JP7288766B2 - Smoke detection device and smoke detection method - Google Patents

Description

The present invention relates to a smoke detection device and a smoke detection method for detecting smoke caused by a fire from an image of a surveillance area captured by a camera.

2. Description of the Related Art Conventionally, various apparatuses and systems have been proposed that detect a fire by performing image processing on an image of a surveillance area captured by a surveillance camera.

In such a smoke detector, early detection of fire is important from the viewpoint of initial extinguishing and evacuation guidance.

For this reason, in the conventional apparatus (Patent Document 1), as phenomena caused by the smoke accompanying the fire from the image, the decrease in transmittance or contrast, the convergence of the luminance value to a specific value, the narrowing of the luminance distribution range and the dispersion of luminance , a change in the average luminance value due to smoke, a decrease in the total amount of edges, and an increase in intensity in the low-frequency band.

However, in the conventional fire detection system that detects fire from images of smoke accompanying fires, the entire image captured by a surveillance camera is processed to detect characteristic changes due to smoke to determine fires. Therefore, there is a problem that the processing load for judging the fire from the entire image increases and the processing takes time.

In order to solve this problem, in Patent Document 2, there are areas of high density and areas of low density inside the smoke that rises from the fire, and the areas of high density smoke appear sequentially and move upward. Therefore, a class is arranged according to the rule of a particle filter, for example, in the part of smoke with high density from the captured image, and position information is obtained by tracking each class for each image. Detecting changes moving upward enables reliable fire judgment.

JP 2008-046916 A JP 2017-191544 A

By the way, in the method of arranging classes in a smoke area using such a conventional particle filter and tracking them, when the classes are arranged by scattering particles in the target image, the particles are concentrated in one place, Since there is a problem that it is not possible to arrange and track multiple classes, we irradiate the monitoring area with many spotlights and capture the image of the area where the spotlights scattered in the image of the monitoring area captured by the surveillance camera are captured. By extracting and tracking, it is possible to generate and track multiple classes.

However, a special spot irradiation device is required to irradiate a large number of spot lights on the monitoring area, and an infrared light source is used to make it difficult to see that the spot lights are hitting the area. There is a problem that it can not be used in , and lacks versatility.

In the present invention, for an image captured in a visible lighting environment, class tracking by a particle filter prevents particles from concentrating in one place, and arranges a plurality of classes indicating a smoke area in time series, which is unique to smoke. It is an object of the present invention to provide a smoke detection device and a smoke detection method capable of detecting a fire at an early stage.

(smoke detector)
The present invention is a smoke detection device comprising:
imaging means for sequentially capturing images of the monitored area;
a class tracking means for repeating a process of arranging classes by obtaining the center of gravity of the smoke area in each image captured by the imaging means and tracking the classes to detect movement trajectories;
fire determination means for determining a fire by detecting a predetermined characteristic change due to smoke from the movement trajectory of the class detected by the class tracking means;
is provided,
Class tracking means
For each image captured by the imaging means, a predetermined number of particles are scattered in the image according to a predetermined rule, and a predetermined class is arranged by obtaining the center of gravity of the scattered predetermined particles as the center of gravity of the smoke area. do the processing,
In the case of an image in which no class has been placed, a new class is placed by scattering a predetermined number of particles over the entire image area,
In the case of an image in which classes are arranged, for each class that has already been arranged, a predetermined number of particles are scattered around a predetermined position in the area above the center of gravity of the class to be processed. By scattering, the particles of each class are rearranged to rearrange the classes, after rearranging all the classes, a new class is arranged for the area excluding the classes ,
When rearranging a class, change the distance from the center of gravity of the particles when the class was last placed to the scattering center, depending on the movement speed corresponding to the color of the smoke, or the movement speed of the class in the past. It is characterized by

(Smoke detection method)
Another aspect of the invention is a smoke detection method comprising:
The imaging means sequentially captures images of the monitored area,
The class tracking means repeats the process of finding the center of gravity of the smoke area in the image and arranging the classes for each image generated by the imaging means, tracking the classes to detect movement trajectories,
The fire determination means detects a characteristic predetermined change due to smoke from the moving trajectory of the class detected by the class tracking means to determine a fire,
Class tracking means
For each image captured by the imaging means, a predetermined number of particles are scattered in the image according to a predetermined rule, and a predetermined class is arranged by obtaining the center of gravity of the scattered predetermined particles as the center of gravity of the smoke area. do the processing,
In the case of an image in which no class has been placed, a new class is placed by scattering a predetermined number of particles over the entire image area,
In the case of an image in which classes are arranged, for each class that has already been arranged, a predetermined number of particles are scattered around a predetermined position in the area above the center of gravity of the class to be processed. By scattering, the particles of each class are rearranged to rearrange the classes, after rearranging all the classes, a new class is arranged for the area excluding the classes ,
When rearranging a class, change the distance from the center of gravity of the particles when the class was last placed to the scattering center, depending on the movement speed corresponding to the color of the smoke, or the movement speed of the class in the past. It is characterized by

Other features of the smoke detection method are the same as those of the smoke detection device described above.

(basic effect)
In the smoke detection apparatus of the present invention, the imaging means for sequentially imaging the image of the monitoring area, and the process of arranging the classes for each image generated by the imaging means are repeated, and the classes are tracked to determine the movement trajectory. Class tracking means for detecting and fire judgment means for judging a fire by detecting a predetermined characteristic change due to smoke from the moving trajectory of the class detected by the class tracking means are provided. In the case of an unarranged image, a new class is arranged for the entire image area, and in the case of an image with a class arranged, the class is rearranged for the area above the center of gravity of the class, and all After rearranging the classes, a new class is arranged for the area excluding the class. As an example, by using a particle filter, the class tracking means is imaged by the imaging means. For each image obtained, a predetermined number of particles are scattered in the image according to a predetermined rule, each particle is weighted by the likelihood of being smoke-like, and particles with a likelihood less than a predetermined value are deleted. The process of arranging classes while leaving particles of a predetermined value or more is repeated, and the classes are tracked to detect movement trajectories. A new class is randomly scattered in the image, and in the case of the image in which the class is placed, a predetermined number of particles are placed at a predetermined probability density centering on a predetermined position above the particles left in the previous class placement. Scatter the classes according to the distribution and rearrange the classes. After rearranging all the classes, randomly scatter the specified number of particles in the area excluding the classes and arrange the new classes, so the classes are rearranged. In this case, the classes are rearranged by scattering them around the position above the particles left last time. By arranging such classes, it is possible to continuously track the arrangement and rearrangement of classes without interfering with each other, for a mass of smoke with a certain concentration that rises due to the rising smoke vortex. It is possible to reliably determine that it is a fire by capturing the unique characteristics of the smoke from the moving trajectory of the class.

Also, since smoke is generally difficult to descend, when rearranging a class, it is necessary to place particles in an area slightly above the position of each particle when the class was last arranged, and in the area of particles belonging to other classes. By scattering particles in an area that does not include a predetermined range centered on the center of gravity and rearranging the classes, it is possible to reliably prevent the overlapping of particles between classes, and the classes that accurately follow the rising mass of smoke. A movement trajectory can be detected.

(Effect of smoke detection by particle filter using difference image)
Further, a difference image generating means for generating a difference image for each of the plurality of images captured by the image capturing means is provided, and the class tracking means generates a particle-based class for each difference image generated by the difference image generating means. Since classes are tracked by placement and rearrangement to detect movement trajectories, it is possible to emphasize only moving parts such as smoke in the image and prevent erroneous detection.

(Effect of rearrangement based on class priority)
Also, when the classes are rearranged, the class tracking means sets a different priority for each class, rearranges the classes in order from the higher class, and among the plurality of particles scattered for the class rearrangement, already Since the center of gravity of the remaining particles is calculated from the likelihood of the remaining particles, excluding the particles that overlap with the smoke area of the relocated upper class, the classes are overlapped corresponding to the continuously rising mass of smoke. It can accurately detect the movement trajectory of the class without repositioning.

(Effect of scattering center position of rearranged particles)
Also, the class tracking means corresponds the distance from the position of each particle when the class was placed last time to the center where the particles are scattered above to the moving speed corresponding to the color of the smoke or the moving speed of the past class. Since the movement of the smoke between images changes depending on the speed at which the smoke rises, if the movement speed of the smoke is high, when rearranging the class, from the center of gravity of the class to the center of scattering particles By increasing the distance of , and shortening it if the moving speed of the smoke is low, it is possible to scatter particles in the optimal area for class rearrangement according to the moving speed of the smoke, and the accuracy of class tracking by class rearrangement can be increased to detect more accurate class movement trajectories.

For example, black smoke from combustion fire rises at a high speed, so the distance to the center of particle scattering is increased. White smoke from smoldering fire rises slowly, so the distance to the center of particle scattering is increased Shortening allows for accurate class repositioning as the smoke moves.

(The effect of removing particles that move the class downwards)
In addition, the class tracking means determines the likelihood of remaining particles, excluding particles existing in an area below the center of gravity of the class, among the particles scattered for class rearrangement, or the horizontal line passing through the center of gravity of the class. Since the center of gravity of the particles including the particles existing in the area below the center of gravity of the moved class is calculated based on the likelihood of particles existing in the area below the center of gravity of the moved class, the center of gravity is calculated and relocated. Particles scattered below have a likelihood of zero (zero weight) so that they do not contribute to the calculation of the center of gravity of the class to be rearranged. A smoke-specific class of motion trajectories can be detected.

(Effect of class disappearance and reuse)
In addition, the class tracking means is capable of arranging new classes up to a predetermined maximum number, and when a stagnation of classes is detected, the stagnation classes are deleted and used for arranging new classes. Even if you limit the maximum number of classes you can place, you can keep smoke tracking smooth by placing classes in a circular fashion.

(Particle scattering area when the staying class is erased)
When the retained class is erased, the class tracking means randomly scatters a predetermined number of particles under the area where other classes exist, or under the erased class when no other classes exist. As the smoke area rises and disappears from the screen, the classes remaining at the top of the screen are erased, and the newly appearing smoke area affects other classes. You can place your class early without any problems and resume smoke tracking smoothly.

(Likelihood of particles)
In addition, since the class tracking means performs weighting using the pixel value of the particle as the likelihood, it is possible to appropriately obtain the center of gravity of the smoke area by weighting the particle by setting the likelihood corresponding to the smoke density.

(Effect of differential image generation)
Further, the differential image generating means detects the maximum value and minimum value of each pixel in a predetermined number of images including the current image, generates a maximum value image and a minimum value image among a plurality of images, Since the difference image is generated from the image and the minimum value image, smoke can be visualized from the moving image captured by the imaging means by generating the difference image, and the smoke can be reliably tracked by the particle filter. In addition, by changing the number of images (number of frames) for generating the difference image, even if the smoke has different flow velocities, a difference image in which the movement of the smoke between the difference images is substantially constant is generated. Regardless of the movement speed, the center for scattering particles can be set to a substantially constant distance from the center of gravity of the class and relocated when the class is relocated.

The effect of the smoke detection method is the same as that of the smoke detection device described above.

Explanatory drawing showing a monitoring area in which the smoke detection device of the present invention is installed Block diagram showing an outline of the functional configuration of the smoke detection device Explanatory diagram showing a frame image captured by a surveillance camera and its difference image Explanatory diagram showing the procedure up to placement of a new class by scattering particles in the first smoke area. Explanatory diagram showing the particle scattering area during class rearrangement FIG. 5 is an explanatory diagram showing placement of new classes following the class rearrangement of FIG. 4 ; Explanatory diagram showing time-series changes in class placement and class movement trajectories for time-series difference images Explanatory diagram showing rearrangement of stagnant classes Flowchart showing smoke detection processing by the smoke detection device of FIG. Flowchart showing details of class tracking by particle filter in step S2 of FIG.

[Basic concept of the embodiment]
FIG. 1 is an explanatory diagram showing a monitoring area in which the smoke detection device of the present invention is installed, and FIG. 2 is a block diagram showing an outline of the functional configuration of the smoke detection device.

The basic concept of this embodiment is to detect a fire at an early stage by inputting a moving image captured by a surveillance camera 10 functioning as an imaging means to a smoke detection device 12 . From the images input to the smoke detection device 12, a differential image generating unit 24, which functions as differential image generating means, generates a differential image for each of a plurality of images, and a class tracking unit, which functions as class tracking means using a particle filter. 26, for each differential image, a plurality of particles are scattered in the image according to a predetermined rule, each particle is weighted by the likelihood of indicating smoke-likeness, and the center of gravity of the smoke area is determined to arrange classes. The process is repeated to track the class and detect the movement trajectory. Subsequently, the fire judgment unit 28 functioning as fire judgment means detects a characteristic predetermined characteristic of smoke from the movement trajectory of the class detected by the class tracking unit 26. Fire is determined by detecting changes in

In particular, in the case of a difference image in which classes are not arranged, the class tracking unit 26 randomly scatters a plurality of particles in the image to arrange a new class as initialization processing, and obtains a difference image in which classes are arranged. In the case of images, the classes were rearranged by scattering multiple particles according to a predetermined probability density distribution centered on a predetermined position above the position of each particle when the classes were previously arranged, and all classes were rearranged. Later, a new class is placed by randomly scattering multiple particles in the area excluding the class.

In this way, when the class tracking unit 26 rearranges the classes, it scatters particles around the position above the position of each particle when the classes were previously arranged, and rearranges the classes. After the placement is completed, by scattering particles in the area other than the class placement area and placing a new class, the mass of smoke (smoke area) with a certain density rising as a smoke vortex without interfering with each other. , the classes can be arranged and rearranged continuously and tracked, and the specific characteristics of the smoke can be captured from the movement trajectories of multiple classes to reliably determine that it is a fire.

Also, since it is generally difficult for smoke to descend, when rearranging the class, by scattering particles in an area slightly above the position of each particle when the class was placed last time and rearranging the class , it is possible to reliably prevent overlapping of particles between classes, and to detect the moving trajectory of the class that accurately follows the rise of the mass of smoke. A detailed description will be given below.

[Outline of smoke detector]
As shown in FIG. 1, a monitoring camera 10 is installed in a monitoring area 14, and the monitoring camera 10 sequentially captures images of the monitoring area illuminated by the lighting equipment at a frame rate of, for example, 30 frames per second or 60 frames per second. output a moving image consisting of a series of color frame images. The surveillance camera 10 is installed, for example, in the center of the upper corner of a surveillance area (monitoring space) 14 partitioned vertically, horizontally, and forward and backward, and is arranged obliquely downward so that its imaging optical axis faces the opposite wall surface. The entire area 14 can be photographed.

The monitoring camera 10 captures a moving image of the visible image of the monitoring area 14, and the moving image captured by the monitoring camera 10 is transmitted through a transmission line to a smoke detection device 12 installed in a janitor's room or the like. Smoke 16 rising from a fire source 15 such as a canister is detected by image processing using a particle filter to judge a fire, and a fire detection signal is transmitted to a fire alarm equipment 18 to output a fire alarm. The size of the moving image captured by the monitoring camera 10 is, for example, 1280×720 pixels.

[Functional Configuration of Smoke Detection Device]
As shown in FIG. 2, the smoke detection device 12 is composed of a computer circuit having a CPU, a memory, various input/output ports, etc. as hardware. A unit 20, an image input unit 22, a difference image generation unit 24, a class tracking unit 26, and a fire judgment unit 28 are provided. Also, an alarm display unit 30 and an acoustic alarm unit 32 are provided to issue an alarm when a fire is detected.

The control unit 20 performs overall control of each function provided in the monitoring camera 10 and the smoke detection device 12 . The monitoring camera 10 operates in response to an instruction from the control unit 20, and transmits image data of the monitoring area at, for example, 30 frames per second as moving image data by transmission control, and the image input unit 22 of the smoke detection device 12 , and stored in a memory (not shown). The image input unit 22 reads the stored moving image frame by frame according to an instruction from the control unit 20 , and outputs it to the differential image generation unit 24 .

[Difference image generator]
FIGS. 3A and 3B are explanatory diagrams showing frame images captured by a monitoring camera and their differential images, FIG. 3A showing an image of smoke, and FIG. 3B showing a differential image of smoke.

The difference image generator 24 generates a difference image according to the following procedure.
(Procedure 1) A frame image is converted to grayscale and smoothed by a Gaussian filter for noise reduction.
(Procedure 2) Monitor the maximum value and minimum value of each pixel in the latest, for example, 30 frames including the current frame image, and create a maximum value image and a minimum value image between the 30 frames.
(Procedure 3) A difference image is created from the maximum value image and the minimum value image generated in the procedure 2.

Note that the generation of the difference image is not limited to the procedures 1 to 3, and an appropriate method of generating the difference image can be applied.

Here, the length of 30 frames for obtaining the difference is the length of processing for an image of smoke with a slow flow velocity. , parallel processing that creates a difference image from a plurality of frames corresponding to the flow velocity.

As shown in FIG. 3A, in a smoke image 16a of a frame 34 photographing the smoke 16 rising from the fire source 15 shown in FIG. Focusing on one point in the image, the presence or absence of smoke changes on the line connecting the viewpoint and the background. Using the features of , smoke is visualized by differential processing.

A smoke differential image 16b shown in the differential frame 36 of FIG. 3B is generated by the differential processing of steps 1 to 3, and the smoke differential image 16b is a mass of smoke having a certain density. It is visualized that it rises while appearing multiple times. Note that the fire source image 15a disappears in the differential frame 36 as indicated by the dotted line.

[Class tracking part]
FIG. 4 is an explanatory diagram showing the procedure up to placement of a new class by scattering particles in the first appearing smoke area, and FIG. 5 is an explanation showing the particle scattering area during class rearrangement. FIGS. 6A and 6B are explanatory diagrams showing placement of new classes following the rearrangement of classes in FIG .

(Overview of class tracking)
The class tracking unit 26 uses a particle filter to arrange smoke regions, which are clusters of smoke, as classes in the differential image, and tracks the classes in time series to detect movement trajectories of the classes.

The class tracking unit 26 repeats the processes of sampling, prediction, and observation, which are the processing procedures of the particle filter. That is, the class tracking unit 26 scatters a plurality of particles in the image according to a predetermined rule for each difference image generated by the difference image generation unit 24, and determines the degree of smoke-likeness for each particle. The process of weighting, finding the center of gravity of the smoke area, and arranging the classes is repeated, and the classes are tracked to detect the movement trajectory. Here, one particle corresponds to, for example, one pixel, the density value of the particle is used as the likelihood, and weighting is performed according to the pixel value.

Note that the particle size may be a predetermined number of pixels, and in this case, the sum of pixel values of the pixels included in the particle is weighted as likelihood.

(initial placement of classes)
If a class has not been arranged by the processing up to that point, the class tracking unit 26 randomly scatters a predetermined number of particles, for example, 40 particles in the read difference image to create a new class. is placed.

For example, as shown in FIG. 4, for a processing frame 38-1 in which a smoke area 41 encircled by a dotted line, which is a clump of smoke, appears, 40 particles 40 are randomly scattered over the entire image area.

Subsequently, as shown in a processing frame 38-2, the pixel value of each particle is weighted as a likelihood, and particles with a low likelihood are represented by white circles and particles with a high likelihood are represented by a black circle. 34 low-likelihood particles are extinguished to leave the high-likelihood estimated smoke particles 42, and the gravity center calculation is performed on the weight values of the remaining estimated smoke particles 42, as shown in the processing frame 38-4. , a class 44-1 having a center of gravity 46-1 is arranged.

Here, the class 44-1 indicates a smoke area that is a mass of smoke, but is actually a set of estimated smoke particles 42 and is shown as a circular area for the sake of simplicity of explanation.

(class rearrangement)
As shown in FIG. 4, the class tracking unit 26, after arranging the classes in the first appearing smoke area, when reading the next differential image, the remaining 6 particles from the previous class arrangement are , centering on a predetermined position above the position (center of gravity) 46-1 of each particle when it was placed last time, determined by the likelihood of the particles that are the reference of the position, and the total number of rearranged particles is set to 40. are scattered at coordinate positions generated according to random numbers of a predetermined probability density distribution, for example, a normal distribution, and 40 particles are scattered to rearrange the classes. At this time, the number of particles to be rearranged based on that position is increased as the likelihood of the particles at the time of previous placement is higher.

Here, the distance L from the position (center of gravity) 46-1 of each particle arranged last time to the scattering center 48 is changed according to the moving speed corresponding to the color of the smoke or the moving speed of the past class. This is because the movement of smoke between images changes depending on the speed at which the smoke rises . By increasing the distance L to the smoke moving speed and shortening the distance L if the moving speed of the smoke is low, it is possible to scatter the particles in the optimal area for class rearrangement corresponding to the moving speed of the smoke.

For example, since the black smoke caused by combustion fire rises at a high speed, the distance L to the center of particle scattering is increased. Shorter distances allow for more accurate class repositioning as smoke moves.

Furthermore, the class tracking unit 26 selects particles below a horizontal line 52 passing through the position (center of gravity) 46 -1 of each particle previously placed among the particles scattered for class rearrangement in the scattering area 50 shown in FIG. is defined as a non-scattering area 54 as indicated by a dotted line, particles existing in the non-scattering area 54 are excluded from the calculation of the center of gravity position of the class to be rearranged, and the scattering area 50 existing above the horizontal line 52 is The center of gravity is obtained from the likelihood of existing particles and the process of rearranging them is performed.

Particles existing in the non-scattering area 54 can be excluded by setting the weight of the particles to zero. Further, by moving all the particles existing in the non-scattering area 54 onto the horizontal line 52 , the center of gravity of the class may be calculated as the particles in the scattering area 50 .

When repositioning the classes in this way, the smoke is usually does not move downward, it is possible to detect the moving trajectory of the smoke-specific class by rearranging the class so as not to move it downward.

Processed frames 38-5 and 38-6 in FIG. 6 show class rearrangement for the next difference image following processed frame 38-4 in FIG. In the processing frame 38-5, the particles are scattered so that the total number is 40, centering on a predetermined position above the position (center of gravity) 46-1 of each particle when it was placed last time. As shown in a processing frame 38-6, particles with low likelihood are extinguished, particles with high likelihood indicated by black circles are left, and the center of gravity 46-2 of the class of particles with high likelihood is obtained, thereby forming a class 44-2. are rearranged.

In rearranging the classes, the distance L from the position (center of gravity) 46-1 of each particle arranged last time shown in FIG. The distance L can be made substantially constant by changing the number of frames from which the difference is taken in the difference image generation unit 24 shown in FIG. 2 according to the moving speed of the smoke.

For example, if the difference image generation unit 24 generates a difference image in units of 30 frames for slow-flowing smoke, and generates a difference image in units of 6 frames for fast-flowing smoke, the difference image for slow-flowing smoke in units of 30 frames is generated. , the amount of movement of the smoke area in the difference image in units of, for example, 6 frames with a high flow velocity can be made substantially the same, and the processing of the class tracking unit 26 can be simplified.

(Class initial placement after class relocation)
The class tracking unit 26 scatters 40 particles at random in an area excluding the rearranged class area when all the classes have been rearranged for the image in which the classes have been arranged in the previous image, and a new Perform class initialization placement that places the class.

In the rearrangement of classes shown in processing frames 38-5 and 38-6 in FIG. 6, the number of classes arranged in the previous image is one, and the class rearrangement is completed in one process. there is When the class rearrangement is completed in this way, as shown in a processing frame 38-7, 40 particles 40 are randomly placed in the area excluding the class 4 4-2 having the rearranged class center of gravity 46-2. In the same way as in the class initialization arrangement shown in FIG. classes can be placed.

(Detection of class movement trajectory)
The class tracking unit 26 repeats class placement and class rearrangement on the difference image read in time series, thereby performing processing for detecting the movement trajectory of the class moving in time series.

FIG. 7(A) shows processing frames 60-1 to 60-5 processed in time series from time t1 to t5, in which classes C1 to C5 are arranged in order and rise due to rearrangement.

As a result, for class C1, for example, the class tracking unit 26 can detect the movement trajectory of the center of gravity Gt1 to Gt5 of class C1 shown in FIG. 7B. For the remaining classes C2 to C5, the movement trajectories of the centers of gravity of the classes are similarly detected.

(Elimination of retained classes)
As shown in processing frame 60-5 in FIG. 7A, class C1, which has moved due to rising smoke, does not rearrange when it reaches the top of the image and stays there.

If the class tracking unit 26 detects the stagnated class C1 as shown in the processing frame 60-5 of FIG. 8 when the class rearrangement processing ends, the class tracking unit 26 deletes the class C1 as indicated by the dotted line ( reset) and made available for placement to the new class.

For example, if the maximum number of classes is 5, as shown in FIG. 8, when class C1 is detected to be retained and deleted in processing frame 60-5, the next sub-class can be reused after deleting the retained class. As shown in the processing frame 60-6 for the image, the area below the currently existing smoke areas of classes C2 to C5, for example, below a horizontal line 62 passing under class C5, is defined as a scattering area 64. , 40 particles 40 are randomly scattered to generate a new class, and the newly generated class is arranged as the class C1 erased by retention.

Further, when the class C1 is retained and deleted, and no other classes exist, the area below the horizontal line passing under the deleted class C1 is used as a scattering area 64 for arranging a new class. Perform initialization processing.

In this embodiment, the maximum number of classes arranged by the class tracking unit 26 is, for example, 16 classes. Classes can be placed without being restricted by the maximum number of classes.

(Effect of rearrangement based on class priority)
As shown in processing frames 60-1 to 60-5 in FIG. 7, the class tracking unit 26 sets a different priority to each class when repeatedly arranging and rearranging classes for smoke areas that appear one after another. From the likelihood of the remaining particles excluding the particles that overlap the smoke area of the already rearranged upper class among the multiple particles scattered for class rearrangement in order from the upper class by A process of obtaining the center of gravity and relocating is performed.

For example, assuming that a higher priority is set for an older generation of a class, taking the rearrangement from the processing frame 60-4 to the processing frame 60-5 in FIG. degree is set.

At this time, the class tracking unit 26 first rearranges the class C1, which has the highest priority, but at this time, the particle scattering range is not restricted by the other classes C2 to C4, which have lower priorities. Next, the class tracking unit 26 rearranges the class C2 with the next highest priority. At this time, the rearranged high-priority class C1 smoke area is masked, the particles scattered for class C2 rearrangement are removed from the class C1 smoke area, and the class C1 and C2 smoke areas are removed. avoid duplication.

As a result, classes can be rearranged without overlapping in correspondence with continuously rising smoke regions, and the moving trajectory of the classes can be accurately detected.

[Fire Judgment Department]
The fire determination unit 28 detects a characteristic predetermined change due to smoke based on the movement trajectory of the position of the center of gravity of each class detected by the class tracking unit 26, and determines the generation of smoke due to fire.

In addition, the fire determination unit 28 sets an accumulation condition for fire determination based on the movement trajectory of the class. For example, when the movement trajectory of the class continues for a predetermined time or longer, it is determined that smoke is generated due to a fire. to output a fire alarm and transmit a fire detection signal to the outside.

[Processing operation of the smoke detection device]
9 is a flow chart showing an outline of smoke detection processing by the smoke detection apparatus of FIG. 2, and FIG. 10 is a flow chart showing details of class tracking by the particle filter in step S2 of FIG.

(Outline of smoke detection processing operation)
As shown in FIG. 9 , the control unit 20 of the smoke detection device 12 shown in FIG. , the maximum value and minimum value of each pixel are detected to create a maximum value image and a minimum value image between a plurality of frames, and a difference image is generated from the maximum value image and the minimum value image.

Subsequently, the control unit 20 advances to step S2, operates the class tracking unit 26, scatters a plurality of particles in the image according to a predetermined rule for each difference image generated by the difference image generation unit 24, and is weighted by the likelihood of indicating smoke-likeness, the center of gravity of the smoke area is obtained, and the process of arranging classes is repeated, and the class is traced to detect the movement trajectory indicating the movement of the position of the center of gravity of the class.

Subsequently, the control unit 20 advances to step S3, operates the fire determination unit 28, and if the movement trajectory of the center of gravity of the class detected by the class tracking unit 26 continues for a predetermined time or longer, the fire is detected as smoke caused by the fire. to judge.

Subsequently, the control unit 20 proceeds to step S4 to determine whether or not the fire determination in step S3 satisfies a predetermined accumulation condition. Next, the alarm display unit 30 and the sound alarm unit 32 are operated to output a fire alarm, and a fire detection signal is transmitted to the fire alarm equipment 18 shown in FIG. 1 to output a fire alarm.

(class tracking process)
The class tracking unit 26 operated by the instruction of the control unit 20 in step S2 of FIG. 9 performs the class tracking process shown in FIG . The class tracking unit 26 reads the difference image generated by the difference image generation unit 24 in step S11. At step S12, it is determined that there is no existing class, and the process advances to step S13 to perform class placement as initialization processing.

Class placement as initialization processing by the class tracking unit 26 involves randomly scattering, for example, 40 particles in the differential image in step S13 as shown in FIG. A weight is set as a likelihood, particles with a low likelihood are deleted, particles with a high likelihood are left, the center of gravity of the weight of the remaining particles is obtained by calculation in step S15, and a new class having the obtained center of gravity is arranged. Then, the center of gravity (coordinate position) of the class arranged in step S16 is stored, and the process returns to the main routine of FIG.

On the other hand, if the class tracking unit 26 determines the existence of an existing class in step S12, the process proceeds to step S17 to perform class rearrangement processing. In the class rearrangement processing by the class tracking unit 26, the center of gravity of the existing class is extracted in step S17. Here, when a plurality of classes have already been arranged, since the priority is set to the classes in order of oldest, the class tracking unit 26 calculates the center of gravity of the class with the highest priority (the class with the oldest appearance). take out.

Subsequently, the class tracking unit 26 advances to step S18 to determine whether or not the class from which the center of gravity has been extracted is a class that does not resemble smoke. available for placement. For example, as shown in the processing frame 60-5 in FIG. 7, the class that reaches the upper end of the screen and stays is erased as a non-smoky class. Also, if the width of the distribution of particles used to calculate the center of gravity of the class exceeds a predetermined range (if the standard deviation exceeds a predetermined value), the class is deleted as a non-smoke-like class.

When the class tracking unit 26 determines in step S18 that the class is likely to be smoke, the process proceeds to step S20, and as shown in FIG . A scattering center 48 is set, and the number of particles erased in the previous process is scattered by generating coordinates based on random numbers of a normal distribution. In step S22, particles with low likelihood are deleted, particles with high likelihood are left, and the weight centers of the remaining particles are calculated to rearrange the classes. coordinate position).

Subsequently, the class tracking unit 26 advances to step S24 to determine whether or not the rearrangement of all classes has been completed. , the classes are rearranged in steps S20 to S23.

If the class tracking unit 26 determines in step S24 that the rearrangement of all classes has ended, it proceeds to step S26. As shown in the processing frame 60-6 of FIG. 8, 40 particles are randomly scattered under the existing class, and the density values of the particles are weighted as likelihoods in steps S14 to S16. Initialization processing for arranging a new class is performed by erasing low-level particles and leaving particles with high-likelihood to obtain the center-of-gravity position.

In the scattering of particles in step S27, if there is no class other than the erased class, 40 particles are randomly scattered under the erased class, and the density of the particles is reduced in steps S14 to S16. Values are weighted as likelihoods, particles with low likelihoods are deleted, particles with high likelihoods are left, and the center of gravity is obtained, thereby arranging new classes.

If the class tracking unit 26 determines in step S26 that there is no deleted class, the process proceeds to step S28, and as shown in the processing frame 38-7 of FIG. 40 particles are scattered in the area, and an initialization process for arranging a new class is performed in steps S14 to S16. Note that the scattering of particles in step S28 can be performed by randomly scattering 40 particles while masking the smoke area of the existing class.

[Modification of the present invention]
(class placement)
The above embodiment also eliminates the dormant class and places the new class below the area where the other class resides, or below the eliminated class if the other class does not exist. Although the classes are arranged by randomly scattering particles on the side, the present invention is not limited to this, and for example, a predetermined number of particles may be uniformly and randomly scattered over the entire screen.

(fire detection)
Further, in the above-described embodiment, classes are arranged and tracked in characteristic regions by a particle filter as characteristic time-series changes due to smoke. A fire may be determined by tracking time-series changes that indicate characteristic movement due to smoke.

(Image processing during normal operation and when smoke is generated)
Also, during normal times, the number of images to be processed is reduced (for example, one image per five shots), and when it is determined that there is a high possibility of smoke, the number of images to be processed is increased (all shot images). You may make it process. Here, the number of images specifically means the number of frames.

In addition, when it is determined that there is a high possibility of smoke, images that were not processed (for example, the remaining 4 images from the 5 times of imaging) are included. All the image processing may be performed again from a time slightly earlier.

Furthermore, as another image processing, a process of simply changing the imaging interval, for example, the imaging interval may be lengthened in normal times and shortened when it is determined that there is a high possibility of smoke.

(others)
Moreover, the present invention is not limited to the above-described embodiments, includes appropriate modifications that do not impair the objects and advantages thereof, and is not limited by the numerical values shown in the above-described embodiments.

10: Surveillance camera 12: Smoke detection device 14: Monitoring area 15: Fire source 16: Smoke 16a: Smoke image 16b: Smoke difference image 18: Fire alarm equipment 20: Control unit 22: Image input unit 24: Difference image generation unit 26: Class tracking unit 28: Fire judgment unit 30: Alarm display unit 32: Acoustic alarm unit 34: Frame 36: Differential frame 40: Particle 42: Smoke estimation particles 44-1, 44-2: Class 48: Scattering center 50 , 58 , 64 : scattered area 54 : non-scattered area

Claims (2)

imaging means for sequentially capturing images of the monitored area;
class tracking means for repeating a process of arranging a class by obtaining the center of gravity of a smoke area in the image for each image captured by the imaging means, and tracking the class to detect a movement trajectory;
fire determination means for determining a fire by detecting a characteristic predetermined change due to smoke from the moving trajectory of the class detected by the class tracking means;
is provided,
The class tracking means comprises:
For each image captured by the imaging means, a predetermined number of particles are scattered in the image according to a predetermined rule, and the center of gravity of the scattered predetermined particles is obtained as the center of gravity of the smoke region, thereby obtaining the predetermined class. is placed,
In the case of an image in which the class is not arranged, a new class is arranged by scattering a predetermined number of particles over the entire image area,
In the case of an image in which the classes are arranged, for each class that has been arranged, a predetermined number of particles are scattered around a predetermined position in the region above the center of gravity of the class to be processed. By scattering, rearrange the particles of each class and rearrange the classes, after rearranging all the classes, place a new class targeting the area excluding the class ,
When relocating the class, the distance from the center of gravity of the particles when the class was placed last time to the scattering center corresponds to the moving speed corresponding to the color of the smoke, or the moving speed of the class in the past. A smoke detection device characterized by changing by
The imaging means sequentially captures images of the monitored area,
class tracking means for each image generated by said imaging means, repeating the process of obtaining the center of gravity of the smoke area in the image and arranging classes, tracking said classes and detecting movement trajectories;
the fire determination means detects a characteristic predetermined change due to smoke from the movement trajectory of the class detected by the class tracking means to determine a fire;
The class tracking means comprises:
For each image captured by the imaging means, a predetermined number of particles are scattered in the image according to a predetermined rule, and the center of gravity of the scattered predetermined particles is obtained as the center of gravity of the smoke area, thereby determining the predetermined class. Do the processing to place,
In the case of an image in which the class is not arranged, a new class is arranged by scattering a predetermined number of particles over the entire image area,
In the case of an image in which the classes are arranged, for each class that has been arranged, a predetermined number of particles are scattered around a predetermined position in the region above the center of gravity of the class to be processed. By scattering, rearrange the particles of each class and rearrange the classes, after rearranging all the classes, place a new class targeting the area excluding the class ,
When relocating the class, the distance from the center of gravity of the particles when the class was placed last time to the scattering center corresponds to the moving speed corresponding to the color of the smoke, or the moving speed of the class in the past. A method of detecting smoke, characterized in that the

Images