Forest fire detection system

Abstract

The present invention relates to a forest fire detection system using real- and thermal-imaging camera modules to accurately detect fires, thereby allowing detection and rapid responses to in cases of forest fires.

Description

Forest fire detection system

[0001] The present invention relates to a technology related to a system for detecting fire.

[0002] The convergence of information and communication technologies (ICTs), such as artificial intelligence (AI), deep learning, and the Internet of Things (IoT), which are driving the Fourth Industrial Revolution, with video security is gaining momentum. While CCTV was previously used solely for surveillance purposes, its use has recently expanded to include AI technology, enabling it to be used as a device to prevent accidents.

[0003] Among these technologies, the one that's gaining attention is the fire detection system. When a pre-determined flame or spark is detected, the system provides the detected flame to monitoring personnel, enabling efficient monitoring.

[0004] Currently, many developed fire detection systems have problems in accurately detecting fires and excluding false fires.

[0005] Accordingly, when a fire breaks out in the mountains, the exact occurrence and progression of the fire cannot be accurately predicted, making it difficult to extinguish the fire in its early stages, resulting in damage.

[0006] [Prior Art Document] Republic of Korea Patent Publication No. 10-2023-0147831 (Publication Date: October 24, 2023)

[0007] The present invention aims to solve a problem that occurs when a forest fire cannot be extinguished in the early stages due to the inability to detect fire and exclude false fires using a real-image camera module and a thermal-image camera module.

[0008] The problems to be solved by the present invention are not limited to the problems mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art from the description below.

[0009] The forest fire detection system of the present invention for achieving the above-mentioned problem comprises a camera unit including a real-image camera module for photographing a certain area to generate a first image, and a thermal-image camera module for photographing the same area as the photographing area photographed by the real-image camera module to generate a second image, an object classification and extraction module for receiving the first image image and the second image image from the camera unit, including a preset reference flame object, detecting a first image-extracted flame object matching the preset reference flame object in the first image image, detecting an area that does not match the reference flame object as a first image-extracted background object, detecting a second image-extracted flame object matching the preset reference flame object in the second image image, and detecting an area that does not match the reference flame object as a second image-extracted background object, and a temperature extraction module for extracting a temperature of a second image-extracted flame object located at a position overlapping the first image-extracted flame object detected by the object classification and extraction module, The main server unit includes a fire detection module that generates a fire signal when at least one temperature among the temperatures extracted from the temperature extraction module is higher than a preset fire reference temperature, and generates a non-fire signal when both the temperature extracted from the first image extraction flame object and the temperature extracted from the second image extraction flame object are lower than the fire reference temperature.

[0010] The main server unit includes a fire location extraction module that extracts location information of the area where the first and second video images were captured from the first and second video images when a fire signal is generated from the fire detection module.

[0011] The main server unit may include a notification alarm module that outputs a preset message when a fire signal is generated, and a shelter location confirmation module in which location information of at least one shelter is pre-stored.

[0012] The fire evacuation app may be installed, and a wireless terminal unit that communicates with the main server unit and receives messages from the main server unit's notification alarm module may be further included.

[0013] The fire evacuation app includes a map image, and when the wireless terminal is turned on after receiving a message from the alarm module, the location of the wireless terminal and the location of the shelter received from the shelter location confirmation module can be displayed on the map image.

[0014] The main server unit performs data communication with a satellite and receives data observed from the satellite to produce vegetation data including a vegetation index and canopy temperature, receives brightness temperature or backscattering coefficients observed through a microwave installed in the satellite to produce soil moisture data, receives wind data of a location corresponding to the location information extracted from the fire location extraction module while performing data communication with the satellite, receives vegetation data and soil moisture data produced by the calculation module, and receives a variable data reception module, generates a plurality of movement paths from the location of the wireless terminal unit to the location of the shelter and transmits them to the wireless terminal unit, and calculates a soil spread possibility index indicating the degree of possibility that a fire in a set section will spread by inputting the vegetation data and soil moisture data received from the variable data reception module into a vegetation data / soil moisture data calculation formula. A fire spread path extraction module may be included that extracts an expected path along which a fire spreads from the fire location extraction module by calculating a wind spread possibility index by performing a vector product operation on the location information extracted from the fire location extraction module and calculating an inner product of the soil spread possibility index and the wind spread possibility index.

[0015] The main server unit can detect a movement path among multiple movement paths generated by the movement path generation module that has a low wind spread possibility index, does not overlap with the expected path generated by the fire spread path extraction, and has the shortest distance from the wireless terminal unit to the shelter as a fire avoidance movement path and transmit it from the wireless terminal unit.

[0016] The present invention enables long-distance fire detection and accurate detection and identification of fire hazards through an object algorithm, thereby enabling 24-hour continuous confirmation of the occurrence of a fire. Furthermore, the present invention divides a video image into a plurality of segments, and then pulls the segments in which objects are detected from the segmented regions to perform secondary segmentation to identify the characteristics of the objects. In addition, the present invention allows the main server unit to receive fire occurrence location data transmitted from the camera unit, calculate the location of a preset shelter, movement path data between the location of a fire occurrence and the shelter, calculate the movement time according to the movement path, and calculate soil moisture data, vegetation data, and wind direction data according to the movement data path, thereby providing people near the location of a fire occurrence with a safe path to a shelter by avoiding the path through which the fire spreads.

[0017] Figure 1 is a usage diagram of a forest fire detection system according to one embodiment of the present invention.

[0018] FIG. 2 is a block diagram of a forest fire detection system according to one embodiment of the present invention.

[0019] Figure 3 is a perspective view of the camera section of Figure 2.

[0020] FIG. 4 is a drawing showing a first video image and a second video image generated by the camera unit of the present invention.

[0021] Figure 5 is a diagram showing a state in which the object classification extraction module classifies the first image extraction background object and the first image extraction flame object from the first image.

[0022] Figure 6 is a drawing showing a state in which the object classification extraction module classifies the second image extraction background object and the second image extraction flame object from the second image.

[0023] FIG. 7 is a drawing showing a first segmented image obtained by segmenting the first image using the region segmentation module of FIG. 2 and an enlarged image of an area including a first image extraction flame object in the first segmented image.

[0024] FIG. 8 is a drawing showing a second segmented image obtained by segmenting a second image using the region segmentation module of FIG. 2 and an enlarged image of an area including a second image extraction flame object in the second segmented image.

[0025] FIG. 9 is a diagram showing a state in which the composite image generation module of FIG. 2 synthesizes the first segmented image of FIG. 7 and the second segmented image of FIG. 8.

[0026] Figure 10 is a drawing showing the fire evacuation app installed in the wireless terminal of Figure 2 and the map provided by the fire evacuation app.

[0027] Figure 11 is a diagram showing wind information, vegetation information, soil moisture information, etc. included in the main server section of Figure 3.

[0028] Figure 12 is an operation flow chart of a forest fire detection system according to one embodiment of the present invention.

[0029] Figures 13 to 16 are drawings showing the operating state of a forest fire detection system according to one embodiment of the present invention.

[0030] The advantages and features of the present invention, as well as the methods for achieving them, will become clearer with reference to the embodiments described in detail below together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below and may be implemented in various different forms. These embodiments are provided solely to ensure that the disclosure of the present invention is complete and to fully inform those skilled in the art of the invention of the scope of the invention.

[0031] The scope of the present invention can be defined by the claims and the supporting description. Furthermore, like reference numerals refer to like elements throughout the specification.

[0032] Hereinafter, with reference to FIGS. 1 to 16, a forest fire detection system according to an embodiment of the present invention will be described in detail. However, to ensure that the description of the present invention is concise and clear, the forest fire detection system will be described in general terms with reference to FIG. 1. Subsequently, with reference to FIGS. 2 to 11, components constituting the forest fire detection system will be described in detail. Furthermore, with reference to the described contents and FIGS. 12 to 16, the operation of the forest fire detection system according to an embodiment of the present invention will be described in detail.

[0033] Figure 1 is a usage diagram of a forest fire detection system according to one embodiment of the present invention.

[0034] Hereinafter, referring to FIG. 1, the forest fire detection system (1) has a main server unit (20) that accumulates data based on artificial intelligence technology, learns the accumulated data, and then detects newly introduced and generated objects in the camera unit (10) based on the accumulated data. Then, the detected objects are analyzed and the corresponding data is transmitted to the wireless terminal unit (30). That is, the forest fire detection system (1) processes the first video image (A10) and the second video image (A20) transmitted from the camera unit (10) by the main server unit (20), compares the extracted objects with previously stored data, detects the characteristics of the objects included in the first video image (A10) and the second video image (A20), and transmits the characteristics to the wireless terminal unit (30) so that the user can utilize them as useful information. In addition, the forest fire detection system (1) of the present invention pulls the part where an object is detected to check whether there is a fire, and if it is checked as a fire, it divides it into multiple divided areas, and then pulls the area where an object is detected from the divided areas to divide it again for a second time and determines the characteristics of the object. In addition, the forest fire detection system (1) receives the fire occurrence location data transmitted by the main server unit (20) from the camera unit (10), calculates the location of a preset shelter, the movement path data between the fire occurrence location and the shelter, calculates the movement time according to the movement path, and calculates soil moisture data, vegetation data, and wind direction data according to the time for each movement data path.

[0035] The forest fire detection system (1) calculates the characteristics of the object identified in this way and various data to provide the optimal route to a shelter to the wireless terminal (30) preset in the main server (20) adjacent to the location where the fire occurred, thereby enabling people to prepare safely.

[0036] Hereinafter, components constituting the forest fire detection system will be specifically described with reference to FIGS. 2 to 11.

[0037] FIG. 2 is a block diagram of a forest fire detection system according to an embodiment of the present invention, FIG. 3 is a perspective view of a camera unit of FIG. 2, FIG. 4 is a diagram showing a first image image and a second image image generated by the camera unit of the present invention, FIG. 5 is a diagram showing a state in which an object classification extraction module classifies a first image extraction background object and a first image extraction flame object in a first image image, FIG. 6 is a diagram showing a state in which an object classification extraction module classifies a second image extraction background object and a second image extraction flame object in a second image image, and FIG. 7 is a diagram showing a first segmented image image obtained by segmenting a first image image by a region segmentation module of FIG. 2 and an enlarged image of an area including a first image extraction flame object in the first segmented image image. And FIG. 8 is a drawing showing a second segmented image obtained by segmenting a second image by the region segmentation module of FIG. 2 and an enlarged image of an area including a second image extracted flame object in the second segmented image image, FIG. 9 is a drawing showing a state in which the composite image generation module of FIG. 2 synthesizes the first segmented image of FIG. 7 and the second segmented image of FIG. 8, and FIG. 10 is a drawing showing a fire evacuation app installed in the wireless terminal of FIG. 2 and a map provided by the fire evacuation app. And FIG. 11 is a drawing showing wind information, vegetation information, soil moisture information, etc. included in the main server unit of FIG. 3.

[0038] A forest fire detection system (1) such as this includes a camera unit (10), a main server unit (20), and a wireless terminal unit (30) as components.

[0039] The camera unit (10) photographs a certain area from a place where the user has installed it, such as a building or a forest. Then, it photographs the certain area and generates a thermal image, that is, a first image image (A10), and a real image, that is, a second image image (A20), corresponding to the area. As illustrated in FIG. 3, the camera unit (10) includes a case module (100), a real image camera module (110) installed inside the case module (100) to photograph the outside and generate the first image image (A10), and a thermal image camera module (120) installed inside the case module (100) next to the real image camera module (110) to photograph the outside and generate the second image image (A20). This camera unit (10) can detect a fire source measuring 70 cm (width) X 70 cm (length) X 5 cm (height) at a straight-line distance of 5,000 m within 9 seconds, and can detect a fire source in a wide area reaching 10,000 m in the foreground. The first video image (A10) captured by this camera unit (10) is an image generated by a CCD image sensor, and can be an image captured of a mountain as shown in (a) of FIG. 4, and an image captured when a fire broke out on the mountain as shown in (b) of FIG. 4. In addition, the second video image (A20) is a thermal image that visualizes infrared rays emitted by the subject. In other words, the second video image (A20) is an image that visualizes the heat of an object and an area, and allows for identifying a high heat source such as a fire through heat distribution. At this time, the thermal image may be a thermal image taken of a mountain as shown in (c) of FIG. 4, and a thermal image taken when a fire broke out on the mountain as shown in (d) of FIG. 4.

[0040] Both the first video image (A10) and the second video image (A20) are images captured from the same area. The camera unit (10) transmits the first video image (A10) and the second video image (A20) in real time to the main server unit (20) connected to a wireless network.

[0041] The position of such a camera unit (10) can be controlled by data transmitted from the main server unit (20).

[0042] Accordingly, the camera unit (10) can focus on shooting areas with a high risk of fire depending on the local situation. In addition, it can receive latitudes and longitudes transmitted from the main server unit (20) for up to 10 locations and monitor them in rotation from the farthest to the nearest location. At this time, the camera unit (10) sets a number of arbitrary reference locations in the surveillance area, accurately receives the latitude and longitude for each specific set reference location, compares the measured distance of the fire occurrence with the actual distance, identifies errors in the measured data, corrects the occurrence location, and shoots a certain area to generate a first video image (A10) and a second video image (A20).

[0043] The main server unit (20) applies YOLO's object detection technology that detects the locations of multiple objects within an image, collects image data of fire and non-fire, labels and preprocesses (size, normalizes) data as fire and non-fire, converts the data based on the YOLO model, builds a deep learning model based on the YOLO algorithm, and performs learning. In addition, the main server unit (20) processes the first video image (A10) and the second video image (A20) transmitted from the camera unit (10), detects flame objects in the first video image (A10) and the second video image (A20), and transmits them to the wireless communication unit (30).

[0044] This main server unit (20) includes an object classification extraction module (201), a temperature extraction module (202), a fire detection module (203), a fire location extraction module (204), a notification alarm module (205), and a shelter location confirmation module (206).

[0045] The object classification extraction module (201) receives a first image (A10) and a second image (A20) from the camera unit (10), including a preset reference background object and a preset reference flame object. The object classification extraction module (201) more accurately classifies the background object (A12, A22) and the flame object (A11, A11) in the received first image (A10) and second image (A20), including an object detection body (2011), a region segmentation body (2012), and an object alignment body (2013).

[0046] Hereinafter, the object classification and extraction module (201) through the object detection body (2011) and the region segmentation body (2012) detects the image extraction base object and the image extraction flame object in the first image image (A10) and the second image image (A20) will be described in detail.

[0047] The object detector (2011) performs frame segmentation to divide the video into frames for fire detection, flame object detection to accurately extract the background by removing dynamic elements (such as moving objects other than fire) from the video, and color space conversion to convert from RGB to color spaces such as HSV and LAB to utilize color information. In addition, the object detector (2011) detects an object that matches a preset reference flame object in the first video image (A10), calculates a matching value between the detected object and the reference flame object, and classifies the detected object as a flame object if the matching value is greater than or equal to the reference value. In addition, the object that is not classified as a flame object or an object that matches a preset reference background object is detected, and calculates a matching value between the detected object and the reference background object, and classifies the detected object as a background object if the matching value is greater than or equal to the reference value. For example, the object detector (2011) detects a first image extraction flame object (A11) that matches a preset reference flame object in the first image image (A10), as illustrated in (a) of FIG. 5. In addition, as illustrated in (a) of FIG. 5, the object detector (2011) detects a first image extraction background object (A12) that matches a preset reference background object in the first image image (A10) or an object that does not match the reference flame object as the first image extraction background object (A12). In addition, the object detector (2011) detects a second image extraction flame object (A21) that matches a preset reference flame object in the second image image (A20), as illustrated in (a) of FIG. 6. And as shown in (a) of Fig. 6, the second image extraction background object (A22) that matches the preset reference background object in the second image (A20) or the object that does not match the reference flame object is detected as the second image extraction background object (A22). At this time, the reference background object and the reference flame object stored in the object classification extraction module (201) are not limited to the mountain shape and flame shape shown in Figs. 5 and 6, and may be objects including mountains and flames of various shapes.

[0048] The region segmenter (2012) divides the first image image (A10) and the second image image (A20) into a plurality of segmented regions. For example, the region segmenter (2012) divides the first image image (A10) into 3x3 regions to generate a first segmented image image (A13), and divides the second image image (A20) into 3x3 regions to generate a second segmented image image (A23). At this time, the first segmented image image (A13) and the second segmented image image (A23) include the first segmented region (①) to the ninth segmented region (⑨), as illustrated in FIGS. 7 and 8. In addition, the region segmenter (2012) enlarges the region including the object detected by the object detector (2011), and then secondarily divides it into 3x3, generating a part of the first segmented image image (A13) as a first segmented partial enlarged image (A14), and generating a part of the second segmented image image (A23) as a second segmented partial enlarged image (A24).

[0049] The object alignment body (2013) aligns the first image-extracted flame object (A11) and the first image-extracted background object (A12) classified from the real image (A10) and the second image-extracted flame object (A21) and the second image-extracted background object (A22) classified from the thermal image (A20) so that the first image-extracted flame object (A11) and the second image-extracted flame object (A21) overlap.

[0050] The object classification extraction module (201) enables the position of the first image extraction flame object to be accurately identified in the first image (A10), and the temperature extraction module (202) enables the temperature of the second image extraction flame object (A21) to be accurately extracted overlapping the first image extraction flame object.

[0051] The temperature extraction module (202) extracts the temperature of the second image extraction flame object (A22) located at a position overlapping the first image extraction flame object (A11) detected by the object classification extraction module (201). Furthermore, the temperature extraction module (202) extracts the maximum and minimum temperature values ​​from the second image extraction flame object (A22) through temperature analysis and a set threshold value, and can determine whether a fire has occurred by comparing the temperature values ​​with an arbitrarily set preset fire reference temperature when the temperature rises due to the fire. For example, the temperature extraction module (202) can extract the temperature of the first split-part enlarged area (⑤-①) of the second split-part enlarged image (A24) as 25°C, can extract the temperature of the second split-part enlarged area (⑤-②) as 120°C, and can extract the temperature of the third split-part enlarged area (⑤-③) as 25°C, as illustrated in FIG. 9. And the temperature of the fourth divided partial enlargement area (⑤-④) can be extracted as 110℃, the temperature of the fifth divided partial enlargement area (⑤-⑤) can be extracted as 200℃, and the temperature of the sixth divided partial enlargement area (⑤-⑥) can be extracted as 25℃. And the temperature of the seventh divided partial enlargement area (⑤-⑦) can be extracted as 114℃, the temperature of the eighth divided partial enlargement area (⑤-⑧) can be extracted as 132℃, and the temperature of the ninth divided partial enlargement area (⑤-⑨) can be extracted as 25℃.

[0052] This object classification extraction module (201) includes an object detection body (2011), a region segmentation body (2012), and an object alignment body (2013), so as to accurately determine whether a fire has occurred, and if a fire is determined, so as to accurately determine the location where the fire occurred. In addition, the temperature extraction module (202) enables accurate determination of the temperature of each part of the flame object identified by the object classification extraction module (201).

[0053] The fire detection module (203) generates a fire signal when at least one of the temperatures extracted from the temperature extraction module (202) is higher than the preset fire reference temperature. In addition, if both the temperature extracted from the first image extraction flame object (A11) and the temperature extracted from the second image extraction flame object (A22) are lower than the fire reference temperature, a non-fire signal can be generated. For example, as illustrated in FIG. 9, the fire detection module (230) generates a non-fire signal when all temperatures of the first split-part enlarged area (⑤-①) to the first split-part enlarged area (⑤-①) of the second split-part enlarged image (A24) are lower than the fire reference temperature, whereas it can generate a fire signal when the temperature of any one of the first split-part enlarged area (⑤-①) to the first split-part enlarged area (⑤-①) is higher than the fire reference temperature.

[0054] When a fire signal is generated from the fire detection module (203), the fire location extraction module (204) can extract location information of the area where the first image (A10) and the second image (A20) were captured from the first image (A10) and the second image (A20). For example, the location information of the first image extraction flame object (A11) or the second image extraction flame object (A21) can be extracted through the location information of the camera unit (10) installed in a certain area, the location area captured by the camera unit (10), and the location information identified by the object classification extraction module described above.

[0055] The notification alarm module (205) can output a preset message when a fire signal is generated from the fire detection module (203). Here, the message can be a message having the content 'A fire has occurred at location xx, so turn on the fire evacuation app' to the wireless terminal unit (30).

[0056] The shelter location confirmation module (206) includes location information of at least one shelter (B). The shelter location confirmation module (206) may have location information of a first shelter and a second shelter located near a first location, and location information of a third shelter and a fourth shelter located near a second location, stored therein.

[0057] The wireless terminal unit (30) transmits and receives data wirelessly with the main server unit (20), and becomes a terminal on which various applications can be installed. As illustrated in (a) of FIG. 10, a message app (310) and a fire evacuation app (320) can be installed on the wireless terminal unit (30). Here, the fire evacuation app (320) includes a map image (321), and when the wireless terminal unit (30) is turned on after receiving a message from the alarm module (205), the location of the wireless terminal unit (30) and the location of the shelter (B) received from the shelter location confirmation module (206) can be displayed on the map image (321), as illustrated in (b) of FIG. 10. In addition, the fire evacuation app (320) can be operated by a signal transmitted from the main server unit (20).

[0058] The map image (321) shows the current location (C) of the wireless terminal unit (30), a shelter (B) located near the area where a fire has occurred, and various movement routes (D) from the location of the wireless terminal unit (30) to the shelter (B). More specifically, as shown in (b) of FIG. 10, various movement routes such as a first movement route (D1), a second movement route (D2), a third movement route (D3), and a fourth movement route (D4) may appear. In addition, when the wireless terminal unit (30) receives data corresponding to a fire avoidance movement route transmitted from the main server unit (20), the fire evacuation app (320) can provide the user using the wireless terminal unit (30) with an optimal fire avoidance movement route (E) to a shelter through the data transmitted from the main server unit (20). A detailed description of the fire avoidance movement path (E) data received by the wireless terminal unit (30) and the fire avoidance movement path (E) displayed on the wireless terminal unit (30) will be provided later.

[0059] The main server unit (20) may further include a calculation module (207), a variable data receiving module (208), a movement path generation module (209), a fire spread path extraction module (210), and a database module (211) as components to provide the fire evacuation app (320) with an optimal path to avoid fire, i.e., a fire avoidance movement path (E).

[0060] The operation module (207) communicates with a satellite (F) and receives data observed from the satellite (F) to produce vegetation data (H) including a vegetation index and canopy temperature. In addition, the operation module (207) produces soil moisture data (I) by receiving brightness temperature or backscattering coefficients observed through a microwave installed in the satellite (F).

[0061] The variable data receiving module (208) communicates with a satellite (F) and receives wind data (G) of a location corresponding to the location information extracted from the fire location extraction module (204). In addition, it receives vegetation data (H) and soil moisture data (I) produced by the calculation module (207).

[0062] The movement path generation module (209) generates a plurality of movement paths (D) from the location of the wireless terminal unit (30) to the location of the shelter and transmits them to the wireless terminal unit (30). At this time, the plurality of movement paths (D) may be paths formed by combining various passageways based on a preset passageway.

[0063] The fire spread path extraction module (210) receives wind data (G) received from the variable data receiving module (207) and vegetation data (H) and soil moisture data (I) produced from the calculation module (207).

[0064] The vegetation data (H) and soil moisture data (I) can be calculated using a preset formula to extract the predicted path (J) along which the fire will spread. For example, if a northwesterly wind blows at a point where a fire has broken out, the fire spread path extraction module (210) can predict that the fire will follow the northwesterly wind and spread to the northwest, and can extract the predicted path (J) generated by blocking and bypassing the fire spread path according to the vegetation data (H) and soil moisture data (I) from the predicted path. Here, the wind data (G) can be wind data provided by the Korea Meteorological Administration. As shown in (a) of FIG. 11, the wind data (G) is data that reflects direction changes over time, and when reflected in the extraction of the predicted path, the reliability of the extracted predicted path can be increased. Such wind data (G) can be received in real time from the Korea Meteorological Administration. Additionally, vegetation data (H) is the moisture content of vegetation using data observed from satellites to determine the degree of dryness of vegetation covering the ground surface.

[0065] This vegetation data (H) is information about the plant population growing on the ground surface, as illustrated in (b) of Fig. 11. This vegetation data (H) indirectly contains information about the vegetation of a certain area by using data observed from artificial satellites. In particular, since the vegetation covering the ground surface contains moisture for growth, the vegetation data (H) can be used to predict the risk of forest fires and the possibility of fire movement paths by identifying the degree of drying of the vegetation, along with information about soil moisture, which will be described later. This vegetation data (H) may include a vegetation index, canopy temperature, etc. In this case, the vegetation index is an index that emphasizes the characteristic information of the vegetation by using signals in wavelengths sensitive to vegetation observed from artificial satellites and signals in wavelengths that are not sensitive to vegetation. Examples of such indexes include a normalized differential vegetation index (NDVI) and an enhanced vegetation index (EVI). Here, the normalized differential vegetation index (NDVI) is an index that shows a high value in the near infrared region where the reflectance is high, and a value close to 0 in the soil where the difference between the two reflectance values ​​is small. Accordingly, the vegetation data (H) can be expressed as a formula that includes the vegetation temperature (Tc). For example, the vegetation data (H) is T C =[T R (θ) n -(1-f)T S N ] / f 1 / n It can be expressed as a formula. Here, T C is the vegetation temperature (K), T R : The directional surface temperature according to the view angle (θ). And the view angle is T R Silver T R={[T B (θ) n -(1-ε(θ)T n SKY ] / ε(θ)} 1 / n becomes. Here, T B (θ) is the directional brightness temperature measured by the satellite at a specific field of view (θ), and ε(θ) is the directional emissivity measured at the field of view θ. And T SKY is the hemispherical melting point (=2.7K), and T S is the soil surface temperature n=(C2 / λT O )*(1+1 / exp(C2 / λT O )-1). Here, λ is the wavelength of the satellite-mounted sensor, and C2 is 1.4388*10 4 It becomes μmK. T O is the brightness temperature input to the Plank function. f=1-exp(-0.5LAI / cosΨ). Here, LAI: Leaf Area Index, Ψ is the view zenith angle at which LAI is measured.

[0066] As can be seen from the equation, in order to calculate the vegetation temperature (Tc), the directional surface temperature (T) according to the view angle R ), soil surface temperature (T s ) are required. Directional surface temperature (T) according to view angle R ) is calculated as follows, where T B (θ) is the directional brightness temperature measured by a satellite at a specific field of view θ. And ε(θ) is the directional emissivity measured at field of view θ, which represents the efficiency of energy emission from the surface of an object during thermal radiation. And T SKYThe celestial hemispherical melting point is applied, which is the blackbody radiation amount of 2.7K that fills the universe. Here, n can be calculated according to the above equation using the wavelength (λ) of the sensor mounted on the satellite that observed the brightness temperature and the Plank function. T0 is the brightness temperature input to the Plank function, and in the case of the Soil Moisture Active Passive (SMAP) satellite, n = 1 can be applied in the range of brightness temperature TO = 190 ~ 315K measured at a wavelength of 21 cm, i.e., λ = 210000 μm.

[0067] Soil moisture data refers to the amount of water contained in soil. This soil moisture data can be expressed as a percentage of the total volume of soil, water, and air. This soil moisture data can be produced on a global scale via satellite. For example, soil moisture data can be obtained using brightness temperature acquired through a passive microwave sensor installed on a satellite, and backscattering coefficients acquired through an active microwave sensor installed on a satellite.

[0068] These soil moisture data can indirectly confirm the moisture contained in the soil, and can be used as information on the combustibility of vegetation near the ground surface, along with the vegetation data described above.

[0069] The movement path generation module (209) generates multiple movement paths from the location of the wireless terminal unit (30) to the location of the shelter.

[0070] The fire spread path extraction module (210) inputs the vegetation data (H) and soil moisture data (I) received from the variable data reception module (207) into the vegetation data / soil moisture data formula to calculate a soil spread possibility index indicating the degree of possibility of fire spread in a set section. The soil spread possibility index calculated in this way can indicate the high or low possibility of forest fire outbreak risk depending on the size of the value. Here, if the soil spread possibility index value is 120K / % or more, the risk of forest fire outbreak can be determined to be high. And if the soil spread possibility index value is 20K / % or less, the risk of forest fire outbreak can be determined to be low. In addition, the fire spread path extraction module (210) can calculate a wind spread possibility index that scores points with a possibility of movement according to the wind data by performing a vector product operation on the wind data (G) from the location information extracted from the fire location extraction module (204). In this way, the fire spread path extraction module (210) can calculate the wind spread possibility index as data in three-dimensional space by performing a vector product operation. In addition, the fire spread path extraction module (210) can extract the expected path (J) along which the fire spreads from the fire location extraction module (204) by calculating the inner product of the calculated soil spread possibility index and the wind spread possibility index. In addition, the fire spread path extraction module (210) can detect the movement path with the smallest total spread possibility index among the total spread possibility indices calculated from a plurality of movement paths as a fire avoidance movement path (K) and transmit it from the wireless terminal unit (30). Here, the soil moisture data, which serves as the denominator in the vegetation data / soil moisture data calculation, is a value calculated using the brightness temperature or backscattering index described above, and the vegetation data, which serves as the numerator, can be the vegetation temperature. At this time, the vegetation temperature (canopy temperature) is a factor indicating the moisture level of the vegetation, and can be calculated by the fire spread path extraction module (210).

[0071] The database module (211) can store data observed from satellites, vegetation information including vegetation temperature, soil moisture, etc. In addition, data observed from satellites equipped with microwave sensors can be stored. Of course, the database unit (12) can also store various geographic, meteorological, vegetation information, soil moisture, etc. data in addition to data observed from microwave sensors.

[0072] Hereinafter, with reference to FIGS. 12 to 16, the operation of a forest fire detection system according to one embodiment of the present invention will be described in detail.

[0073] Fig. 12 is an operational flowchart of a forest fire detection system according to one embodiment of the present invention. Figs. 14 to 16 are diagrams illustrating the operational status of a forest fire detection system according to one embodiment of the present invention.

[0074] The forest fire detection system (1) of the present invention can be started by acquiring a first image (A10) and a second image (A20) from a camera unit (10), as illustrated in FIG. 12. In the acquired first image (A10) and second image (A20), objects that match and do not match a reference flame object are first detected. When an object that matches a reference flame object is first detected, the area where the object is detected is zoomed in, the object is enlarged, and then the object is compared with the reference flame object again. At this time, when the enlarged detected object matches the reference flame object by a value higher than a reference value, the temperature of the flame object classified in the second image is measured to determine once again whether the object is a flame.

[0075] As illustrated in Fig. 13, when the location of a forest fire is identified, the forest fire detection system (1) generates a fire signal from the main server unit (20) and transmits it to the wireless terminal unit (30). At this time, a text message guiding the activation of the forest fire evacuation app is transmitted to the wireless terminal unit (30).

[0076] At this time, when a fire signal is received at the wireless terminal unit (30), the fire evacuation app (320) as shown in (a) of Fig. 14 can be activated by the user or the fire signal. When the fire evacuation app (320) is activated, a map image (321) indicating the location of the wireless terminal unit (30) and the location of the shelter received from the shelter location confirmation module (206) is output as shown in (b) of Fig. 14. At this time, the map image (321) shows, as shown, the current location (C) of the wireless terminal unit (30), a shelter (B) located near the area where the fire has occurred, and various movement routes (D) from the location of the wireless terminal unit (30) to the shelter (B). At this time, the main server unit (20) can calculate the quantified value, i.e., the value obtained by calculating the soil spread possibility index and the wind spread possibility index, i.e., the inner product value, in the fire spread path extraction module (210) for multiple compartment areas including the sections of the movement paths shown in the map image (321). For example, the main server unit (20) can calculate an inner product value 'C' for the first compartment area (①-①) including the first movement path (D1), as illustrated in FIG. 15, can calculate an inner product value 'B' for the second compartment area (①-②) including the first movement path (D1), can calculate an inner product value 'A' for the third compartment area (①-③) including the first movement path (D1), and can calculate an inner product value 'D' for the fourth compartment area (①-④) including the first movement path (D1) and the fifth compartment area (①-⑤) including the first movement path (D1).

[0077] For the first compartment area (②-①) of the second movement path (D2), the inner product value 'D' can be calculated, for the second compartment area (②-②), the inner product value 'B' can be calculated, for the third compartment area (②-③) of the second movement path (D2), the inner product value 'A' can be calculated, for the fourth compartment area (②-④), the inner product value 'B' can be calculated, and for the fifth compartment area (②-⑤), the inner product value 'D' can be calculated. And, the inner product operation value 'D' can be calculated for the first compartment area (③-①), the second compartment area (③-②), the third compartment area (③-③), the fourth compartment area (③-④), the fifth compartment area (③-⑤), the sixth compartment area (③-⑥), and the seventh compartment area (③-⑦) of the third movement path (D3). And, the inner product operation value 'D' can be calculated for the first compartment area (④-①), the second compartment area (④-②), the third compartment area (④-③), the fourth compartment area (④-④), the fifth compartment area (④-⑤), the sixth compartment area (④-⑥), the seventh compartment area (④-⑦), the eighth compartment area (④-⑧), and the ninth compartment area (④-⑨) of the fourth movement path (D4). Here, if the calculated inner product value is 'A', it means the area is at a very high risk of forest fire spread. If it is 'B', it means the area is at a high risk of forest fire spread. If it is 'C', it means the area requires caution regarding forest fire spread. If it is 'D', it means the area is at a low risk of forest fire spread.

[0078] The main server unit (20) calculates the inner product operation value for the partition area of ​​the first to fourth movement paths, and based on this, among the multiple movement paths generated by the movement path generation module (209), the movement path that has a small wind spread possibility index, does not overlap with the expected path (J) generated by the fire spread path extraction module (210), and has the shortest distance from the wireless terminal unit (30) to the shelter can be detected as the fire avoidance movement path (K). Then, the detected fire avoidance movement path (K) is transmitted to the wireless terminal unit (30) and displayed on the map image (321), as illustrated in FIG. 16.

[0079] Through this, the forest fire detection system (1) can enable a user using a wireless terminal (30) to safely reach a shelter along the route indicated on the map image (321).

[0080] As such, a forest fire detection system (1) can secure high reliability in detecting flames. Furthermore, by extracting information on the temperature, size, and direction of the detected flames, the presence or absence of a fire can be more accurately determined. In addition, by using wind data (G) acquired from a satellite, vegetation data (H) and soil moisture data (I) generated from a computational module, the point of spread from the point of fire occurrence is calculated, thereby allowing evacuees (L) to move along a safe path rather than the path of fire spread and reach a shelter (B) without being injured by fire.

[0081] Although embodiments of the present invention have been described with reference to the attached drawings, those skilled in the art will appreciate that the present invention can be implemented in other specific forms without altering the technical spirit or essential characteristics thereof. Therefore, the embodiments described above should be understood to be illustrative in all respects and not restrictive.

[0082] [Explanation of symbols]

[0083] 1: Forest fire detection system

[0084] 10: Camera section

[0085] 110: Real-time camera module 111: First video image

[0086] 120: Thermal imaging camera module 121: Second video image

[0087] 20: Main server

[0088] 201: Object Classification Extraction Module 202: Temperature Extraction Module

[0089] 203: Fire detection module 204: Fire location extraction module

[0090] 205: Notification and Alert Module 206: Shelter Location Confirmation Module

[0091] 207: Variable data reception module 208: Fire spread path extraction module

[0092] 209: Path generation module

[0093] 30: Wireless terminal donation

[0094] 310: Messaging app 320: Fire evacuation app

[0095] 321: Map image

[0096] A10: First video image A20: Second video image

[0097] A11: First image extraction flame object A12: First image extraction background object

[0098] A21: Second image extraction flame object A22: Second image extraction background object

[0099] A13: First segmented image A14: First segmented partial enlarged image

[0100] A23: Second segmented image A24: Second segmented partial enlarged image

[0101] B: Shelter C: Current location

[0102] D: Various movement routes

[0103] D1: First movement path D2: Second movement path

[0104] D3: Third movement path D4: Fourth movement path

[0105] E: Fire escape route F: Satellite

[0106] G: Wind data H: Vegetation data

[0107] I: Soil moisture data J: Expected path

[0108] K: Fire escape route

Claims

1. A camera unit (10) including a real-image camera module (110) that captures a certain area to create a first image (A10) and a thermal-image camera module (120) that captures the same area captured by the real-image camera module (110) to create a second image (A20); By including a preset standard flame object and receiving a first image (A10) and a second image (A20) from the camera unit (10), In the first image image (A10), a first image extraction flame object (A11) matching a preset standard flame object is detected, and an area that does not match the standard flame object is detected as a first image extraction background object (A12). An object classification and extraction module (201) that detects a second image extraction flame object (A22) that matches a preset standard flame object in a second image image (A20) and detects an area that does not match the standard flame object as a second image extraction background object (A22), A temperature extraction module (202) that extracts the temperature of a second image extraction flame object (A22) located at a position overlapping with a first image extraction flame object (A11) detected by an object classification extraction module (201), and A forest fire detection system (1) including a main server unit (20) equipped with a fire detection module (203) that generates a fire signal when at least one temperature extracted from a temperature extraction module (202) is higher than a preset fire reference temperature, and generates a non-fire signal when both the temperature extracted from the first image extraction flame object (A11) and the temperature extracted from the second image extraction flame object (A22) are lower than the fire reference temperature.

2. In paragraph 1, the main server unit (20) A forest fire detection system (1) including a fire location extraction module (204) that extracts location information of the area where the first image (A10) and the second image (A20) were captured from the first image (A10) and the second image (A20) when a fire signal is generated from the fire detection module (203).

3. In the second paragraph, the main server unit (20) A forest fire detection system (1) including a notification alarm module (205) that outputs a preset message when a fire signal is generated, and a shelter location confirmation module (206) in which location information of at least one shelter (B) is pre-stored.

4. In paragraph 3, A forest fire detection system (1) further comprising a wireless terminal unit (30) that has a fire evacuation app (310) installed, communicates with the main server unit (20) through data communication, and receives a message (C) from the notification alarm module (205) of the main server unit (20).

5. In paragraph 4, Fire evacuation app (310) When the wireless terminal unit (30) is turned on after receiving a message from the alarm module (205) including a map image (321), A forest fire detection system (1) that displays the location of a wireless terminal (30) and the location of a shelter received from a shelter location confirmation module (206) on a map image (321).

6. In paragraph 5, the main server unit (20) An operation module (207) that receives data observed from a satellite (F) while communicating with the satellite (F) and calculates vegetation data (H) including a vegetation index and canopy temperature, and calculates soil moisture data (I) by receiving brightness temperature or backscattering coefficients observed through a microwave installed in the satellite (F). A variable data receiving module (208) that receives wind data (G) corresponding to the location information extracted from the fire location extraction module (204) while communicating with a satellite (F) and receives vegetation data (H) and soil moisture data (I) produced by the calculation module (207), A movement path generation module (209) that generates multiple movement paths from the location of the wireless terminal unit (30) to the location of the shelter and transmits them to the wireless terminal unit (30), The vegetation data (H) received from the operation module (207) and the soil moisture data (I) received from the variable data receiving module (208) are input into the vegetation data / soil moisture data formula to calculate the soil spread possibility index indicating the degree of possibility that the fire in the set section will spread. The wind spread possibility index is calculated by performing a vector product operation on the location information extracted from the fire location extraction module (204) and the wind data (G). A forest fire detection system (1) including a fire spread path extraction module (210) that extracts an expected path (J) along which a fire spreads from a fire location extraction module (204) using a value obtained by calculating the inner product of a soil spread potential index and a wind spread potential index.

7. In paragraph 6, the main server unit (20) Among the multiple movement paths generated by the movement path generation module (209), the wind diffusion possibility index is low. It does not overlap with the expected path (J) generated by the fire spread path extraction module (210), A forest fire detection system (1) that detects the shortest distance from a wireless terminal (30) to a shelter as a fire avoidance path (K) and transmits it from the wireless terminal (30).

Images