JP2015114930A - Fire detection system and fire detection method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make it possible to ensure detection of a fire from a certain scale of a fire on the basis of a fire observation image, irrespective of a distance to the fire.SOLUTION: A fire image determination unit 24 determines a fire image in which a fire is present from images picked up by a monitoring camera 10, and an angle-of-view detection unit 26 detects an angle of view of the fire from the fire image. Furthermore, a frequency detection unit 28 detects a fluctuation frequency of the fire, and an equivalent-diameter calculation unit 30 calculates an equivalent diameter of the fire based on the detected fluctuation frequency of the fire by means of a predetermined experimental formula. A length calculation unit 32 calculates a fire length from the monitoring camera 10 to the fire on the basis of the angle of view of the fire and the equivalent diameter of the fire. A fire determination unit 34 determines that the fire is a fire, and outputs a fire detection signal as well as information about the fire length and/or the equivalent diameter if the equivalent diameter of the fire is equal to or larger than a predetermined threshold.

Description

The present invention relates to a fire detection system and a fire detection method for detecting a flame in the early stage of a fire from an image of a monitoring area captured by a monitoring camera.

  Conventionally, various fire detection systems have been proposed in which a flame caused by a fire is detected by performing image processing on an image of a monitoring area captured by a monitoring camera.

  In such a fire detection system, early detection of a fire is important from the viewpoint of initial extinguishing and evacuation guidance for the occurrence of a fire.

Therefore, in the conventional system, in order to prevent false detection by light from a light source other than the flame, autocorrelation is calculated for the circularity of the object image area, the temporal dispersion rate of the size, and the temporal change of the size. The method of determining whether or not the object is a flame (Patent Document 1), or the infrared monitoring image has fluctuations over a certain area of high brightness, the frequency of change intensity is not constant, the movement of the high temperature region A method of discriminating a flame on condition that the condition is random (Patent Document 2), a method of determining a flame from a captured image based on a change in temporal enlargement and reduction and / or fluctuation frequency (Patent Document 3) ) Etc. are known.

JP-A-8-305980 JP 2002-279545 A JP 2003-270037 A

  However, in such a conventional fire detection system that detects a fire from an image of a flame, the size when a small flame near the surveillance camera is imaged and the size when a large large flame is imaged are large. For example, a lighter flame lit in the immediate vicinity of a surveillance camera may be mistaken as a flame from a large fire at a distance.

  In addition, even if the flames are the same size, if they are far away, they will be small flame images, and if they are nearby, they will be large flame images. If the flame does not increase, there is a problem that a fire cannot be detected from the flame image, and the time delay until the fire is detected increases.

The present invention provides a fire detection system and a fire detection method capable of reliably detecting a fire and estimating a distance to a fire point from an observation image of a flame whose size changes due to a difference in distance and disaster scale. Objective.

(Fire detection system)
The fire detection system of the present invention is
Imaging means for photographing the surveillance area;
Flame image determination means for determining a flame image in which a flame is present from an image captured by the imaging means;
Expected angle detection means for detecting the expected angle of the flame from the flame image determined by the flame image determination means,
A frequency detection means for detecting the fluctuation frequency of the flame present in the monitoring area;
Equivalent diameter calculation means for calculating the equivalent diameter of the flame based on the fluctuation frequency of the flame detected by the frequency detection means;
A distance calculating means for calculating a flame distance from the imaging means to the flame based on the expected angle of the flame detected by the expected angle detecting means and the equivalent diameter of the flame calculated by the equivalent diameter calculating means;
When the equivalent diameter of the flame calculated by the equivalent diameter calculation means is greater than or equal to a predetermined threshold value, it is determined as a fire and in other cases it is determined as a non-fire, and when a fire is determined, the flame distance calculated by the distance calculation means and Or a fire determination means for outputting a fire detection signal together with information indicating the equivalent diameter of the flame calculated by the equivalent diameter calculation means,
Is provided.

(Calculation of flame equivalent diameter)
The equivalent diameter calculation means sets the equivalent diameter 2R of the flame to 2R = (Z / 18.622) ^{1 / −0.554,} where Z is the fluctuation frequency.
Calculated by

(Calculation of flame equivalent diameter)
When the equivalent diameter of the flame is 2R and the prospective angle of the flame is θ, the distance calculation means calculates the flame distance L,
L = R / tan (θ / 2)
Calculated by

(Flame image judgment)
The flame image determination unit detects a flame candidate region that is a set of pixels having a predetermined luminance or higher from the image captured by the imaging unit, and determines that the image is a flame image when the number of horizontal pixels of the flame candidate region is equal to or greater than a predetermined threshold.

(Detection of the expected angle of flame)
The expected angle detection means detects the expected angle θ of the flame by multiplying the expected angle α per pixel determined by the horizontal viewing angle and the number of horizontal pixels of the imaging means and the horizontal pixel number X of the flame measured from the flame image. To do.

(Fluctuation frequency detection)
The frequency detection means detects the fluctuation frequency of the flame from successive flame images based on a time-series change in the number of pixels with luminance equal to or higher than a predetermined threshold in the entire image or the flame candidate region.

  Further, the frequency detection means may detect the fluctuation frequency of the flame from the successive flame images from the time series change of the luminance sum that is equal to or greater than a predetermined threshold in the entire image or the flame region.

  The frequency detecting means includes a light receiving element means for receiving infrared light or visible light from a flame existing in the monitoring area and outputting a light receiving signal, and detects a fluctuation frequency from the light receiving signal output from the light receiving element means. You may do it.

(Fire judgment)
The fire determination means determines a fire when the equivalent diameter of the flame is equal to or greater than a predetermined threshold and the flame distance is equal to or greater than the predetermined threshold distance, and otherwise determines as a non-fire.

(Method)
The present invention provides a fire detection method,
A flame image determination step for determining a flame image in which a flame is present from an image captured by an imaging unit that captures a monitoring region;
A prospective angle detection step of detecting a prospective angle of the flame from the flame image determined in the flame image determination step;
A frequency detection step for detecting the fluctuation frequency of the flame present in the monitoring area;
An equivalent diameter calculating step for calculating the equivalent diameter of the flame based on the fluctuation frequency of the flame detected in the frequency detecting step;
A distance calculating step for calculating a flame distance from the imaging means to the flame based on the estimated angle of the flame detected in the estimated angle detection step and the equivalent diameter of the flame calculated in the equivalent diameter calculating step;
If the equivalent diameter of the flame calculated in the equivalent diameter calculation step is greater than or equal to a predetermined threshold, it is determined as a fire and in other cases it is determined as a non-fire, and if a fire is determined, the flame distance calculated in the distance calculation step and Or a fire determination step for outputting a fire detection signal together with information indicating the equivalent diameter of the flame calculated in the equivalent diameter calculation step;
Is provided.

  Other features of the fire detection method according to the present invention are basically the same as those of the fire detection system.

(Basic effect)
The fire detection system of the present invention includes an imaging unit that captures a monitoring area, a flame image determination unit that determines a flame image in which a flame is present from an image captured by the imaging unit, and a flame from a flame image determined by the flame image determination unit The expected angle detection means for detecting the expected angle, the frequency detection means for detecting the fluctuation frequency of the flame existing in the monitoring area, and the equivalent diameter for calculating the equivalent diameter of the flame based on the fluctuation frequency of the flame detected by the frequency detection means A calculation means; a distance calculation means for calculating a flame distance from the imaging means to the flame based on the expected angle of the flame detected by the expected angle detection means and the equivalent diameter of the flame calculated by the equivalent diameter calculation means; and an equivalent diameter calculation means If the equivalent flame diameter calculated in step 1 is greater than or equal to a predetermined threshold, it is determined that there is a fire and in other cases it is determined that there is no fire. Since there is provided a fire determination means for outputting a fire detection signal together with information indicating the flame distance calculated by the calculation means and / or the equivalent diameter of the flame calculated by the equivalent diameter calculation means, a captured image of a flame caused by a fire of the same scale is provided. Estimate the size of the flame by calculating the equivalent diameter of the flame from the fluctuation frequency of the flame, based on the relationship that the fluctuation frequency of the flame decreases as the flame obtained from the empirical formula increases, even if the size varies depending on the fire point distance. This makes it possible to reliably detect a fire due to a fire regardless of the fire point distance.

  In addition, even if the flame images of fires of different scales are the same depending on the fire point distance and it is not possible to distinguish between a near flame and a far flame, the flame obtained from the empirical formula also becomes large. The flame size can be estimated by obtaining the equivalent diameter of the flame from the flame fluctuation frequency based on the relationship that the flame fluctuation frequency becomes lower in accordance with the Detect and enable alarm.

  Further, based on the equivalent flame diameter obtained from the fluctuation frequency, the distance from the imaging means to the flame is obtained and the flame distance and / or the equivalent flame diameter are output together with the fire detection signal. For example, a fire alarm is output and the flame is outputted. By displaying the distance and the equivalent diameter (size) of the flame, it is possible to easily grasp the fire occurrence state in the monitoring area and to take appropriate measures.

(Effects of calculation of flame equivalent diameter)
The equivalent diameter calculation means sets the equivalent diameter 2R of the flame to 2R = (Z / 18.622) ^{1 / −0.554,} where Z is the fluctuation frequency.
By calculating by this, it is possible to easily estimate the size of the flame.

(Effects of calculation of flame equivalent diameter)
Further, the distance calculating means calculates the flame distance L when the equivalent diameter of the flame is 2R and the prospective angle of the flame is θ.
L = R / tan (θ / 2)
Because it is calculated by the above, it is possible to estimate the fire point distance to the flame from the flame image that changes in size according to the distance. Make it possible to accurately grasp the state of the fire.

(Effect by flame image judgment)
The flame image determination unit detects a flame candidate region that is a set of pixels having a predetermined luminance or higher from the image captured by the imaging unit, and determines that the image is a flame image when the number of horizontal pixels of the flame candidate region is equal to or greater than a predetermined threshold. Therefore, for a flame image with a high possibility of a flame due to a fire, the process calculates the equivalent diameter from the expected angle of the flame image to determine the fire, and calculates the flame distance from the fluctuation frequency and the equivalent diameter It is possible to reduce the processing burden by eliminating the need for processing for determining a fire for an image with a low possibility of flame.

(Effect by detecting the expected angle of flame)
Further, the prospective angle detection means multiplies the prospective angle α per pixel determined by the horizontal viewing angle and the number of horizontal pixels of the image pickup means and the horizontal pixel number X of the flame measured from the flame image, thereby calculating the expected angle θ of the flame. Therefore, it is possible to easily calculate the fire point distance based on the equivalent diameter of the flame detected from the estimated angle of the flame and the fluctuation frequency.

(Effects of fluctuation frequency detection)
Further, the frequency detection means detects the fluctuation frequency of the flame from the continuous flame image, from the time series change of the number of pixels of the brightness that is equal to or higher than the predetermined threshold in the entire image or the flame candidate area, or from the continuous flame image Since the fluctuation frequency of the flame is detected from the time series change of the luminance sum that exceeds the predetermined threshold in the entire image or the flame candidate area, the fluctuation frequency that changes depending on the size of the flame is accurately detected by processing the flame image. Make it possible.

The frequency detecting means includes light receiving element means for receiving infrared light or visible light from a flame existing in the monitoring region and outputting a light receiving signal, and detects a fluctuation frequency from the light receiving signal output from the light receiving element means. In this case, it is possible to detect the fluctuation frequency directly from the flame image from the detection signal of the light receiving element means, and it is not necessary to detect the time series change in luminance as in the flame image, Becomes easier.

Explanatory drawing which showed the monitoring area | region which installed the fire detection system of this invention A graph showing the variation of the fluctuation frequency of the flame with respect to the equivalent diameter of the flame A graph showing the characteristics of the approximate equation showing the relationship between the equivalent diameter of the flame and the fluctuation frequency Explanatory drawing showing the relationship between the expected angle of flame and the flame distance for the horizontal monitoring area of the surveillance camera The block diagram which showed embodiment of the fire detection system by this invention Explanatory drawing showing two flames with the same angle of view and different flame distance and size in comparison with the horizontal monitoring area of the surveillance camera Explanatory drawing showing two flames with the same size but different angle of view and flame distance in comparison with the horizontal monitoring area of the surveillance camera Flow chart showing fire detection process

[Outline of fire detection system]
FIG. 1 is an explanatory diagram showing a monitoring area where a fire detection system according to the present invention is installed.

  As shown in FIG. 1, a monitoring camera 10 functioning as an imaging unit is installed in the monitoring area 16, for example, obliquely downward, and the monitoring area 16 is imaged to obtain an observation image.

  An observation image of the monitoring area 16 captured by the monitoring camera 10 is transmitted to the image processing apparatus 12 installed in a manager's room or the like via a transmission path, and a fire occurs in the monitoring area 16 from the fire source 18 to the flame 20. , The observation image in which the flame 20 is present (hereinafter referred to as “flame image”) is processed by image processing, the equivalent diameter is calculated from the fluctuation frequency of the flame 20, and a fire is determined. A distance up to 20 (hereinafter referred to as “flame distance”) L is calculated, a fire detection signal is sent to the fire alarm device 14 together with information indicating the equivalent diameter and / or flame distance of the flame, a fire alarm is output, and the magnitude of the flame, for example The equivalent diameter and flame distance are displayed.

[Detection principle]
(Flame fluctuation frequency and equivalent diameter)
It is known that the frequency of flame fluctuation decreases as the flame increases. Ono et al. Experimentally determined the frequency of flame fluctuation with respect to the equivalent diameter of the flame. According to this, as the combustion diameter of the flame increases to 4 to 50 cm, the frequency of the flame changes in a low direction to 10 to 2 Hz.

  FIG. 2 is a graph showing the experimental results by Ono et al., Where the horizontal axis represents the flame equivalent diameter and the vertical axis represents the flame fluctuation frequency. According to the research by Ono et al., The relationship between the flame equivalent diameter and the fluctuation frequency in FIG. 2 shows almost the same tendency, although there is a slight deviation depending on the material.

In the present invention, an approximate straight line is obtained from the experimental results shown in the graph of FIG. The approximate straight line equation obtained on the logarithm of the fluctuation frequency and the equivalent diameter logarithmically by the least square method is as follows: the fluctuation frequency of the flame is Z (Hz) and the equivalent diameter of the flame is 2R (cm).
Z = 18.622 (2R) ^-0.554 (Formula 1)
It becomes. By converting this and obtaining the equivalent diameter 2R,
2R = (Z / 18.622) ^{1 / −0.554} (Formula 2)
It becomes.

  FIG. 3 is a logarithmic graph showing the relationship between the equivalent diameter of the flame and the fluctuation frequency according to the approximate expression of Expression 2. Therefore, the magnitude of the flame can be estimated by obtaining the equivalent diameter 2R of the generated flame from the observed flame frequency Z according to Equation 2.

(Flame distance)
If the equivalent diameter 2R of the flame can be obtained from the fluctuation frequency of the flame, the flame distance L from the monitoring camera 10 shown in FIG. 1 to the flame 20 can be obtained based on the equivalent diameter 2R.

  FIG. 4 is an explanatory diagram showing the relationship between the expected angle of flame and the flame distance in the horizontal monitoring area of the monitoring camera.

  In order to calculate the flame distance L, first, the prospective angle θ of the flame 20 existing in the observed flame image is obtained. Here, the prospective angle means an angle of view when a specific object is imaged from the surveillance camera.

  The size of the flame in the flame image captured by the monitoring camera 10 represents the expected angle of the flame. Even if the expected angle is the same, the size of the flame image 20 depends on the distance from the monitoring camera 10. Will change.

Now, assuming that the screen size of the monitoring camera 10 is (horizontal) × (vertical) = 1280 × 960 pixels and the horizontal one-side angle of the imaging region 15 of the monitoring camera 10 is 45 °, the expected angle α per pixel is α = 45 ° / 640 = 0.070 (° / pixel)
It becomes. In the following description, a pixel may be referred to as a pixel, and (pxl) may be expressed as a unit of the number of pixels.

Here, assuming that the number of pixels in the horizontal direction of the flame candidate region that is equal to or higher than the predetermined threshold luminance of the flame image captured by the monitoring camera 10 is X pixels, the expected angle θ of the flame is
θ = αX (°) (Formula 3)
It becomes.

When the expected angle θ of the flame is obtained in this way, the flame distance L from the monitoring camera 10 to the flame 20 based on the equivalent diameter 2R of the flame obtained from the equation 2 is
L = R / tan (θ / 2) (Formula 4)
It becomes.

  FIG. 4 shows an example in which the flame 20 exists on the optical axis of the surveillance camera 10, but the same size flame 20a existing at a position away from the optical axis having the same flame distance L is also used for the light. Similarly to the flame 20 on the shaft, the equivalent diameter 2R of the flame and the flame distance L can be calculated.

[Fire detection system]
(Functional configuration of fire detection system)
FIG. 5 is a block diagram showing an outline of the functional configuration of the fire detection system according to the present invention. As shown in FIG. 5, the fire detection system includes a monitoring camera 10 and an image processing device 12. The image processing device 12 includes a computer circuit having a CPU, a memory, various input / output ports and the like as hardware. Consists of.

  The image processing apparatus 12 functions as a flame image determination unit 24 that functions as a flame pixel measurement unit, a prospective angle detection unit 26 that functions as a prospective angle detection unit, and a fluctuation frequency detection unit as functions realized by execution of a program by the CPU. A frequency detecting unit 28, an equivalent diameter calculating unit 30 functioning as an equivalent diameter calculating unit, a distance calculating unit 32 functioning as a distance calculating unit, and a fire determining unit 34 functioning as a fire determining unit. The transmission unit 22 uses an appropriate transmission interface for receiving image data captured by the monitoring camera 10.

  The monitoring camera 10 that functions as an imaging unit transmits image data of a monitoring area, for example, 30 frames per second as moving image data by transmission control of the transmission unit 22, and stores it in a memory (not shown) provided in the image processing device 12. . In the present embodiment, for example, the surveillance camera 10 having a resolution of 1280 pixels in horizontal pixels and 960 pixels in vertical pixels is used.

  The flame image determination unit 24 determines a flame image in which a flame exists from the image captured by the monitoring camera 10. The determination of the flame image by the flame image determination unit 24 is performed by, for example, detecting a flame candidate area formed by a set of pixels having a predetermined threshold luminance or higher that can be determined as a flame, and the horizontal pixel count X of the detected flame candidate area Is determined to be a flame image in which a flame is present, and the horizontal pixel number X of the candidate fire area in the determined flame image is output to the prospective angle detector 26 and the flame determined to the frequency detector 28 An image is sent to instruct the fluctuation frequency detection operation.

  The expected angle detection unit 26 calculates the expected angle θ of the flame from the horizontal pixel number X of the flame candidate area in the flame image determined by the flame image determination unit 24, and outputs it to the distance calculation unit 32.

  On the other hand, the frequency detector 28 detects the fluctuation frequency Z of the flame present in the monitoring area. The frequency detection unit 28 is, for example, a time series of the number of pixels having a luminance equal to or higher than a predetermined threshold in the entire image or the flame candidate region determined by the flame image determination unit 24 from the continuous flame images determined by the flame image determination unit 24. The fluctuation frequency Z of the flame is detected from the change based on, for example, Fourier transform.

  Further, as another embodiment of the frequency detection unit 28, a luminance sum that is equal to or greater than a predetermined threshold in the entire image or the flame candidate region determined by the flame image determination unit 24 from the continuous flame images determined by the flame image determination unit 24. For example, the fluctuation frequency Z of the flame is detected from the time series change by Fourier transform.

  Here, the detection of the flame fluctuation frequency Z based on the luminance of the flame candidate region is performed by dividing the entire image into a plurality of blocks and detecting the luminance by paying attention to the divided blocks where the flame is detected. Can be increased.

  As another embodiment of the frequency detector 28, a light receiving element such as a pyroelectric element or a photodiode that receives infrared light or visible light from a flame existing in the monitoring region and outputs a light receiving signal is provided. The fluctuation frequency Z is detected from the received light signal output from the element by Fourier transform or the like.

  As another embodiment of the frequency detection unit 28, the area and height of a flame candidate region formed by a set of pixels having a predetermined threshold luminance or higher determined by the flame image determination unit 24, The fluctuation frequency of the flame may be detected by a Fourier transform from a time-series change such as an expected angle of direction.

  The equivalent diameter calculation unit 30 calculates the equivalent diameter 2R of the flame from the flame fluctuation frequency Z detected by the frequency detection unit 28 based on the above equation 2.

  The distance calculation unit 32 calculates the flame distance from the monitoring camera 10 to the flame based on the equation 4 based on the expected angle θ of the flame detected by the expected angle detection unit 26 and the equivalent diameter 2R of the flame calculated by the equivalent diameter calculation unit 30. L is calculated.

  The fire determination unit 34 determines a fire when the equivalent diameter 2R of the flame calculated by the equivalent diameter calculation unit 30 is equal to or greater than a predetermined threshold value. Otherwise, the fire determination unit 34 determines a non-fire and determines a fire. The fire detection signal is output to the external fire alarm device 14 together with the data of the equivalent diameter 2R calculated by the equivalent diameter calculator 30 and / or the flame distance L calculated by the distance calculator 32, and a fire alarm is output and, for example, a flame The equivalent diameter 2R and / or the flame distance L are displayed.

  By grasping the size and distance of the flame in the monitoring area in such a fire alarm facility 14, it is possible to accurately determine the state of the fire, for example, whether it is in the initial fire stage or whether the fire has spread over a certain period of time. It is possible to grasp the situation and explain the situation to the initial fire fighting and the fire brigade that arrived.

[Specific examples of flame detection]
Now, assuming that the number X of horizontal pixels of the flame candidate region determined by the flame image determination unit 24 is X = 57 pixels, the expected angle θ calculated by the expected angle calculation unit 26 is θ = 0.070 × 57 = 4 °
It becomes.

At this time, assuming that the flame fluctuation frequency Z detected from the flame image by the frequency detection unit 28 is, for example, Z = 2 (Hz), the equivalent diameter calculation unit 30 sets the equivalent diameter 2R of the flame to 2R = 56 according to the above equation 2. .1 (cm)
And calculate.

Further, the distance calculation unit 32 calculates the flame distance L from the monitoring camera 10 to the flame by L = 1607 (cm) according to the equation 4.
And calculate.

  If the predetermined threshold value 2Rth for determining fire from the equivalent diameter 2R of the flame calculated by the equivalent diameter calculation unit 30 is 2R = 10 cm, for example, the fire determination unit 34 calculates the equivalent diameter 2R = 56.1 ( cm) is equal to or greater than the threshold value 2Rth = 10 (cm), so it is determined that the flame is a fire.

On the other hand, based on the calculated equivalent flame diameter 2R = 56.1 (cm), the distance calculating unit 32 sets the flame distance L to L = 1607 (cm) according to the above equation 4.
And calculate.

  For this reason, the fire determination unit 34 outputs a fire detection signal together with the equivalent diameter data and flame distance data of the flame to the fire alarm equipment 14, and together with the output of the fire alarm, the equivalent diameter 2R = 56.1 (cm) and the flame distance L = 1607 (cm) is displayed, and the status of the fire that occurred in the monitoring area can be confirmed.

On the other hand, when the horizontal pixel number X of the flame image is X = 57 pixels and the prospective angle θ is the same θ = 4 °, the fluctuation frequency Z of the flame image is, for example, Z = 8 (Hz). In this case, the equivalent diameter 2R of the flame is 2R = 4.6 (cm) from Equation 2
The flame distance L from the monitoring camera 10 to the flame is L = 13.2 (cm) from the equation 4.
It becomes.

  In this case, the equivalent diameter 2R = 4.6 (cm) of the flame is less than the threshold value Rth = 10 (cm) for determining fire, and it is determined that there is no fire. Further, since the flame distance L at this time is L = 13.2 (cm), which is close to the monitoring camera 10, the flame determined as non-fire is a small flame caused by a lighter or the like near the monitoring camera 10. I understand that.

(Flames with the same angle of view and different distances and sizes)
FIG. 6 is an explanatory view showing two flames having the same angle of view and different flame distances and sizes in comparison with the horizontal monitoring area of the monitoring camera.

As shown in FIG. 6, the prospective angle θ of the flame 20-1 generated at a position distant from the monitoring camera 10 is calculated from the above equation 3 based on the horizontal pixel number X of the fire candidate area in the flame image captured by the monitoring camera 10. Is the angle to be. The flame 20-1 is equivalent diameters 2R ₁ calculated by the formula 2 from the detection of the fluctuation frequency Z, flame distance L1 is calculated from the equation 3, and the equivalent diameter 2R ₁ flame 20-2 is equal to or more than the threshold The fire is judged by becoming.

For example, a small flame 20-2 generated near the monitoring camera 10 is used as the flame having the same prospective angle θ as that of the flame 20-1, and is smaller by the above equation 2 from the detection of the fluctuation frequency Z of the flame 20-2. The equivalent diameter 2R ₂ is detected, and a short flame distance L2 is calculated from the above equation 3. For example, the non-fire is determined when the equivalent diameter 2R _{2 of the} flame 20-2 is less than the threshold value.

As described above, when the monitoring camera 10 captures the flame 20-1 and the flame 20-2 having the same angle of view, the sizes of the flame 20-1 and the flame 20-2 existing in the respective images are the same. Then, by estimating the equivalent diameters 2R ₁ and 2R ₂ from the different fluctuation frequencies, the fire is determined from the flame 20-1 having the same angle of view and the same size on the screen. For -2, non-fire can be determined.

(Flames with the same size but different angles and distances)
FIG. 7 is an explanatory view showing two flames having the same size but different angles of view and flame distances in comparison with the horizontal monitoring area of the monitoring camera.

  As shown in FIG. 7, the flame 20-1 and the flame 20-2 have the same equivalent diameter 2R because the fluctuation frequency Z is the same. However, since the distance from the monitoring camera 10 is different, The number of horizontal pixels of the existing fire candidate area is different from X1 and X2, and thus the prospective angle is different from θ1 and θ2. As described above, when the prospective angles of the flames 20-1 and 20-2 are different from θ1 and θ2, the fire source distance calculated by the above equation 4 from the same equivalent diameter 2R of the flames 20-1 and 20-2 is L1, L2. Is different.

  In this way, even in the case of a fire of the same size, it can be seen that the fire source distance is different due to the difference in the number of horizontal pixels of the flame candidate area present in the flame image, and in addition to the output of the fire alarm based on the fire detection signal By displaying the equivalent diameter of the flame and the flame distance, it is possible to appropriately grasp the occurrence of the fire, for example, that the fire is in the initial stage.

[Fire detection processing]
FIG. 8 is a flowchart showing a fire detection process according to the embodiment of FIG. As shown in FIG. 8, first, in step S1 (hereinafter, “step” is omitted), a threshold Xth of the number of horizontal pixels for determining a flame candidate area in the flame image, a prospective angle α per pixel in the monitoring camera 10, an equivalent of flame A fire determination threshold value 2Rth for determining a fire from the diameter is set as a constant.

  Subsequently, the process proceeds to S2, where a flame candidate area formed by a pixel set having a predetermined luminance or higher is detected from the observation image captured by the monitoring camera 10, the horizontal pixel number X of the flame candidate area is obtained, and a predetermined threshold value Xth is obtained in S3. If it is above, it determines with a flame image and progresses to S4.

  In S4, a prospective angle θ is calculated from the horizontal pixel number X of the flame detected in S2, and then the process proceeds to S5, where a flame fluctuation frequency Z is detected from successive flame images. Subsequently, the process proceeds to S6, where the equivalent diameter 2R of the flame is calculated from the fluctuation frequency Z, and then, the process proceeds to S7, where the flame distance L is calculated from the expected angle θ of the flame and the equivalent diameter 2R.

  In S8, if the equivalent flame diameter 2R is equal to or greater than the fire threshold value 2Rth, it is determined that there is a fire. In S9, a fire detection signal is output together with data on the equivalent flame diameter 2R and / or the flame distance L.

[Application of flame distance and equivalent diameter]
(Application based on estimation of fire source distance)
By using the flame distance L that is output when the image processing apparatus 12 determines a fire in the above embodiment, the position of the flame 20 in the monitoring region 10 of FIG. 1 can be specified.

  For example, since the inclination angle β with respect to the vertical direction of the optical axis from the monitoring camera 10 is determined, the horizontal distance to the flame can be calculated as Lsinβ, and the height from the flame 20 to the monitoring camera 10 is calculated as Lcosθ. Further, as shown in FIG. 4, the horizontal angle with respect to the optical axis of the flame 20 viewed from the monitoring camera 10 is known, so that the flame in a three-dimensional coordinate space with a predetermined position in the monitoring space as the origin is obtained. Twenty positions can be determined.

  When the position of the flame 20 in the monitoring region 16 is obtained in this manner, for example, proper fire extinguishing can be performed by sending flame position information to the fire extinguishing device. For example, when a sprinkler fire extinguishing equipment having a remotely controllable sprinkler head is installed, only the sprinkler head close to the fire source position is operated. A movable fire extinguishing robot is automatically moved to near the flame. In addition, the position of the water gun is controlled.

  In an automated large space facility such as a factory or an automatic warehouse, knowing the flame position enables appropriate measures such as stopping the operation of the target equipment and operating the attached fire extinguishing device.

  In addition, even during the initial fire extinguishing activity by a person, it is possible to go straight to the flame position, and a quick response is possible. In addition, an accurate route can be set for evacuation guidance. Furthermore, it is possible to provide information on the exact location of the fire point to fire fighters, and to improve the fire extinguishing efficiency, leading to life saving and damage suppression. Besides this, various measures can be taken by knowing the flame distance.

(Application based on estimation of equivalent diameter of flame)
The following application is possible by using the equivalent diameter 2R of the flame output when the image processing apparatus 12 determines a fire in the above embodiment, that is, the estimated value of the size (scale) of the flame. .

  For example, it is possible to use an appropriate extinguishing agent that matches the size (scale) of the flame, and it is possible to reliably extinguish with a small amount of extinguishing agent, leading to prevention of pollution and property protection.

  In addition, the sudden expansion of the equivalent diameter of the flame reveals a rapid expansion of the fire, and it is possible to start evacuation guidance in an appropriate manner at an early stage according to the size of the flame, leading to lifesaving. Besides this, various measures can be taken by knowing the size of the flame.

[Modification of the present invention]
(Equivalent diameter of flame)
In the above embodiment, when a fire is determined from the flame image, in addition to the fire detection signal, the estimated flame equivalent diameter 2R and / or the distance data of the flame distance L are output. However, only the fire detection signal is output. May be output.

(Flame non-fire judgment)
In the above embodiment, when the equivalent diameter of the flame is equal to or greater than a predetermined threshold value, it is determined as a fire, and data indicating the flame distance and / or the equivalent diameter is output together with the fire detection signal. In addition, data indicating the flame distance and / or equivalent diameter is output together with the non-fire flame detection signal, for example, a warning warning is output for the lighter flame by detecting the flame, and flame is prohibited by displaying the flame distance and equivalent diameter. It is possible to notify the use of fire in the monitoring area and to deal with, for example, arson.

(Surveillance camera)
In the above embodiment, the case where the viewing angle θ of the surveillance camera is set to 45 ° is taken as an example in order to simplify the explanation, but the embodiment can be applied to an appropriate viewing angle. Further, the number of horizontal and vertical pixels of the surveillance camera is not limited to (640 × 480) pixels, and those having an appropriate resolution can be used.

(Image processing device)
In the above embodiment, the surveillance camera and the image processing apparatus are separately arranged and connected by a transmission path, but an apparatus in which both are integrated may be used.

(Other)
The present invention is not limited to the above-described embodiment, includes appropriate modifications without impairing the object and advantages thereof, and is not limited by the numerical values shown in the above-described embodiment.

10: surveillance camera 12: image processing device 14: fire alarm equipment 15: imaging area 16: monitoring area 18: fire source 20, 20A, 20-1, 20-2: flame 22: transmission unit 24: flame image determination unit 26 : Expected angle calculation unit 28: Frequency detection unit 30: Equivalent diameter calculation unit 32: Distance calculation unit 34: Fire determination unit

Claims (16)

Imaging means for photographing the surveillance area;
Flame image determination means for determining a flame image in which a flame exists from an image captured by the imaging means;
Expected angle detection means for detecting the expected angle of the flame from the flame image determined by the flame image determination means,
A frequency detection means for detecting the fluctuation frequency of the flame present in the monitoring area;
Equivalent diameter calculating means for calculating an equivalent diameter of the flame based on the fluctuation frequency of the flame detected by the frequency detecting means;
A distance calculating means for calculating a flame distance from the imaging means to the flame based on the expected angle of the flame detected by the expected angle detecting means and the equivalent diameter of the flame calculated by the equivalent diameter calculating means;
The flame calculated by the distance calculating means when the equivalent diameter of the flame calculated by the equivalent diameter calculating means is determined to be a fire when the equivalent diameter is greater than or equal to a predetermined threshold and otherwise determined to be a non-fire, and when a fire is determined Fire determining means for outputting a fire detection signal together with information indicating the distance and / or the equivalent diameter of the flame calculated by the equivalent diameter calculating means;
A fire detection system characterized by the provision of
2. The fire detection system according to claim 1, wherein the equivalent diameter calculation means sets the equivalent diameter 2R of the flame to 2R = (Z / 18.622) ^{1 / −0.554} when the fluctuation frequency is Z.
Fire detection system characterized by calculating by
The fire detection system according to claim 1, wherein the distance calculation means calculates the flame distance L when the equivalent diameter of the flame is 2R and the expected angle of the flame is θ.
L = R / tan (θ / 2)
Fire detection system characterized by calculating by
2. The fire detection system according to claim 1, wherein the flame image determination unit detects a flame candidate region which is a set of pixels having a predetermined luminance or higher from an image captured by the imaging unit, and a horizontal pixel of the flame candidate region. A fire detection system, wherein a flame image is determined when the number is equal to or greater than a predetermined threshold.
2. The fire detection system according to claim 1, wherein the prospective angle detecting means includes a prospective angle α per pixel determined by a horizontal viewing angle and the number of horizontal pixels of the imaging means, and a horizontal flame measured from the flame image. A fire detection system that detects an expected angle θ of a flame by multiplication with the number of pixels X.
5. The fire detection system according to claim 4, wherein the frequency detection means detects a fluctuation of the flame from a continuous flame image based on a time-series change in the number of pixels having a luminance equal to or higher than a predetermined threshold in the entire image or the flame candidate region. A fire detection system characterized by detecting the frequency.
5. The fire detection system according to claim 4, wherein the frequency detection means calculates a flame fluctuation frequency from a continuous flame image based on a time-series change in luminance sum that is equal to or greater than a predetermined threshold in the entire image or the flame candidate region. A fire detection system characterized by detection.
2. The fire detection system according to claim 1, wherein the frequency detection means includes light receiving element means for receiving infrared light or visible light from a flame existing in a monitoring area and outputting a light reception signal. A fire detection system that detects a fluctuation frequency from a light-receiving signal output from a light source.
A flame image determination step for determining a flame image in which a flame is present from an image captured by an imaging unit that captures a monitoring region;
A prospective angle detection step of detecting a prospective angle of the flame from the flame image determined in the flame image determination step;
A frequency detection step for detecting the fluctuation frequency of the flame present in the monitoring area;
An equivalent diameter calculating step for calculating the equivalent diameter of the flame based on the fluctuation frequency of the flame detected in the frequency detecting step;
A distance calculating step for calculating a flame distance from the imaging means to the flame based on the estimated angle of the flame detected in the estimated angle detection step and the equivalent diameter of the flame calculated in the equivalent diameter calculating step;
The flame calculated in the distance calculating step when the equivalent diameter of the flame calculated in the equivalent diameter calculating step is determined to be a fire when the equivalent diameter of the flame is equal to or larger than a predetermined threshold and otherwise determined to be a non-fire. Fire determining means for outputting a fire detection signal together with information indicating the distance and / or the equivalent diameter of the flame calculated in the equivalent diameter calculating step;
The fire detection method characterized by providing.
10. The fire detection method according to claim 9, wherein the equivalent diameter calculation step sets the equivalent diameter 2R of the flame to 2R = (Z / 18.622) ^{1 / −0.554} when the fluctuation frequency is Z.
The fire detection method characterized by calculating by the above.
10. The fire detection method according to claim 9, wherein the distance calculating step sets the flame distance L when the equivalent diameter of the flame is 2R and the expected angle of the flame is θ.
L = R / tan (θ / 2)
The fire detection method characterized by calculating by the above.
The fire detection method according to claim 9, wherein the flame image determination step detects a flame candidate region that is a set of pixels having a predetermined luminance or higher from an image captured by the imaging unit, and detects a flame of the flame candidate region. A fire detection method characterized by determining a flame image when the number of horizontal pixels is equal to or greater than a predetermined threshold.
10. The fire detection method according to claim 9, wherein the expected angle detection step includes a predicted angle α per pixel determined by a horizontal viewing angle and a number of horizontal pixels of the imaging means, and a horizontal flame measured from the flame image. A fire detection method characterized by detecting an expected flame angle θ by multiplication with the number of pixels X.
13. The fire detection method according to claim 12, wherein the frequency detection step includes, from the continuous flame image, a flame fluctuation based on a time-series change in the number of pixels having a luminance equal to or higher than a predetermined threshold in the entire image or the flame candidate region. A fire detection method characterized by detecting a frequency.
13. The fire detection method according to claim 12, wherein the frequency detection step calculates a flame fluctuation frequency from a continuous flame image based on a time series change of a luminance sum that is equal to or greater than a predetermined threshold in the entire image or the flame candidate region. A fire detection method characterized by detecting.
  10. The fire detection method according to claim 9, wherein the frequency detection step includes light receiving element means for receiving infrared light or visible light from a flame existing in a monitoring area and outputting a light reception signal. A fire detection method, wherein a fluctuation frequency is detected from a light reception signal output from a light source.

Images