Fire detection system

Abstract

To provide a fire detection system which can detect fire quickly and facilitates identification of an ignition source.SOLUTION: In a fire detection system 1, an imaging means group 20 having a visible image capturing means 21 and thermal image capturing means 22 is disposed in each monitoring area 11. The visible image capturing means 21 images a monitoring area 11 to acquire a visible image. The thermal image capturing means 22 images the monitoring area 11 to acquire a thermal image. Visible image determination means determines whether the visible image includes at least one of flame A1 and smoke A2. Thermal image determination means determines temperature information included in the thermal image on the basis of a reference temperature. When at least one of the results determined by the visible image determination means and the thermal image determination means is affirmative, notification means 31, 32 issues a notification about occurrence of a fire. Handling method determination means determines a handling method for the fire.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a fire detection system for detecting a fire in a monitoring area.

Background Art

[0002] Conventionally, in a plurality of monitoring areas set at major locations of a building, a system for detecting a fire has been used (see, for example, Patent Documents 1 to 5). In addition, such a system is provided with a fire detector for detecting and notifying a fire.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Patent Document 2

Patent Document 3

Patent Document 4

Patent Document 5

Summary of the Invention

Problems to be Solved by the Invention

[0004] By the way, as a fire detector, usually, a smoke type or a heat type is used. However, a smoke type fire detector cannot detect a fire until smoke flows from the ignition source. In addition, a heat type fire detector cannot detect a fire until the surrounding air becomes high temperature. Therefore, even if the ignition source is slightly away, it may take time from the occurrence of the fire to the detection. Further, even if a fire is notified by the fire detector, it is difficult to specify the ignition source.

[0005] This invention has been made in view of the above-mentioned problems, and its objective is to provide a fire detection system that can quickly detect a fire and easily identify the source of the ignition. [Means for solving the problem]

[0006] To solve the above problems, the invention described in claim 1 is a system for detecting a fire in a monitoring area, comprising: a group of imaging means having visible image imaging means for imaging the monitoring area and acquiring a visible image, and a group of imaging means for imaging the monitoring area and acquiring a thermal image, provided for each of the plurality of monitoring areas; a display means for displaying the visible image and the thermal image for each group of imaging means; a visible image determination means for determining whether or not at least one of flames and smoke is visible in the visible image; a thermal image determination means for determining temperature information contained in the thermal image based on a reference temperature; a notification means for notifying the occurrence of a fire in the corresponding monitoring area when at least one of the determination result by the visible image determination means and the determination result by the thermal image determination means is positive; and a response method determination means for determining a response method to the fire when at least one of the determination result by the visible image determination means and the determination result by the thermal image determination means is positive. A marker board having multiple through holes arranged regularly in the vertical and horizontal directions, which is imaged by the visible image capturing means and the thermal image capturing means while heated or cooled, and an image processing means that performs arithmetic processing using a coordinate transformation matrix to align the visible image and the thermal image in order to connect the overhead coordinate systems of the visible image capturing means and the thermal image capturing means by making them common via a marker coordinate system centered on the marker board that exists in the visible image and the thermal image. The essence of this invention is a fire detection system characterized by comprising the following features.

[0007] In the invention described in claim 1, the visible image determination means determines whether or not at least one of flames and smoke is visible in the visible image captured and acquired by the visible image acquisition means. At the same time, the thermal image determination means determines the temperature information contained in the thermal image captured and acquired by the thermal image acquisition means based on a reference temperature. If at least one of the determination results from the visible image determination means and the determination results from the thermal image determination means is positive, the notification means notifies of the occurrence of a fire. In other words, the fire detection system does not have to wait for the arrival of smoke or the temperature to rise, so it can quickly detect a fire based on the image captured of the monitoring area. Furthermore, the notification means notifies of the occurrence of a fire in a monitoring area where the determination results from the visible image determination means and the thermal image determination means are positive among multiple monitoring areas. This makes it easier to identify the source of ignition.

[0008] Examples of notification means for reporting the occurrence of a fire include light-emitting means such as lamps that report the occurrence of a fire by emitting light (lighting up, flashing, etc.), sound output means such as alarms that report the occurrence of a fire by sound (announcement sound, etc.), and display means such as liquid crystal display devices that report the occurrence of a fire by displaying images such as letters, symbols, or pictures.

[0009] The invention described in claim 2 is characterized in that, in claim 1, the temperature information is temperature difference information obtained by comparing a first thermal image, which is a thermal image acquired by imaging the monitoring area, with a second thermal image, which is a thermal image acquired by imaging the monitoring area at different times.

[0010] In the invention described in claim 2, the temperature difference information obtained by comparing two thermal images taken at different times becomes the temperature information contained in the thermal image. Therefore, based on the temperature information and the reference temperature, temperature changes within the monitoring area corresponding to the captured thermal image can be accurately managed. As a result, the administrator can accurately recognize which monitoring area has a large temperature change, that is, which monitoring area has a fire.

[0011] The invention described in claim 3 is characterized in that, in claim 1 or 2, it comprises a learning means for learning a method for determining whether a living organism is reflected in the visible image and the thermal image, and a storage means for storing the learning results by the learning means as learned data, wherein the visible image determination means determines whether or not a living organism is reflected in the visible image based on the learned data stored in the storage means, the thermal image determination means determines whether or not a living organism is reflected in the thermal image based on the learned data stored in the storage means, and the notification means notifies of the presence of a living organism in the corresponding monitoring area when at least one of the determination results by the visible image determination means and the determination results by the thermal image determination means is positive.

[0012] In the invention described in claim 3, the presence or absence of living organisms can be accurately determined by determining the presence or absence of living organisms based on the learning results of the learning means. Moreover, increasing the learning opportunities of the learning means improves the accuracy of the determination of the presence or absence of living organisms. Furthermore, in claim 3, the presence or absence of living organisms is confirmed by a visible image determination means determining whether or not living organisms are present in the visible image, and a thermal image determination means determining whether or not living organisms are present in the thermal image. Therefore, the presence or absence of living organisms can be confirmed with greater accuracy compared to when the presence or absence of living organisms is confirmed by either the visible image determination means or the thermal image determination means alone.

[0013] The invention described in claim 4 is characterized in that, in claim 3, the means for determining the countermeasure method determines, as the countermeasure method, to guide the living organism to an evacuation site using the notification means when the presence of a living organism is recognized by the visible image determination means and the thermal image determination means.

[0014] In the invention described in claim 4, in addition to identifying the source of ignition using the visible image determination means and the thermal image determination means, the presence of living organisms is also recognized. Therefore, based on the identified source of ignition and the recognized living organisms, living organisms can be guided to an evacuation site via a safer route. Furthermore, even if living organisms are difficult to capture in the visible image due to poor visibility caused by smoke, they will reliably appear in the thermal image. Therefore, living organisms can be quickly recognized by the thermal image determination means and guided to an evacuation site. Here, living organisms can include, for example, humans and animals.

[0015] The invention described in claim 5 is characterized in that, in claim 3 or 4, the means for determining the countermeasure method determines, as the countermeasure method, to notify the outside of the fire detection system using the notification means when the presence of a living organism is recognized by the visible image determination means and the thermal image determination means.

[0016] In the invention described in claim 5, when the presence of a living organism is recognized by the visible image determination means and the thermal image determination means, an alert is sent to the outside of the fire detection system, so that a request for rescue of the living organism can be made to the outside.

[0017] The invention described in claim 6 is characterized in that, in claim 3, the learning means learns a method for determining whether flames and smoke are reflected in the visible image, and the visible image determination means determines whether or not flames and smoke are reflected in the visible image based on the learned data stored in the storage means.

[0018] In the invention described in claim 6, the presence or absence of flames and smoke can be accurately determined by determining the presence or absence of flames and smoke based on the learning results of the learning means. Moreover, increasing the learning opportunities of the learning means improves the accuracy of the determination of the presence or absence of flames and smoke.

[0019] The invention according to claim 7 is characterized in that, in any one of claims 3 to 6, it comprises target setting means for setting, as a recognition target, a flame or a living body reflected in the visible image, and target position estimation means for estimating the position of the recognition target in an overhead coordinate system based on the visible image imaging means.

[0020] In the invention according to claim 7, the target position estimation means automatically estimates the position of the recognition target in an overhead coordinate system based on the visible image imaging means. As a result, based on the estimated position of the recognition target, the position of the flame or the living body can be easily estimated.

[0021] The invention according to claim 8 is characterized in that, in claim 7, the display means displays an image of an overhead plane, and calculates an evacuation route of the living body based on information indicating the position of the flame and information indicating the position and number of the living bodies, and displays an icon indicating the recognition target and the evacuation route of the living body on the image of the overhead plane.

[0022] In the invention according to claim 8, the display means displays an image of an overhead plane and also displays an icon indicating the recognition target and the evacuation route of the living body on the image. Thereby, the estimated position of the flame or the living body and the evacuation route of the living body can be displayed clearly.

[0023] The invention according to claim 9 is characterized in that, in claim 8, the notification means notifies the evacuees of the evacuation route by voice or image.

[0024] In the invention according to claim 9, since the notification means automatically notifies the evacuees of the evacuation route, the evacuees can evacuate after surely recognizing the evacuation route.

[0025] Claim 1 etc. The invention described So, Imaging a marker board having a plurality of through holes in a heated or cooled state .Also,Image processing means that aligns the visible image and the thermal image based on the shape information of the marker board present in the visible image and the thermal image. It is equipped with.

[0026] Therefore, the above In this invention, the visible image and thermal image are aligned by an image processing means, enabling accurate detection of the location of flames and smoke. Furthermore, since a marker board with multiple through holes is used for aligning the visible and thermal images, the shape information is simplified. As a result, alignment control based on the shape information of the marker board becomes easier, reducing the burden on the image processing means. Moreover, since the marker board is heated or cooled during alignment, the marker board can be reliably reflected in the thermal image acquired by the thermal image acquisition means. As a result, alignment can be reliably performed based on the shape information of the marker board present in both the visible and thermal images. [Effects of the Invention]

[0027] As detailed above, claims 1 to 9 According to the invention described, fires can be detected quickly, and the source of ignition can be easily identified. [Brief explanation of the drawing]

[0028] [Figure 1] A schematic diagram showing the fire detection system in this embodiment. [Figure 2] A floor plan showing a specific floor of a building. [Figure 3] A schematic diagram showing a visible image. [Figure 4] A schematic diagram showing a thermal image. [Figure 5] Front view showing the marker board. [Figure 6] An explanatory diagram showing how the same marker board is captured from different cameras. [Figure 7] A flowchart illustrating the processes for detecting a fire and guiding evacuation. [Modes for carrying out the invention]

[0029] Hereinafter, one embodiment embodying the present invention will be described in detail with reference to the drawings.

[0030] As shown in Figures 1 and 2, the fire detection system 1 of this embodiment is a system that detects fires in multiple monitoring areas 11 (Area A, Area B, Area C, ...) set in key locations of a building 10. The fire detection system 1 is provided with an imaging means group 20 having a visible camera 21 (visible image acquisition means) and a thermal camera 22 (thermal image acquisition means) for each of the multiple monitoring areas 11. The visible camera 21 captures the monitoring area 11 to acquire a visible image 24 (see Figure 3) and outputs the image data of the acquired visible image 24. The thermal camera 22 captures the monitoring area 11 to acquire a thermal image 25 (see Figure 4) and outputs the image data of the acquired thermal image 25. The visible image 24 is a color image, and the thermal image 25 is an infrared image.

[0031] Furthermore, a monitor 31 (display means) constituting the fire detection system 1 is provided in the control room 12 of building 10. Visible images 24 and thermal images 25 are displayed on the display screen of the monitor 31 for each imaging means group 20. The visible image 24 is set to show flames A1, smoke A2, and a person A3 (living organism) (see Figure 3). When a person A3 is shown in the thermal image 25, a rectangular frame 26 recognizing the person A3 is displayed in the thermal image 25 so as to surround the person A3 (see Figure 4). In this embodiment, the frame 26 consists of red straight lines, but it does not have to be red, and it may be a dashed line. Also, the frame 26 does not have to be displayed. Furthermore, when flames A1 are shown in the thermal image 25, a rectangular frame recognizing the flames A1 may be displayed in the thermal image 25 so as to surround the flames A1. Additionally, when smoke A2 is shown in the thermal image 25, a rectangular frame recognizing the smoke A2 may be displayed in the thermal image 25 so as to surround the smoke A2.

[0032] As shown in Figure 2, the monitor 31 displays a floor plan 27, which is an overhead view image. The floor plan 27 is an image showing a specific floor of the building 10 as seen from directly above. Also displayed on the floor plan 27 are an icon 41 representing flame A1 and an icon 42 representing human A3. In this embodiment, the icon 41 representing flame A1 is a picture of flame A1, and the icon 42 representing human A3 is a picture of human A3 as seen from directly above. Furthermore, the floor plan 27 displays an icon 43 representing the imaging means group 20 (visible camera 21 and thermal camera 22). In this embodiment, the icon 43 representing the imaging means group 20 is a picture of a camera.

[0033] Furthermore, as shown in Figure 1, an alarm 32, which constitutes the fire detection system 1, is installed in the control room 12. In addition, the fire detection system 1 is equipped with a speaker 33 (notification means) for each of the multiple monitoring areas 11.

[0034] Next, the electrical configuration of the fire detection system 1 will be described.

[0035] As shown in Figure 1, the fire detection system 1 is equipped with a personal computer 51 in the control room 12, and the personal computer 51 is equipped with a control device (not shown) that comprehensively controls the entire system. The control device is composed of a well-known computer consisting of a CPU, ROM, RAM, etc. The monitor 31, alarm 32, and speaker 33 are electrically connected to the CPU. The monitor 31, alarm 32, and speaker 33 may also be connected to the CPU wirelessly.

[0036] Furthermore, the visible-lens camera 21 and the thermal camera 22 are electrically connected to the CPU. The RAM stores the visible-lens image 24 acquired by the visible-lens camera 21 and the thermal image 25 acquired by the thermal camera 22. The ROM stores the program for controlling the fire detection system 1. The visible-lens camera 21 and the thermal camera 22 may also be connected to the CPU wirelessly.

[0037] Next, we will explain how to align the visible image 24 and the thermal image 25.

[0038] First, the method for connecting the visible light camera 21 and the thermal camera 22 will be explained based on Figure 6. As a prerequisite, the positional relationship between the cameras 21 and 22 and the floor surface of the corresponding monitoring area 11 must be known. Also, a marker board 60 (see Figure 5) with multiple through-holes 61 is prepared.

[0039] First, the marker board 60 is imaged using the respective cameras 21 and 22 while it is heated. Then, the CPU of the personal computer 51 aligns the visible image 24 and the thermal image 25 based on the shape information of the marker board 60 present in the captured visible image 24 and thermal image 25. In other words, the CPU functions as an "image processing means".

[0040] Specifically, the CPU estimates the positional relationship between the marker board 60 and the visible camera 21, and the positional relationship between the marker board 60 and the thermal camera 22. When the marker board 60 is imaged in this way, the direct positional relationship between cameras 21 and 22 is unknown, but the coordinate system centered on the marker board 60 itself (marker coordinate system) is shared. Therefore, it is possible to convert points seen from each camera 21 and 22 to points seen from the marker board 60 for comparison, and then convert them to points seen from the other camera.

[0041] Therefore, the CPU performs calculations to connect the overhead coordinate systems of cameras 21 and 22 via the marker board 60, thereby aligning the visible image 24 captured by the visible camera 21 with the thermal image 25 captured by the thermal camera 22. Here, the "overhead coordinate system" refers to the coordinate system viewed vertically from above the cameras 21 and 22, and is represented as an xy coordinate system with the same direction as the cameras 21 and 22 as the y-axis, and the direction perpendicular to the y-axis as the x-axis. Furthermore, the "process of aligning the visible image 24 with the thermal image 25" refers to the process of understanding the positional relationship between the cameras 21 and 22 and converting a point (x2, y2) in the overhead coordinate system of the thermal camera 22 to a point (x1, y1) in the overhead coordinate system of the visible camera 21. Furthermore, the system may perform a process to convert points in the overhead coordinate system of the visible light camera 21 to points in the overhead coordinate system of the thermal camera 22, or it may perform a process to convert both points in the overhead coordinate system of the visible light camera 21 and points in the overhead coordinate system of the thermal camera 22 to points in another overhead coordinate system.

[0042] Since the overhead coordinate systems of each camera 21 and 22 are located on the same floor plane, transformations between the overhead coordinate systems can be performed using affine transformations, although other methods may also be used. Specifically, a conventionally known affine transformation matrix for performing coordinate transformations is obtained by utilizing the properties of the marker coordinate system described above. Then, the transformation from a point (x2, y2) in the overhead coordinate system of the thermal camera 22 to a point (x1, y1) in the overhead coordinate system of the visible camera 21 is calculated using the affine transformation matrix.

[0043] Therefore, according to this embodiment, the overhead coordinate systems of cameras 21 and 22 are connected via a marker board 60 that is visible to both cameras 21 and 22. Specifically, points in the overhead coordinate system of the thermal camera 22 are converted to points in the overhead coordinate system of the visible camera 21, thereby unifying the overhead coordinate systems. This improves the accuracy of estimating the positions of flame A1, smoke A2, and person A3. Moreover, since the overhead coordinate systems are unified to match a single reference camera, the positions of flame A1, smoke A2, and person A3 that have been identified become easier for administrators to understand.

[0044] Next, I will explain how to detect a fire.

[0045] First, the fire detection system 1 continuously monitors multiple monitoring areas 11 set inside and outside the building 10 using a visible-lens camera 21 and a thermal camera 22. For example, in a specific monitoring area 11, the visible-lens camera 21 of the imaging means group 20 installed in that monitoring area 11 captures the monitoring area 11 and acquires a visible image 24. In addition, the thermal camera 22 of the same imaging means group 20 captures the monitoring area 11 and acquires a thermal image 25. The imaging of the monitoring area 11 by the visible-lens camera 21 and the thermal camera 22 is performed periodically.

[0046] Each time imaging of the monitoring area 11 is completed, the visible camera 21 outputs the acquired visible image data 24 to the CPU of the personal computer 51, and the thermal camera 22 outputs the acquired thermal image data 25 to the CPU. The CPU stores the visible image 24 and thermal image 25 indicated by the input image data in RAM.

[0047] The CPU then outputs a drive signal to the monitor 31 and controls it to display the visible image 24 and thermal image 25 stored in RAM on the monitor 31. The monitor 31 is configured to constantly display information from cameras 21 and 22 (visible image 24 and thermal image 25).

[0048] Next, in step S11 of Figure 7, the CPU, which is the "visible image determination means," determines whether or not at least one of flame A1 and smoke A2 is visible in the visible image 24. The CPU also functions as a "learning means." More specifically, the CPU learns in advance how to determine whether flame A1 is visible in the visible image 24 or how to determine whether smoke A2 is visible in the visible image 24, based on the image data of the visible image 24 acquired by the visible camera 21, and obtains the learning results. The CPU then stores the obtained learning results in the RAM as first learned data. In other words, the RAM functions as a "storage means." As a result, the CPU can determine whether or not at least one of flame A1 and smoke A2 is visible in the visible image 24 based on the first learned data stored in the RAM.

[0049] In the following step S12, the CPU, which is the "thermal image determination means," determines the temperature information contained in the thermal image 25 based on the reference temperature. Specifically, the CPU compares a first thermal image 25, which is a thermal image 25 acquired by imaging the monitoring area 11 in the past, with a second thermal image 25, which is a thermal image 25 acquired by imaging the monitoring area 11 at a different time (in this embodiment, the present time). In this embodiment, the second thermal image is compared with five first thermal images acquired by imaging 1 second, 10 seconds, 30 seconds, 1 minute, and 10 minutes ago. More specifically, the CPU calculates the difference (temperature difference) between the temperature at each pixel of the first thermal image and the temperature at each pixel of the second thermal image. Then, the temperature difference information indicating the temperature difference at each pixel is used as the temperature information contained in the thermal image 25. After the temperature difference has been calculated for all pixels, the CPU determines whether or not there are any pixels whose temperature difference is higher than the reference temperature (100°C in this embodiment).

[0050] Then, if at least one of the results of the visible image determination means and the thermal image determination means is positive, the CPU determines in step S13 that a fire has occurred in the corresponding monitoring area 11. In the following step S14, the CPU outputs a drive signal to the monitor 31 and controls it to display a message indicating that a fire has occurred (for example, the words "Fire occurred in area A"). This notifies the administrator of the occurrence of a fire in the corresponding monitoring area 11. At the same time, the CPU outputs a drive signal to the alarm 32 and controls it to activate. This causes the light (light-emitting means) of the alarm 32 to turn on and the alarm (sound output means) of the alarm 32 to announce the temperature, notifying the administrator of the occurrence of a fire in the corresponding monitoring area 11. In other words, the monitor 31 and the alarm 32 have the function of "notification means".

[0051] Next, we will explain how to guide person A3 to an evacuation location during a fire.

[0052] The fire detection system 1 continuously monitors multiple monitoring areas 11 using a visible light camera 21 and a thermal camera 22, even during a fire. Specifically, the visible light camera 21 captures images of the monitoring areas 11 to acquire visible images 24, and the thermal camera 22 captures images of the monitoring areas 11 to acquire thermal images 25. The acquired visible images 24 and thermal images 25 are stored in RAM by the CPU.

[0053] Next, in step S21 of Figure 7, the CPU, which is the "visible image determination means," determines whether or not human A3 is visible in the visible image 24. More specifically, the CPU learns in advance a method for determining whether or not human A3 is visible in the visible image 24 based on the image data of the visible image 24 acquired by the visible camera 21, and obtains the learning result. The CPU then stores the obtained learning result as second learned data in RAM. As a result, the CPU can determine whether or not human A3 is visible in the visible image 24 based on the second learned data stored in RAM.

[0054] In the following step S22, the CPU, which is the "thermal image determination means," determines whether or not human A3 is visible in the thermal image 25. More specifically, the CPU pre-learns a method for determining whether or not human A3 is visible in the thermal image 25 based on the image data of the thermal image 25 acquired by the thermal camera 22, and obtains the learning result. The CPU then stores the obtained learning result in RAM as third learned data. As a result, the CPU can determine whether or not human A3 is visible in the thermal image 25 based on the third learned data stored in RAM.

[0055] Next, if the CPU recognizes the presence of human A3 using the visible image detection means and the thermal image detection means, it guides human A3 to an evacuation location. Specifically, first, the CPU estimates the position of flame A1 (ignition source) and human A3 from the visible image 24, using the visible camera 21 as a reference. As a prerequisite, the visible camera 21 used must have a known positional relationship with the floor surface of the corresponding monitoring area 11.

[0056] Furthermore, in order to perform position estimation, the intrinsic parameters of the visible camera 21 are required. The intrinsic parameters are a matrix consisting of the focal length and optical center of the visible camera 21. The intrinsic parameters allow a point in the camera coordinate system to be transformed into a point on the image plane.

[0057] Furthermore, the internal parameters can be calculated using the CalibrateCamera function provided by OpenCV (Open Source Computer Vision Library) or by calculations from the datasheet of the visible camera 21. The CPU then stores the information of the visible camera 21's internal parameters in RAM.

[0058] Next, the CPU sets the flame A1 and the human A3, both visible in the captured visible image 24, as recognition targets. In other words, the CPU functions as a "target setting means."

[0059] Next, the CPU estimates the position of the object to be recognized on the overhead coordinate system based on the visible camera 21, based on the internal parameter information stored in RAM. In other words, the CPU functions as an "object position estimation means". Specifically, it calculates the position (x,y) of the object to be recognized on the coordinate system based on the visible camera 21 as follows: First, it determines the object point that is assumed to exist on the overhead plane by imaging the flame A1 or the person A3, for example.

[0060] The CPU then calculates a three-dimensional direction vector for the target point. Next, based on the direction vector, it calculates a vector indicating the position from the origin to the recognized object on the overhead plane in the camera coordinate system. As a result, the position (x,y) of the recognized object is obtained in polar coordinates from the calculated vector. This allows the position of flame A1 or human A3 to be estimated relative to the visible camera 21. Alternatively, the position of flame A1 or human A3 may be estimated using a conventionally known method, such as the one shown in Japanese Patent Application Publication No. 2020-136700.

[0061] Subsequently, the CPU displays the floor plan 27 on the monitor 31 and processes the display of an icon 41 representing the recognition target (flame A1) and an icon 42 representing the recognition target (human A3) on the floor plan 27. This displays the location of flame A1 on the monitor 31, notifying the administrator of the occurrence of a fire in the corresponding monitoring area 11. In addition, the location of human A3 is displayed on the monitor 31, notifying the administrator of the presence of human A3 in the corresponding monitoring area 11.

[0062] Furthermore, the CPU determines a course of action to take in response to the fire if at least one of the results obtained by the visible image determination means and the thermal image determination means is positive. In other words, the CPU functions as a "course of action determination means". Specifically, when the CPU recognizes the presence of person A3, it decides that guiding person A3 to an evacuation location using speaker 33 is the course of action. Then, in step S23, the CPU calculates an evacuation route R1 (see Figure 2) for each person A3 based on information indicating the location of flame A1 and information indicating the location and number of people A3.

[0063] Next, the CPU processes the calculation of the evacuation route R1 on the floor plan 27. At the same time, in step S24, the CPU outputs a drive signal to the speaker 33 and controls the speaker 33 to operate. As a result, the speaker 33 makes an audio guidance announcement to the evacuee, person A3, informing them of the evacuation route R1. Consequently, the direction of evacuation is indicated based on the positions of the flame A1 and person A3, and person A3 is guided to the evacuation exit (step S25).

[0064] Therefore, according to this embodiment, the following effects can be obtained.

[0065] (1) In the fire detection system 1 of this embodiment, the visible image determination means (CPU) determines whether or not at least one of flame A1 and smoke A2 is visible in the visible image 24. At the same time, the thermal image determination means (CPU) determines whether or not there is a temperature difference in each pixel of two thermal images 25 (first thermal image and second thermal image) taken at different times that is higher than the reference temperature. Then, if at least one of the determination results from the visible image determination means and the determination results from the thermal image determination means is positive, the monitor 31 and alarm 32 report the occurrence of a fire. In other words, the fire detection system 1 does not have to wait for the arrival of smoke A2 or for the temperature to rise, so it can quickly detect a fire based on the images 24 and 25 that have been captured in the monitoring area 11. Furthermore, the monitor 31 and alarm 32 report the occurrence of a fire in the monitoring area 11 where the determination results from the visible image determination means and the thermal image determination means are positive, among the multiple monitoring areas 11. This makes it easy to identify the source of ignition.

[0066] (2) In this embodiment, in addition to identifying the ignition source using the visible image determination means and the thermal image determination means, the presence of person A3 is also recognized. Therefore, based on the identified ignition source and the recognized person A3, person A3 can be guided to an evacuation location via a safer route. Furthermore, even if person A3 is difficult to see in the visible image 24 due to poor visibility caused by smoke A2, person A3 will be reliably captured in the thermal image 25. Therefore, the CPU can quickly recognize person A3 and guide them to an evacuation location.

[0067] (3) The fire detection system 1 of this embodiment controls two functions (a function to detect a fire and a function to guide human A3 to an evacuation location) based on images 24 and 25 acquired using two types of cameras 21 and 22. Therefore, the fire detection system 1 can be implemented at a lower cost than if a separate system were prepared for each function.

[0068] The above embodiment may be modified as follows.

[0069] In the above embodiment, the first thermal image (thermal image 25) was an image acquired by capturing the monitoring area 11 1 second, 10 seconds, 30 seconds, 1 minute, and 10 minutes ago, while the second thermal image (thermal image 25) was an image of the monitoring area 11 at the present time. However, the time at which the first thermal image is acquired can be changed as appropriate. Specifically, the first thermal image may be an image acquired by capturing the monitoring area 11, for example, 30 minutes or 60 minutes ago.

[0070] In the above embodiment, the occurrence of a fire was detected when the temperature difference (temperature difference information) obtained by comparing two thermal images 25 (specifically, the first thermal image and the second thermal image) exceeded the reference temperature (here, 100°C). However, the occurrence of a fire may also be detected when the temperature of any pixel in the thermal image 25 exceeds the reference temperature (for example, 100°C).

[0071] • In the above embodiment, the visible camera 21 was a video camera that captures video, but it may also be a camera that captures still images. Also, in the above embodiment, the visible camera 21 was a monocular camera, but it may also be other cameras such as a compound-lens camera. Furthermore, a depth camera or stereo camera that can measure the depth of the image may also be used.

[0072] Furthermore, the visible light camera 21 may be an existing camera already installed in the building 10. In this way, it is not necessary to install a large number of visible light cameras 21 to detect fires. As a result, the installation cost of the fire detection system 1 can be significantly reduced.

[0073] The CPU in the above embodiment confirmed the presence or absence of human A3 by determining whether or not human A3 is visible in the visible image 24 and whether or not human A3 is visible in the thermal image 25 (see steps S21 and S22 shown in Figure 7). However, the presence or absence of human A3 may be confirmed by determining whether or not human A3 is visible in the visible image 24, or by determining whether or not human A3 is visible in the thermal image 25.

[0074] The CPU in the above embodiment may also notify an external party of the fire detection system 1 when it recognizes the presence of human A3. Examples of external parties of the fire detection system 1 include a security company, a fire station, or a smartphone held by the administrator.

[0075] In the above embodiment, the notification means was a speaker 33 that notified evacuees of the evacuation route R1 by voice. However, a display means that notifies evacuation route R1 by image (specifically, a monitor or a smartphone carried by the evacuee, etc.) may also be used as the notification means.

[0076] In the above embodiment, when aligning the visible image 24 and the thermal image 25, the visible camera 21 and the thermal camera 22 were imaging with the marker board 60 heated. However, the visible camera 21 and the thermal camera 22 may also image with the marker board 60 cooled. Alternatively, the thermal camera 22 may image with the marker board 60 heated or cooled, while the visible camera 21 images with the marker board 60 neither heated nor cooled.

[0077] In the above embodiment, a two-dimensional marker board 60 was used to align the visible image 24 and the thermal image 25. However, a three-dimensional object such as a protrusion present within the monitoring area 11 may be used to align both images 24 and 25. Alternatively, two-dimensional markers such as ArUco markers, QR codes (registered trademark of DENSO WAVE INCORPORATED), AprilTag, or other two-dimensional objects such as pictures, letters, or symbols may be used to align both images 24 and 25.

[0078] • The fire detection system 1 in the above embodiment was a system for detecting fires in building 10, but it may also be a system for detecting fires in buildings in general, such as factories, hospitals, residences, and public facilities.

[0079] Next, in addition to the technical ideas described in the claims, the technical ideas that can be grasped by the embodiments described above are listed below.

[0080] (1) The fire detection system according to claim 2, characterized in that the temperature information is temperature difference information calculated when the temperature shown in the second thermal image is higher than the temperature shown in the first thermal image.

[0081] (2) The fire detection system according to claim 8 or 9, characterized in that, in addition to the icon indicating the object to be recognized, an icon indicating the group of imaging means is displayed on the overhead view image. [Explanation of symbols]

[0082] 1… Fire detection system 11…Monitoring area 20… Group of imaging means 21…Visible Camera as a Means of Acquiring Visible Images 22. Thermal cameras as a means of acquiring thermal images 24…Visible images 25… Thermal images 27…Floor plan as an overhead view image 31... Monitors as display and notification means 32… Alarm devices as a means of notification 33. Speakers as a means of notification 41, 42… Icons indicating the object to be recognized 60… Marker board 61…Through hole A1…flame A2…Smoke A3... Human beings as living organisms and evacuees R1...Evacuation Route

Claims

1. A system for detecting fires within a monitored area, A group of imaging means, each having visible image imaging means for capturing images of the monitoring area and acquiring visible images, and thermal image imaging means for capturing images of the monitoring area and acquiring thermal images, is provided for each of the multiple monitoring areas. A display means for displaying the visible image and the thermal image for each group of imaging means, A visible image determination means for determining whether or not at least one of flames and smoke is visible in the visible image, A thermal image determination means that determines the temperature information contained in the thermal image based on a reference temperature, A notification means that notifies of the occurrence of a fire in the corresponding monitoring area when at least one of the determination result by the visible image determination means and the determination result by the thermal image determination means is positive, A means for determining a method of dealing with a fire, which determines a method of dealing with a fire when at least one of the determination results from the visible image determination means and the determination results from the thermal image determination means are positive. A marker board having multiple through holes arranged regularly in the vertical and horizontal directions, and which is imaged by the visible image capturing means and the thermal image capturing means while heated or cooled, Image processing means that performs alignment of the visible image and the thermal image by performing calculations using a coordinate transformation matrix in order to connect the overhead coordinate systems of the visible image capturing means and the thermal image capturing means by making them common via a marker coordinate system centered on the marker board present in the visible image and the thermal image, and A fire detection system characterized by being equipped with the following features.

2. The fire detection system according to claim 1, characterized in that the temperature information is temperature difference information obtained by comparing a first thermal image, which is a thermal image acquired by imaging the monitoring area, with a second thermal image, which is a thermal image acquired by imaging the monitoring area at different times.

3. The system comprises a learning means for learning a method for determining whether living organisms are present in the visible image and the thermal image, and a storage means for storing the learning results from the learning means as learned data. The thermal image determination means determines whether or not a living organism is visible in the thermal image based on the learned data stored in the storage means. The notification means notifies of the presence of a living organism in the corresponding monitoring area when at least one of the determination results from the visible image determination means and the determination results from the thermal image determination means is positive. The fire detection system according to claim 1 or 2, characterized by the above.

4. The fire detection system according to claim 3, characterized in that the countermeasure method determination means determines, as the countermeasure method, to guide the living organism to an evacuation site using the notification means when the presence of a living organism is recognized by the visible image determination means and the thermal image determination means.

5. The fire detection system according to claim 3 or 4, characterized in that the countermeasure determination means determines, as the countermeasure, to notify the outside of the fire detection system using the notification means when the presence of a living organism is recognized by the visible image determination means and the thermal image determination means.

6. The learning means learns a method for determining whether flames and smoke are visible in the visible image. The visible image determination means determines whether or not flames and smoke are visible in the visible image based on the learned data stored in the storage means. The fire detection system according to claim 3.

7. A target setting means for setting flames or living organisms visible in the aforementioned visible image as targets for recognition, A target position estimation means for estimating the position of the recognized object on an overhead coordinate system based on the visible image acquisition means, and A fire detection system according to any one of claims 3 to 6, characterized by comprising:

8. The display means displays an overhead view image, Based on the information indicating the location of the flame and the information indicating the location and number of the living organisms, the evacuation route of the living organisms is calculated. On the overhead view image, an icon representing the object to be recognized and the evacuation route of the living organism are displayed. The fire detection system according to feature 7.

9. The fire detection system according to claim 8, characterized in that the notification means notifies evacuees of the evacuation route by voice or image.

Images