and yellow as the main tones; optionally, the first characteristic image can be expressed as:
[0070]
[0071] Among them, A ry represents the first feature image; Represents the first H channel image; Represents the first S channel image; Represents the first V channel image; It is the union area of the first S channel image and the first V channel image, the union of the target pixel in the first S channel image and the target pixel in the first V channel image, and further the intersection area of the union area and the first H channel image, that is, the intersection of the target pixel, so as to obtain the first feature image, which is a unit unified binary image. For example, the pixel value of the target pixel in the intersection finally screened out is set to 1, and the other pixel values are set to 0, so as to obtain the binary first feature image of 1 and 0.
[0072] S204, binarizing the S channel image and the V channel image according to a second preset condition to obtain a second S channel image and a second V channel image; wherein the second preset condition includes a second S channel threshold and a second V channel threshold.
[0073] In some implementations, a target pixel point whose pixel value is less than a second S channel threshold value in the S channel image is obtained, the pixel value of the target pixel point is set to the maximum value, and other pixels are set to the minimum value to obtain a binary second S channel image.
[0074] In some implementations, a target pixel point in the V channel image whose pixel value is greater than a second V channel threshold is obtained, the pixel value of the target pixel point is set to a maximum value, and other pixels are set to a minimum value to obtain a binary second V channel image.
[0075] S205 , determining a second characteristic image according to the second S channel image and the second V channel image.
[0076] In this embodiment, the second characteristic image is a screened high-brightness white area image; optionally, the second characteristic image can be expressed as:
[0077]
[0078] Among them, A w represents a second feature image; Represents the second S channel image; Represents the second V channel image; It is the intersection area of the second S channel image and the second V channel image, that is, the intersection of the target pixel in the second S channel image and the target pixel in the second V channel image. The second feature image is a unit unified binary image. For example, the pixel value of the target pixel in the intersection finally screened out is set to 1, and the other pixel values are set to 0, to obtain a binary second feature image of 1 and 0.
[0079] S206: Determine a union area of the first feature image and the second feature image as the feature image.
[0080] Optionally, the feature image can be expressed as: A = A ry ∪A w , where A is a feature image, which is a binary image with pixel values of 0 and 1, and the union of the pixel points with a pixel value of 1 in the first feature image and the second feature image is the pixel points with a pixel value of 1 in the feature image.
[0081] S207: Determine a first characteristic value of the target area image according to the characteristic image.
[0082] In some implementations, a feature image corresponding to the target area image at a previous moment may be obtained, and a first matching degree between the feature image at a previous moment and the current feature image may be calculated; the first matching degree may be filtered to obtain a first feature value of the target area image.
[0083] In this embodiment, before calculating the first matching degree between the feature image at the previous moment and the current feature image, the feature image at the previous moment and the current feature image can be scaled to images of size 64*64, and the image size can be obtained. and Then obtain and The first degree of matching between is:
[0084]
[0085] Among them, l′ sh,i Indicates the first matching degree; for and The absolute difference of the pixels at the corresponding positions in .
[0086] Furthermore, the first matching degree is filtered, and the first eigenvalue is:
[0087] l sh,i =η1l′ sh,i +(1-η1)l sh,i-1
[0088] Among them, η sh,i is the first eigenvalue of the current target area image; l sh,i-1 is the first eigenvalue of the target area image at the previous moment, which is a known value after real-time recognition and analysis of the image at the previous moment; η1 is the first filter coefficient, which is a set value.
[0089] It should be noted that, in this embodiment, only the target region image is extracted and recorded for the region image at the first moment, and no analysis is performed, and the first characteristic value of the target region image at the first moment is set to 0.
[0090] S208, determining a second eigenvalue and a third eigenvalue of the target area image according to the color information and size information of the suspected flame area in the target area image.
[0091] Optionally, the target area image at a previous moment may be acquired, and a second matching degree between the target area image at a previous moment and the current target area image may be calculated; and the second matching degree may be filtered to obtain a second eigenvalue of the target area image.
[0092] In this embodiment, before calculating the second matching degree between the target area image at the previous moment and the current target area image, the target area image at the previous moment and the current target area image can be scaled to images of size 64*64, and B i-1 and Bi , and then obtain B i-1 and B i The second degree of matching between is:
[0093]
[0094] Among them, l sh,i Indicates the second matching degree; For B i-1 and B i The absolute difference of the pixels at the corresponding positions in .
[0095] Further, the second matching degree is filtered, and the second eigenvalue is:
[0096] l ap,i =η2l′ ap,i +(1-η2)l ap,i-1
[0097] Among them, l ap,i Indicates the second eigenvalue of the current target area image; l ap,i-1 is the second eigenvalue of the target area image at the previous moment. If the previous moment is the first moment, the second eigenvalue is 0; η2 is the second filter coefficient, which is a set value.
[0098] Optionally, the width and height of the target area image can be obtained, and the absolute difference in width and height between the target area image at the previous moment and the current target area image can be calculated; it can be understood that the target area image is a minimum rectangular bounding box image of the suspected flame area extracted based on the neural network, so it can be judged whether the suspected flame area has a size change based on the size of the target area image, and the width and height of the target area image can be determined based on the four-tuple corresponding to the minimum rectangular bounding box, that is, the width and height of the target area image are determined according to the upper left corner coordinates and the lower right corner coordinates. The height and width of the target area image can be expressed as:
[0099] w=r-1
[0100] h = tb
[0101] Among them, w represents the width, h represents the height, (1, t) represents the coordinates of the upper left corner of the minimum rectangular bounding box, and (r, b) represents the coordinates of the lower right corner of the minimum rectangular bounding box.
[0102] The absolute difference in width and height can be expressed as:
[0103] d′ w,i =|w i -w i-1 |
[0104] d′ h,i =|h i -h i-1 |
[0105] Among them, d′ w,i Indicates the absolute difference in width; d′ h,i Indicates the absolute height difference; w i and h i Indicates the width and height of the current target area image; w i-1 and h i-1 Indicates the width and height of the target area image at the previous moment.
[0106] Furthermore, the absolute difference in width and the absolute difference in height are filtered respectively to obtain a width matching feature value and a height matching feature value of the target area image.
[0107] Optionally, the width matching eigenvalue and the height matching eigenvalue can be expressed as:
[0108] d w,i =η3d′ w,i +(1-η3)d w,i-1
[0109] d h,i =η4d′ h,i +(1-η4)d h,i-1
[0110] Among them, d w,i and d h,i Respectively represent the width matching eigenvalue and the height matching eigenvalue; d w,i-1 and d h,i-1 They respectively represent the width matching eigenvalue and height matching eigenvalue of the target area image at the previous moment. If the previous moment is the first moment, the width matching eigenvalue and the height matching eigenvalue are both 0; η3 and η4 are the third filter coefficient and the fourth filter coefficient, respectively, and are both set values.
[0111] S209, determining the flame region in the region image according to the first eigenvalue, the second eigenvalue and the third eigenvalue.
[0112] In some implementations, a feature input vector can be determined based on the first eigenvalue, the second eigenvalue, and the third eigenvalue; the feature input vector is input into a pre-trained classifier to output a flame classification result of the target area image; and the flame area in the area image is determined based on the flame classification result of the target area image.
[0113] Optionally, the classifier can be a multilayer perceptron, with an S-shaped Sigmoid function as an activation function, and a binary cross entropy (BCE) loss function is applied to train the classifier, and a feature input vector is determined by the first eigenvalue, the second eigenvalue and the third eigenvalue, and the feature input vector is input into a pre-trained classifier, and a flame classification result of the corresponding target area image is output. The flame classification result can be a numerical representation, for example, 0 represents no flame and 1 represents flame, so that the flame area in the regional image is determined according to the flame classification results of each target area image, and processing is performed in time.
[0114] Furthermore, in order to avoid false alarms and result jitter, the detection signal value can also be determined based on the flame classification result; the detection signal value is filtered based on a preset weight coefficient to determine the alarm signal value of the target area image; in response to the alarm signal value meeting the alarm condition, the target area image is determined to be a flame area, and a flame alarm is issued.
[0115] In this embodiment, the detection signal value may be a value corresponding to the flame classification result, that is, the detection signal value is 1 and 0, 1 indicates that a flame is detected, and 0 indicates that no flame is detected; the alarm signal value of the target area image can be expressed as:
[0116]
[0117] Among them, a i Indicates the alarm signal value; Represents the detection signal value of the current target area image; a i-1 It represents the alarm signal value of the target area image at the previous moment. After the regional image analysis and recognition at the previous moment is completed, the alarm signal value is a known value. If the previous moment is the first moment, the default detection signal value is 0. η is the preset weight coefficient, and the value range is (0,1).
[0118] In this embodiment, the suspected flame area in the regional image is identified by a neural network, and the suspected flame area is screened, and other interfering regional images are preliminarily removed to reduce the amount of calculation; based on the color characteristics presented by the flame, the images of the three channels H, S and V corresponding to the target regional image are binarized respectively, so as to obtain a feature image that meets the flame characteristics, and the first eigenvalue of the color feature is determined by matching with the feature image at the previous moment; further, the pixel difference between the current target regional image and the target regional image corresponding to the previous moment is obtained to determine the second eigenvalue including the appearance color information, and the third eigenvalue including the size information is determined according to the absolute difference in width and the absolute difference in height between the target regional image and the target regional image corresponding to the previous moment, and the first eigenvalue, the second eigenvalue and the third eigenvalue are composed of a feature input vector and input into a pre-trained classifier for flame recognition, and finally the flame area of the regional image is obtained according to the flame recognition results of each target regional image, and an alarm is issued for the determined flame area, so as to improve the accuracy of flame detection in coal mines and reduce the possibility of accidents.
[0119] In order to implement the above-mentioned embodiment, the present application also proposes a coal mine underground flame detection device based on visible light characteristics.
[0120] Figure 3 The following is a schematic diagram of the structure of a coal mine underground flame detection device based on visible light characteristics provided in an embodiment of the present application. Figure 3 As shown, the coal mine underground flame detection device 300 based on visible light characteristics includes:
[0121] The first acquisition module 301 is used to acquire a regional image of the current well and identify a target regional image of one or more suspected flame regions in the regional image;
[0122] The second acquisition module 302 is used to acquire a hue H channel image, a saturation S channel image and a lightness V channel image corresponding to the target area image, and determine a feature image according to the H channel image, the S channel image and the V channel image;
[0123] A third acquisition module 303 is used to determine a first characteristic value of the target area image according to the characteristic image;
[0124] The fourth acquisition module 304 is used to determine the second eigenvalue and the third eigenvalue of the target area image according to the color information and size information of the suspected flame area in the target area image;
[0125] The detection module 305 is used to determine the flame area in the area image according to the first eigenvalue, the second eigenvalue and the third eigenvalue.
[0126] Furthermore, in a possible implementation of the embodiment of the present application, the second acquisition module 302 includes:
[0127] Binarization is performed on the H channel image, the S channel image and the V channel image according to a first preset condition to obtain a first H channel image, a first S channel image and a first V channel image; wherein the first preset condition includes an H channel threshold, a first S channel threshold and a first V channel threshold;
[0128] Determine a first characteristic image according to the first H channel image, the first S channel image and the first V channel image;
[0129] Binarization is performed on the S channel image and the V channel image according to a second preset condition to obtain a second S channel image and a second V channel image; wherein the second preset condition includes a second S channel threshold and a second V channel threshold;
[0130] Determine a second feature image according to the second S channel image and the second V channel image;
[0131] A union area of the first feature image and the second feature image is determined as the feature image.
[0132] Furthermore, in a possible implementation of the embodiment of the present application, the third acquisition module 303 includes:
[0133] Obtaining a feature image corresponding to the target area image at a previous moment, and calculating a first matching degree between the feature image at a previous moment and the current feature image;
[0134] The first matching degree is filtered to obtain a first eigenvalue of the target area image.
[0135] Furthermore, in a possible implementation of the embodiment of the present application, the fourth acquisition module 304 includes:
[0136] Acquire the target area image at the previous moment, and calculate the second matching degree between the target area image at the previous moment and the current target area image;
[0137] The second matching degree is filtered to obtain a second eigenvalue of the target area image.
[0138] Furthermore, in a possible implementation of the embodiment of the present application, the fourth acquisition module 304 includes:
[0139] Obtain the width and height of the target area image, and calculate the absolute difference in width and height between the target area image at the previous moment and the current target area image;
[0140] The absolute difference in width and the absolute difference in height are filtered respectively to obtain the width matching feature value and the height matching feature value of the target area image.
[0141] Furthermore, in a possible implementation of the embodiment of the present application, the first acquisition module 301 includes:
[0142] Input the region image into a pre-trained target recognition network, and output one or more suspected flame regions in the region image;
[0143] The minimum rectangular bounding box image corresponding to the suspected flame area is extracted as the target area image of the suspected flame area.
[0144] Furthermore, in a possible implementation of the embodiment of the present application, the apparatus 300 further includes:
[0145] Obtain the regional image of the well at the previous moment;
[0146] The suspected flame region in the regional image at the previous moment is matched with the suspected flame region in the current regional image to obtain a regional matching set; the target regional image corresponding to the suspected flame region that does not exist in the regional matching set is deleted.
[0147] Furthermore, in a possible implementation of the embodiment of the present application, the detection module 305 includes:
[0148] Determine a characteristic input vector according to the first eigenvalue, the second eigenvalue and the third eigenvalue;
[0149] Input the feature input vector into the pre-trained classifier and output the flame classification result of the target area image;
[0150] According to the flame classification result of the target area image, the flame area in the area image is determined.
[0151] Furthermore, in a possible implementation of the embodiment of the present application, the detection module 305 further includes:
[0152] Determine the detection signal value according to the flame classification result;
[0153] Filtering the detection signal value based on a preset weight coefficient to determine the alarm signal value of the target area image;
[0154] In response to the alarm signal value satisfying the alarm condition, the target area image is determined to be a flame area, and a flame alarm is issued.
[0155] It should be noted that the aforementioned explanation of the embodiment of the coal mine underground flame detection method based on visible light characteristics is also applicable to the coal mine underground flame detection device based on visible light characteristics of this embodiment, and will not be repeated here.
[0156] In the embodiment of the present application, a suspected flame area in a regional image is identified by a neural network, and the suspected flame area is screened, and other interfering regional images are preliminarily removed to reduce the amount of calculation; based on the color characteristics presented by the flame, the images of the three channels H, S and V corresponding to the target regional image are binarized respectively, so as to obtain a feature image that meets the flame characteristics, and the first eigenvalue of the color feature is determined by matching with the feature image at the previous moment; further, the pixel difference between the current target regional image and the target regional image corresponding to the previous moment is obtained to determine the second eigenvalue including the appearance color information, and the third eigenvalue including the size information is determined according to the absolute difference in width and the absolute difference in height between the target regional image and the target regional image corresponding to the previous moment, and the first eigenvalue, the second eigenvalue and the third eigenvalue are used to form a feature input vector, which is input into a pre-trained classifier for flame recognition, and finally the flame area of the regional image is obtained according to the flame recognition results of each target regional image, and an alarm is issued for the determined flame area, so as to improve the accuracy of flame detection in coal mines and reduce the possibility of accidents.
[0157] In order to implement the above embodiments, the present application also proposes an electronic device, comprising: a processor, and a memory communicatively connected to the processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the method provided by the above embodiments.
[0158] In order to implement the above embodiments, the present application also proposes a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the methods provided by the above embodiments.
[0159] In order to implement the above embodiments, the present application also proposes a computer program product, including a computer program, which implements the methods provided by the above embodiments when executed by a processor.
[0160] The collection, storage, use, processing, transmission, provision and disclosure of user personal information involved in this application are in compliance with relevant laws and regulations and do not violate public order and good morals.
[0161] It should be noted that personal information from users should be collected for legitimate and reasonable purposes and should not be shared or sold outside of these legitimate uses. In addition, such collection / sharing should be carried out after receiving the user's informed consent, including but not limited to notifying the user to read the user agreement / user notice and sign the agreement / authorization including authorization of relevant user information before the user uses the function. In addition, any necessary steps should be taken to protect and safeguard access to such personal information data and ensure that others who have access to personal information data comply with its privacy policy and procedures.
[0162] The present application is expected to provide an implementation scheme for users to selectively block the use or access of personal information data. That is, the present disclosure is expected to provide hardware and / or software to prevent or block access to such personal information data. Once the personal information data is no longer needed, the risk can be minimized by limiting data collection and deleting the data. In addition, when applicable, such personal information is de-identified to protect the privacy of the user.
[0163] In the description of the aforementioned embodiments, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.
[0164] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.
[0165] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.
[0166] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute the instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.
[0167] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.
[0168] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.
[0169] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.
[0170] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.
Claims
1. A method for detecting flames in coal mines based on visible light characteristics, characterized in that: include: Acquire a current downhole regional image, and identify one or more target regional images of suspected flame regions in the regional image; Acquire a hue H channel image, a saturation S channel image, and a lightness V channel image corresponding to the target area image, and determine a feature image based on the H channel image, the S channel image, and the V channel image; Determining a first characteristic value of the target area image according to the characteristic image; Determining a second eigenvalue and a third eigenvalue of the target area image according to color information and size information of the suspected flame area in the target area image; A flame region in the region image is determined according to the first eigenvalue, the second eigenvalue, and the third eigenvalue.
2. The method according to claim 1, characterized in that According to the H channel image, the S channel image and the V channel image, a feature image is determined, including: Binarization is performed on the H channel image, the S channel image and the V channel image according to a first preset condition to obtain a first H channel image, a first S channel image and a first V channel image; wherein the first preset condition includes an H channel threshold, a first S channel threshold and a first V channel threshold; Determine a first characteristic image according to the first H channel image, the first S channel image and the first V channel image; Binarization is performed on the S channel image and the V channel image according to a second preset condition to obtain a second S channel image and a second V channel image; wherein the second preset condition includes a second S channel threshold and a second V channel threshold; Determine a second feature image according to the second S channel image and the second V channel image; A union area of the first feature image and the second feature image is determined as the feature image.
3. The method according to claim 2, characterized in that Determining a first characteristic value of the target area image according to the characteristic image includes: Obtaining a feature image corresponding to the target area image at a previous moment, and calculating a first matching degree between the feature image at a previous moment and the current feature image; The first matching degree is filtered to obtain a first eigenvalue of the target area image.
4. The method according to claim 1, characterized in that The method for obtaining the second characteristic value includes: Acquire the target area image at the previous moment, and calculate the second matching degree between the target area image at the previous moment and the current target area image; The second matching degree is filtered to obtain a second eigenvalue of the target area image.
5. The method according to claim 1, characterized in that The third eigenvalue includes a width matching eigenvalue and a height matching eigenvalue, and a method for obtaining the third eigenvalue includes: Obtaining a width value and a height value of the target area image, and calculating an absolute difference in width and an absolute difference in height between the target area image at a previous moment and the current target area image; The width absolute difference and the height absolute difference are respectively subjected to filtering processing to obtain a width matching feature value and a height matching feature value of the target area image.
6. The method according to any one of claims 3 to 5, characterized in that: Identifying a target area image of one or more suspected flame areas in the area image, comprising: Inputting the region image into a pre-trained target recognition network, and outputting one or more suspected flame regions in the region image; A minimum rectangular bounding box image corresponding to the suspected flame area is extracted as a target area image of the suspected flame area.
7. The method according to claim 6, characterized in that The method further comprises: Obtain the regional image of the well at the previous moment; The suspected flame region in the regional image at the previous moment is matched with the suspected flame region in the current regional image to obtain a regional matching set; and the target regional image corresponding to the suspected flame region that does not exist in the regional matching set is deleted.
8. The method according to claim 1, characterized in that The step of determining the flame region in the regional image according to the first eigenvalue, the second eigenvalue, and the third eigenvalue comprises: Determine a feature input vector according to the first eigenvalue, the second eigenvalue and the third eigenvalue; Inputting the feature input vector into a pre-trained classifier, and outputting a flame classification result of the target area image; According to the flame classification result of the target area image, a flame area in the area image is determined.
9. The method according to claim 8, characterized in that The step of determining the flame region in the region image according to the flame classification result of the target region image further comprises: Determining a detection signal value according to the flame classification result; Performing filtering processing on the detection signal value based on a preset weight coefficient to determine the alarm signal value of the target area image; In response to the alarm signal value satisfying the alarm condition, the target area image is determined to be a flame area, and a flame alarm is issued.
10. A coal mine underground flame detection device based on visible light characteristics, characterized in that: include: A first acquisition module is used to acquire a current downhole regional image and identify a target regional image of one or more suspected flame regions in the regional image; A second acquisition module is used to acquire a hue H channel image, a saturation S channel image and a lightness V channel image corresponding to the target area image, and determine a feature image according to the H channel image, the S channel image and the V channel image; A third acquisition module, used to determine a first characteristic value of the target area image according to the characteristic image; A fourth acquisition module, used to determine the second eigenvalue and the third eigenvalue of the target area image according to the color information and size information of the suspected flame area in the target area image; A detection module is used to determine the flame area in the area image according to the first eigenvalue, the second eigenvalue and the third eigenvalue.
Citation Information
Patent Citations
Double-spectrum forest fire disaster monitoring method and double-spectrum forest fire disaster monitoring device based on infrared-visible light image
CN105488941A
A video flame detection method and device
CN109726620A
Smoke and fire identification method and device, computer equipment and storage medium
CN112560657A
Oxidizing furnace flame recognition detection method and device and storage medium
CN116343040A
Fire detection method and system based on infrared visible light image fusion
CN117689568A
