A method based on fire detection IoT perception response

Abstract

The present invention relates to the technical field of forest fire early warning, and discloses a method of sensing and responding based on the Internet of Things of fire detection. The key points of the technical solution include gridding a forest protection area, setting a detector at each grid point, acquiring an image in each grid through the detector, and forming a regional image at a control console, so that an operator of the control console can observe the forest protection area and select an abnormal state block to determine whether there is an abnormality. An alarm can be issued at the control console, and a smoke or fire warning can be issued in the absence of an operator, and the control console can be connected to the Internet of Things with a patrolman to further determine the true situation of the abnormal state block.

Description

Technical Field

[0001] The present invention relates to the technical field of forest fire early warning, and more specifically to a method based on fire detection Internet of Things perception response. Background Art

[0002] Forest reserves are multi-species nature reserves. In dry weather, forest firefighting is required to prevent forest fires. Forest fires are extremely destructive and will cause certain harm and losses to forest ecosystems and humans.

[0003] Disaster prevention is even more important in forest reserves. Generally, forest patrollers are deployed to carry out fire patrols. However, the number of forest patrollers is limited, and the patrol range is limited, which makes forest fire fighting more difficult. Summary of the invention

[0004] In view of the shortcomings of the prior art, the purpose of the present invention is to provide a method based on fire detection Internet of Things perception response to facilitate the detection of forest fires.

[0005] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: A method based on fire detection IoT sensing response, comprising: dividing the forest protection area into grids, and setting detectors at each grid point;

[0006] Taking the location of one of the detectors as the origin, a spatial rectangular coordinate system is established, and the shooting direction and angle of the detectors are adjusted according to the positional relationship between multiple detectors;

[0007] The detector captures and acquires corresponding regional images at a preset rotation angle rate, and forms four image layers from the regional images obtained at four grid points on a single grid, and performs preprocessing and combination to increase the contrast effect, so as to form a complete grid image and upload it to the console;

[0008] The latter region image replaces the former region image and forms a new grid image;

[0009] Extracting feature quantities in the previous region image and the next region image in a preset point selection mode, selecting the abnormal state block, the first position information of the abnormal state block and the first pixel mean value in the abnormal state block from the feature quantities, judging whether the abnormal state block is expanded and whether the first pixel mean value is greater than a threshold, and obtaining the expansion direction with the maximum expansion speed;

[0010] Determine whether there is a patrolman in the grid where the abnormal state block is located;

[0011] If there is no patrolman, a close detector is selected based on the position relationship between the multiple detectors and the first position information, and the shooting direction and angle of the detector are automatically or manually adjusted;

[0012] If there is a patrolman, locate the grid where the patrolman is located, and generate second position information, and adjust the shooting direction and angle of the camera worn by the patrolman by combining the position relationship between the multiple detectors, the first position information and the second position information;

[0013] Acquire multiple comparison images of the abnormal state block captured by the camera at intervals of unit time and second pixel averages of the comparison images, and determine whether the second pixel average is greater than a threshold;

[0014] A line is connected between the camera and the abnormal state block to form a detection line, and a first detection area is formed between the detection line, the expansion direction, and the vertical line between the camera and the expansion direction. One of the detectors in the grid where the abnormal state block is located is controlled to obtain a safety image of the first detection area.

[0015] As a further improvement of the present invention, the detector includes at least two cameras, and the two cameras are spaced 180 degrees apart, and the preset rotation angle rate is one rotation per minute.

[0016] As a further improvement of the present invention, taking the first abnormal state block as the center point, multiple first detection blocks are selected around it, and the third pixel mean in each first detection block is obtained to determine whether the third pixel mean is greater than a threshold, and multiple first detection blocks form an encirclement state of the abnormal state block.

[0017] As a further improvement of the present invention, multiple third pixel means with values ​​greater than the diffusion value are selected and set as diffusion points, and the direction of the line connecting the center point and the diffusion point is set as the positioning direction. Multiple second detection blocks with the diffusion point as the origin and facing the positioning direction are selected, and the multiple second detection blocks form a semi-encirclement state of the diffusion point.

[0018] As a further improvement of the present invention, the enclosing state encloses the abnormal state block in a spherical form, and the semi-enclosing state encloses the diffusion point in a rectangular parallelepiped form.

[0019] As a further improvement of the present invention, a line is selected between a detector in a grid where one of the abnormal state blocks is located and the abnormal state block to form a second detection area, and a detector adjacent to the detector is controlled to obtain a spreading image of the second detection area and obtain a spreading speed.

[0020] As a further improvement of the present invention, the plants in each image layer are framed and positioned, and positioning information is generated; the rainfall amount is detected, and the positioning information is updated in real time.

[0021] As a further improvement of the present invention, the size of the framed plants is screened according to a preset area, and the plant parts not framed in the selection frame are simulated.

[0022] As a further improvement of the present invention, the latter positioning information is compared with the former positioning information to obtain the movement information of the plant, wherein the movement information at least includes the movement direction, and multiple third detection blocks in a direction perpendicular to the movement direction are selected to obtain the current width and widening trend of the debris flow.

[0023] As a further improvement of the present invention, the information sent to the control console is sent to the patrolman's handheld device.

[0024] Beneficial effects of the present invention: In the present invention, the forest protection area is divided into grids, and detectors are set at each grid point. The images in each grid can be obtained through the detectors, and a regional image is formed at the control console, which is convenient for the operator of the control console to observe the forest protection area and select the abnormal state block to determine whether there is an abnormality. An alarm can be issued at the control console, and smoke or fire warnings can be issued in the absence of an operator. In addition, the Internet of Things can be connected with patrol officers to further determine the true situation of the abnormal state block. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 is a system block diagram of the present invention;

[0026] Figure 2 It is the installation method of the detector and the initial orientation of the camera in the present invention. DETAILED DESCRIPTION

[0027] The present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. The same parts are represented by the same reference numerals. It should be noted that the words "front", "rear", "left", "right", "upper" and "lower" used in the following description refer to directions in the accompanying drawings, and the words "bottom surface" and "top surface", "inner" and "outer" refer to directions toward or away from the geometric center of a specific component, respectively.

[0028] Reference Figure 1 and Figure 2 As shown, a method of fire detection IoT sensing response based on this embodiment includes: gridding the forest protection area, and setting a detector at each grid point; the detector includes at least two cameras, and the two cameras are spaced 180 degrees apart, and a rotating part for driving the camera to rotate is set on the detector, and the rotating part includes at least a motor, a gear set and a rotating member, and the motor drives the rotating member to rotate through the gear set, and the camera, PCB, wire, power supply for the camera, etc. are all set inside and outside the rotating member, and a solar panel can also be set to provide continuous power supply;

[0029] The detector also includes a smoke alarm, which can detect fires caused by the detector itself or fires below it, and an anemometer, which provides the function of measuring wind speed and direction;

[0030] Wind speed and wind direction information can be obtained, and the information is sent to and displayed on a console, which includes at least a server and a display screen. If the wind speed is greater than a preset wind speed threshold, the number of first detection blocks in an encircled state or the number of second detection blocks in a semi-encircled state along the wind direction is adjusted with different weights according to the size of the wind speed;

[0031] The greater the wind speed, the greater the number of first detection blocks or second detection blocks acquired along the wind direction, but the enclosed state is still spherical, and the semi-enclosed state is still rectangular.

[0032] Taking the location of one of the detectors as the origin, a spatial rectangular coordinate system is established, and the X-axis, Y-axis and Z-axis coordinates are established. Each detector forms a coordinate (a, b, c). According to the difference in a, b, c data, the height difference of the detector's position is obtained, and the shooting direction of the camera is adjusted according to the height difference of the position. When the camera rotates, the shooting direction and angle of the detector are adjusted according to the position relationship between multiple detectors; the shooting direction and angle are adjusted by the camera itself, and the camera's cross-grid shooting is achieved by the rotating part;

[0033] The detector shoots and acquires the corresponding regional image at a preset rotation angle rate, which is one rotation angle rate per minute, and forms four image layers from the regional images obtained at four grid points on a single grid, and performs preprocessing and combination to increase the contrast effect, deletes the overlapping parts of the four image layers, and retains the non-overlapping parts to form a complete grid image and upload it to the console for display on the console;

[0034] The image of the previous area is replaced by the image of the next area, and a new grid image is formed, so that the grid image is updated in real time to facilitate observation by the operator in front of the control console;

[0035] Extract the feature quantities in the previous regional image and the next regional image by using a preset point selection mode. The preset point selection mode is: the regional image is gridded, each of which is one ten-thousandth of the width of the regional image, and an interval point selection mode is used. The previous regional image and the next regional image are grouped together, and three regional images form two groups. First, half of the grids in the first group of regional images are selected, and the intervals between the grids are set. The other half of the grids in the second group of regional images are selected for the second time, which can reduce the number of comparisons of feature quantities, thereby reducing the amount of processing;

[0036] Select the abnormal state block in the feature quantity, the first position information of the abnormal state block and the first pixel mean in the abnormal state block, judge whether the abnormal state block is expanded and whether the first pixel mean is greater than the threshold, and obtain the expansion direction of the maximum expansion speed; the abnormal state can be: the occurrence of fire or smoke. When the fire or smoke occurs, the pixel mean will change. The pixel standard deviation of the abnormal state block can also be obtained, which can be used to judge whether the fire or smoke occurs. This is a prior art, which also involves the Laplace operator, the pixel standard deviation, etc.

[0037] By judging whether the abnormal state spreads to the first detection block, the third pixel mean can be used to judge and then obtain the spreading direction. After the first detection block is detected, several first detection blocks at the edge of the spreading direction are taken as the center point, and multiple blocks are selected around them for detection to obtain the pixel mean. There is a difference between the pixel means, and the boundaries of each direction of smoke spreading can be obtained, so as to obtain the expansion speed of each spreading direction, that is, the expansion speed of the boundary; smoke and fire warnings can be issued when there is no operator at the console;

[0038] Determine whether there is a patrolman in the grid where the abnormal state block is located;

[0039] If there is no patrolman, the position relationship between the multiple detectors and the first position information are combined to select the detector that is close to it, and automatically or manually adjust the shooting direction and angle of the detector; manual adjustment is a method in which the operator at the console adjusts the position of the camera on the detector to shoot, and can also perform operations such as zooming in and shooting, which can more conveniently understand the situation of the abnormal state block;

[0040] Automatic adjustment is the cooperation between the rotating part and the camera to change the shooting angle and direction. Through the positional relationship between the detectors and the first position information, a detector closest to the first position information is selected, or a camera that can directly and clearly shoot the abnormal state block is generally selected. The detector closest to the first position information is used for shooting to obtain the latest abnormal state block information and send it to the console for recording. The operator can determine whether the abnormal state block really has an abnormal situation, and can also determine the real situation of the abnormal state block in the absence of an operator;

[0041] If there is a patrolman, the grid where the patrolman is located is located, and the second position information is generated. The position of the patrolman is preliminarily located in the regional image, and the position relationship between multiple detectors, the first position information and the second position information are combined to adjust the shooting direction and angle of the camera worn by the patrolman; the shooting direction adjusted by the camera is small because it needs to be adjusted by itself. According to the second position information, if the patrolman is in an uphill position, the first position information of the abnormal state block is higher than the position of the patrolman, then the camera needs to adjust the shooting direction and angle diagonally upward;

[0042] The camera is worn on the patrolman's waist. If it is worn on the head, the movement of the patrolman's head will have a greater impact on the shooting angle and direction of the camera.

[0043] Information sent to the console is sent to the patrol officer’s handheld device for easy interaction.

[0044] Acquire multiple comparison images of the abnormal state block captured by the camera at interval unit time and the second pixel mean of the comparison images to determine whether the second pixel mean is greater than a threshold value; the interval unit time can be set to 1 second, and the comparison between the second pixel mean and the threshold is used to further determine whether an abnormal state occurs, thereby improving the accuracy of the judgment.

[0045] A line is connected between the camera and the abnormal state block to form a detection line. The detection line, the expansion direction and the vertical line between the camera and the expansion direction form a first detection area. One of the detectors in the grid where the abnormal state block is located is controlled to obtain a safety image of the first detection area. The first detection area is used as a safety area to determine whether the patrolman's safety will be affected when shooting at close range, so as to remind the patrolman and sound an alarm on the patrolman's handheld device.

[0046] The safety range can be determined by determining whether the area of ​​the first detection area in the safety image decreases or increases, whether the pixel mean changes, etc.

[0047] In the case of multiple patrollers, there may be multiple detection lines, each forming a detection area.

[0048] Taking the first abnormal state block as the center point, multiple first detection blocks are selected around it, and the third pixel average value in each first detection block is obtained to determine whether the third pixel average value is greater than a threshold value. Multiple first detection blocks form a surrounding state of the abnormal state block. In the surrounding state, the number of first detection blocks can be expanded, and each direction of expansion can be obtained more accurately, and the overall width of the smoke can also be obtained;

[0049] The encirclement state surrounds the abnormal state block in a spherical form. The selection of the spherical form can detect the abnormal state below when smoke occurs in the air, that is, on a tree, to prevent missed detection. In addition, the spherical form involves multiple directions, which can include all directions of 360 degrees.

[0050] Select multiple third pixel means whose values ​​are greater than the diffusion value, and the diffusion value is a preset value. Under this value, it is predicted that the smoke or fire will be more intense. If the third pixel mean in the first detection block located at the edge is less than the diffusion value, it can be determined that it is located in the diffusion range of the smoke, indicating that the intensity of the smoking point below it has not increased. It is set as the diffusion point, and the direction of the line connecting the center point and the diffusion point is the positioning direction. Select multiple second detection blocks with the diffusion point as the origin and facing the positioning direction. The multiple second detection blocks form a semi-enclosed state for the diffusion point. The semi-enclosed state surrounds the diffusion point in the form of a rectangular parallelepiped, which can be used to determine how large the diffusion range of the smoke is when it is emitted from the bottom and extends in the diffusion direction. The semi-enclosed state subtracts the detection range below the diffusion point. Compared with the spherical form, due to the influence of wind force, in order to improve the detection accuracy, the detection range at the corner is increased to judge whether the second detection block at the corner is in the diffusion range or the top of the outer flame of the flame.

[0051] A line is selected between a detector in a grid where one of the abnormal state blocks is located and the abnormal state block to form a second detection area. A detector adjacent to the detector is controlled to obtain a spread image of the second detection area, and the spread speed is obtained to determine the safe range between the smoke or fire and the detector. Firefighting equipment, such as a fire extinguisher, can be installed on the detector to realize automatic spraying to spray the direction of the spread. It can also be a rainwater collection device to perform watering operations.

[0052] The plants in each image layer are framed and located, and the location information is generated; the rainfall amount is detected through the rainfall sensor installed on the detector, and the location information is updated in real time to determine whether mudslides and other situations occur.

[0053] The size of the selected plants is filtered according to the preset area, and the parts of the plants that are not selected in the selection box are simulated. Because the plants are tall, the tops of the plants are not captured, so they are simulated according to the outer diameter of the plant trunks and the type of trees.

[0054] The latter positioning information is compared with the previous positioning information to obtain the movement information of the plant. The movement information at least includes the moving direction. The moving direction can be obtained by comparing the selected position, and the moving speed can be obtained based on the moving distance, so as to determine whether a mudslide will occur or the degree of danger of the mudslide. Multiple third detection blocks in the direction perpendicular to the moving direction are selected to obtain the current width and widening trend of the mudslide. If the moving direction of the plant is obliquely downward, multiple third detection blocks in the horizontal direction are selected to determine whether the plant in the third detection block moves, so as to splice them to obtain the horizontal width of the mudslide and whether it will have a widening trend. The flow rate of the mudslide can be estimated based on the moving speed of the plant.

[0055] The above are only preferred embodiments of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A method based on fire detection IoT perception response, characterized in that: include: The forest protection area is divided into grids, and detectors are set at each grid point; Taking the location of one of the detectors as the origin, a spatial rectangular coordinate system is established, and the shooting direction and angle of the detectors are adjusted according to the positional relationship between the multiple detectors; The detector captures and acquires corresponding regional images at a preset rotation angle rate, and forms four image layers from the regional images obtained at four grid points on a single grid, and performs preprocessing and combination to increase the contrast effect, so as to form a complete grid image and upload it to the console; The latter region image replaces the former region image and forms a new grid image; Extracting feature quantities in the previous region image and the next region image in a preset point selection mode, selecting the abnormal state block, the first position information of the abnormal state block and the first pixel mean value in the abnormal state block from the feature quantities, judging whether the abnormal state block is expanded and whether the first pixel mean value is greater than a threshold, and obtaining the expansion direction with the maximum expansion speed; Determine whether there is a patrolman in the grid where the abnormal state block is located; If there is no patrolman, a close detector is selected based on the position relationship between the multiple detectors and the first position information, and the shooting direction and angle of the detector are automatically or manually adjusted; If there is a patrolman, locate the grid where the patrolman is located, and generate second position information, and adjust the shooting direction and angle of the camera worn by the patrolman by combining the position relationship between the multiple detectors, the first position information and the second position information; Acquire multiple comparison images of the abnormal state block captured by the camera at intervals of unit time and second pixel averages of the comparison images, and determine whether the second pixel average is greater than a threshold; A line is connected between the camera and the abnormal state block to form a detection line, and a first detection area is formed between the detection line, the expansion direction, and the vertical line between the camera and the expansion direction. One of the detectors in the grid where the abnormal state block is located is controlled to obtain a safety image of the first detection area.

2. According to claim 1, a method based on fire detection IoT perception response is characterized in that: The detector includes at least two cameras, and the two cameras are spaced 180 degrees apart, and the preset rotation angle rate is one circle per minute.

3. The method of fire detection IoT sensing response according to claim 2 is characterized in that: Taking the first abnormal state block as the center point, multiple first detection blocks are selected around it, and the third pixel mean in each first detection block is obtained to determine whether the third pixel mean is greater than a threshold. The multiple first detection blocks form a surrounding state for the abnormal state block.

4. The method of fire detection IoT sensing response according to claim 3 is characterized in that: A plurality of third pixel means whose values ​​are greater than the diffusion value are selected and set as diffusion points, and the direction of the line connecting the center point and the diffusion point is set as the positioning direction, and a plurality of second detection blocks with the diffusion point as the origin and facing the positioning direction are selected, and the plurality of second detection blocks form a semi-encirclement state of the diffusion point, and the wind speed and wind direction information is obtained, and the number of the first detection blocks in the encirclement state along the wind direction or the number of the second detection blocks in the semi-encirclement state is adjusted with different weights according to the size of different wind speeds.

5. The method of fire detection IoT sensing response according to claim 4 is characterized in that: The encircling state encircles the abnormal state block in a spherical form, and the semi-encircling state encircles the diffusion point in a rectangular parallelepiped form.

6. The method of fire detection IoT sensing response according to claim 1 is characterized in that: A line is selected between a detector in a grid where one of the abnormal state blocks is located and the abnormal state block to form a second detection area, and a detector adjacent to the detector is controlled to obtain a spreading image of the second detection area and obtain a spreading speed.

7. The method of fire detection IoT sensing response according to claim 1 is characterized in that: The plants in each image layer are framed and located, and the location information is generated; the rainfall amount is detected and the location information is updated in real time.

8. The method of fire detection IoT sensing response according to claim 7 is characterized in that: The size of the selected plants is filtered according to the preset area, and the parts of the plants not selected in the selection box are simulated.

9. The method of fire detection IoT sensing response according to claim 7, characterized in that: The latter positioning information is compared with the former positioning information to obtain the movement information of the plant, wherein the movement information at least includes the movement direction, and a plurality of third detection blocks in a direction perpendicular to the movement direction are selected to obtain the current width and widening trend of the debris flow.

10. A method based on fire detection IoT perception response according to any one of claims 1 to 9, characterized in that: Information sent to the control console is sent to the patrol officer's handheld device.

Images