JP2020177585A - Fire detection system and fire detection program - Google Patents

Abstract

To provide a fire detection system and a fire detection program capable of quickly detecting a fire.SOLUTION: A sensor 100 that detects a fire in a monitoring region based on a monitoring region image, which is an image of the monitoring region, is comprised of: a photographing unit 11 that captures the monitoring region image; a fire detection unit 142 that detects a fire in the monitoring region based on a detection reference template for detecting the fire and the monitoring region image taken by the photographing unit 11; and a production unit 141 that identifies a scale and a direction of movement of a product associated with an outbreak of a fire, determines an outline template which is a region containing the product based on the scale of the product, determines a product-corresponding region, which is a region corresponding to the product, based on the moving direction of the product, and produces an overlapping region of the outline template and the product-corresponding region as the detection reference template.SELECTED DRAWING: Figure 2

Description

The present invention relates to a fire detection system and a fire detection program.

Conventionally, a fire detector has been known which includes a chamber into which smoke flows and detects a fire by detecting the concentration of smoke flowing into the chamber (see, for example, Patent Document 1).

Japanese Unexamined Patent Publication No. 2011-248547

However, in the fire detector of Patent Document 1, since it is necessary to let the smoke that has reached the fire detector flow into the chamber in order to detect the fire, the smoke fires when the fire source is far away or due to the influence of the wind. If it does not move to the sensor side, the detection of the fire may be delayed, and there is room for improvement from the viewpoint of quickly detecting the fire.

The present invention has been made in view of the above problems, and an object of the present invention is to provide a fire detection system capable of quickly detecting a fire and a fire detection program.

In order to solve the above-mentioned problems and achieve the object, the fire detection system according to claim 1 is a fire detection system that detects a fire in the monitoring area based on a monitoring area image which is an image obtained by photographing the monitoring area. The fire in the monitoring area is detected based on the photographing means for photographing the monitoring area image, the detection reference template for detecting the fire, and the monitoring area image photographed by the photographing means. It is provided with a fire detection means.

Further, the fire detection system according to claim 2 is a region in which a product associated with the occurrence of the fire exists based on the monitoring area image taken by the photographing means in the fire detection system according to claim 1. The scale of the product existence region and the moving direction of the product are specified, and the outer shape template showing the region including the product existence region is determined based on the scale of the product existence region, and the product is determined. A product-corresponding region, which is a region corresponding to the product-existing region, is determined based on the moving direction of the product, and a generation means for generating an overlapping region of the outer shape template and the product-corresponding region as the detection reference template. The fire detecting means detects a fire in the monitoring area based on the detection reference template generated by the generating means and the monitoring area image taken by the photographing means.

Further, the fire detection system according to claim 3 is the fire detection system according to claim 2, wherein the generation means has a predetermined point in the outer shape template as an end point and a predetermined angle with respect to the moving direction. The region determined by the two half lines opened by each is determined as the product corresponding region.

The fire detection program according to claim 4 is a fire detection program that detects a fire in the monitoring area based on a monitoring area image that is an image of the monitoring area, and detects the fire by a computer. It functions as a fire detecting means for detecting a fire in the monitoring area based on the detection reference template for the purpose and the monitoring area image taken by the photographing means for capturing the monitoring area image.

According to the fire detection system according to claim 1, for example, a fire has occurred by detecting a fire in a monitoring area based on a detection reference template for detecting a fire and a captured monitoring area image. Since the fire can be detected without waiting for the smoke to arrive, the fire can be detected quickly. In particular, since the fire can be detected using the detection reference template, the fire can be detected by a relatively simple process such as comparison between the detection reference template and the monitoring area image, and the fire can be detected more quickly. Can be detected.

According to the fire detection system according to claim 2, a detection reference template is generated based on the scale of the product existence area, which is the area where the product exists due to the occurrence of a fire, and the movement direction of the product. For example, when the product of a fire is smoke, it is assumed that the smoke is present in a predetermined area and that the tilt is affected by the wind that appears as the direction of movement. In such a case, it is possible to generate a region where smoke is predicted to exist based on the characteristics of the smoke, the detected scale and the moving direction as a detection reference template, and to compare this detection reference template with the monitoring region image. Therefore, it depends on the product that is truly associated with the outbreak of fire, including whether or not the area detected as the product associated with the outbreak of fire has the same characteristics as the area that is caused by the product that is associated with the outbreak of fire such as smoke. Since the comparison is made, the fire can be detected quickly and reliably.

According to the fire detection system according to claim 3, a region determined by two half lines opened by a predetermined angle with respect to a moving direction with a predetermined point in the outer shape template as an end point is a product corresponding region. By determining, for example, the characteristics of the smoke that is the product (for example, the characteristics that the smoke spreads horizontally as it rises and the way it spreads in the flow direction of the entire smoke is affected by the influence of wind, etc.) are detected as detection criteria. Since it can be reflected in the template, the fire can be detected quickly and reliably.

According to the fire detection program according to claim 4, the fire in the monitoring area is detected based on the detection reference template for detecting the fire and the captured monitoring area image, so that, for example, a fire occurs. Since the fire can be detected without waiting for the smoke to arrive, the fire can be detected quickly. In particular, since the fire can be detected using the detection reference template, the fire can be detected by a relatively simple process such as comparison between the detection reference template and the monitoring area image, and the fire can be detected more quickly. Can be detected.

It is a side view of the sensor installed in the monitoring area which concerns on this embodiment. It is a block diagram of a sensor. It is a flowchart of a fire detection process. It is a figure which illustrated the superimposition binarization candidate area image. It is a figure exemplifying the state in which the circumscribing rectangle is illustrated with respect to the image of FIG. It is the figure which illustrated the outline template. It is a figure which illustrated the outline template and the product correspondence area. It is a figure which illustrated the detection standard template.

Hereinafter, embodiments of the fire detection system according to the present invention will be described in detail with reference to the drawings. The present invention is not limited to this embodiment.

[Basic concept of the embodiment]
First, the basic concept of the embodiment will be described. Embodiments generally relate to fire detection systems.

Here, the "fire detection system" is a system for detecting a fire in a monitoring area, specifically, a system for detecting a fire in a monitoring area based on a detection reference template and a monitoring area image. For example, it is a concept including a dedicated system for fire detection, a system in which a fire detection function is implemented for a general-purpose system, and the like, and one example is a concept including a detector, which is a device for detecting a fire in a monitoring area. Is. This "fire detection system" is a concept including, for example, at least a photographing means and a fire detecting means.

Further, the "monitoring area" is an area to be monitored by the fire detection system, specifically, a space having a certain extent, an indoor space or an outdoor space, for example. , A concept that includes spaces such as corridors, stairs, or rooms in buildings. The "monitoring area image" is an image obtained by capturing the monitoring area. Further, the "detection reference template" is used for detecting a fire, and is generated by, for example, a fire detection system, or is recorded in advance in the fire detection system without being generated by the fire detection system. As an example, it is a concept including an area corresponding to an overlapping portion of an outline template and a product corresponding area.

Further, the "shooting means" is for shooting a surveillance area image, for example, a single still image is shot, a plurality of still images are shot, or a still image is shot continuously. It is a concept that includes things such as shooting moving images. Further, the "fire detecting means" detects a fire in the monitoring area based on the detection reference template and the monitoring area image taken by the photographing means.

Then, in the embodiment shown below, the "fire detection system" is the "sensor", the "monitoring area" is the "room of the building", and the "detection reference template" is generated by the "fire detection system". A case where the "thing" and the "shooting means" are "things for shooting a moving image" will be described.

(Constitution)
First, the configuration of the sensor according to the present embodiment will be described. FIG. 1 is a side view of a sensor installed in the monitoring area, and FIG. 2 is a block diagram of the sensor. Note that FIG. 1 shows a state in which a fire has occurred in a monitoring area, which is a room of a building, and smoke 91 is generated from the fire source for convenience of explanation. Further, in the following description, the XYZ directions shown in the respective figures are orthogonal to each other. Specifically, the Z direction is the vertical direction, and the X and Y directions are the vertical directions (horizontal directions). ) Is orthogonal to, for example, the Z direction is referred to as the height direction (vertical direction), the + Z direction is referred to as the upper side (plane), and the −Z direction is referred to as the lower side (bottom surface). explain.

The detector 100 shown in each of these figures is a fire detection system, which is a device for detecting a fire in a monitoring area. For example, as shown in FIG. 1, it is attached to an installation surface 900 which is a ceiling surface of the monitoring area. It is used to detect a fire by photographing a monitoring area including smoke 91, which is a product of a fire. The sensor 100 includes, for example, a photographing unit 11, an alarm unit 12, a recording unit 13, and a control unit 14 in FIG.

(Composition-shooting section)
The photographing unit 11 of FIG. 2 is a photographing means for photographing a monitoring area image which is an image of a monitoring area, and specifically, shoots a moving image by continuously photographing a still image. The specific type and configuration of the photographing unit 11 are arbitrary, but for example, it is assumed that the photographing region A11 surrounded by the alternate long and short dash line shown in FIG. 1 is photographed, and a known image pickup device (for example, , CMOS image sensor, CCD sensor, etc.) and the like. Actually, in the monitoring area of FIG. 1, a plurality of detectors having the same configuration as the sensor 100 are provided so that the entire monitoring area can be monitored, but they are shown here. The sensor 100 will be described with attention.

(Configuration-alarm unit)
The alarm unit 12 in FIG. 2 is an alarm means that outputs an alarm when the detector 100 detects a fire. The specific type and configuration of the alarm unit 12 is arbitrary, but includes, for example, a speaker that outputs an alarm by voice, an indicator lamp that outputs an alarm by light emission display, and the like.

(Configuration-Recording unit)
The recording unit 13 of FIG. 2 is a recording means for recording a program and various data necessary for the operation of the sensor 100, and is configured by using, for example, a hard disk (not shown) as an external recording device. However, in place of or together with the hard disk, a magnetic recording medium such as a magnetic disk, an optical recording medium such as a DVD or a Blu-ray disk, or an electrical recording medium such as a Flash, ROM, USB memory, or SD card is included. Any recording medium can be used.

(Configuration-Control unit)
The control unit 14 of FIG. 2 is a control means for controlling the sensor 100, and specifically, a CPU, various programs interpreted and executed on the CPU (basic control programs such as an OS, and startup on the OS). It is a computer configured to include an application program that realizes a specific function) and an internal memory such as a RAM for storing the program and various data. In particular, the control program according to the embodiment is installed in the sensor 100 via an arbitrary recording medium or network to substantially configure each part of the control unit 14.

The control unit 14 conceptually includes a generation unit 141 and a fire detection unit 142. Based on the monitoring area image taken by the photographing unit 11, the generation unit 141 specifies the scale of the product existence area and the movement direction of the product, which is the area where the product exists due to the occurrence of the fire, and the product exists. Based on the size of the region, an outline template indicating the region including the product presence region is determined, and based on the moving direction of the product, the product correspondence region, which is the region corresponding to the product presence region, is determined. It is a generation means for generating an overlapping area of the outline template and the product corresponding area as a detection reference template. Further, the fire detection unit 142 is a fire detection means for detecting a fire in the monitoring area based on the detection reference template for detecting the fire and the monitoring area image captured by the photographing unit 11. , The fire in the monitoring area is detected based on the detection reference template generated by the generation unit 141 and the monitoring area image captured by the photographing unit 11. The processing performed by each unit of the control unit 14 will be described later.

The “product” is an arbitrary object that accompanies the occurrence of a fire, but here, for example, the case of smoke generated by the occurrence of a fire will be described as an example. The "outer shape template" is a region including a product existence region and has an arbitrary shape, but here, for example, a case of a rectangular shape will be described as an example. The "product existence region" is an region in which a product exists due to the occurrence of a fire, and specifically, is an region in which it is determined on the image that a product exists. For example, the photographing unit 11 photographs the image. It is a concept including an area determined to actually exist as a product in the superimposed binarization candidate area image processed based on the monitored area image. The “product-corresponding region” is an region corresponding to a product, which is an region where a product is expected to exist, and here, for example, a case where smoke is expected to be present. It will be described by way of example.

(processing)
Next, the fire detection process executed by the sensor 100 configured in this way will be described. FIG. 3 is a flowchart of the fire detection process. The "fire detection process" is a process for detecting a fire in a monitoring area. For example, a monitoring area image is taken, a detection reference template is generated based on the captured monitoring area image, and the generated detection reference template is used. This is a process for detecting a fire based on the above-mentioned captured monitoring area image. The timing of executing this fire detection process is arbitrary, but for example, the photographing unit 11 determines to acquire a monitoring area image at 30 frames per second, and repeatedly starts every few seconds, which is an arbitrary interval longer than this photographing interval. It will be described from the place where the fire detection process is started as the one to be executed. Here, for example, as shown in FIG. 1, a case where the sensor 100 photographs a monitoring area in a state where a fire has occurred and smoke 91 is generated will be described as an example.

First, as shown in FIG. 3, in SA1, the photographing unit 11 photographs the monitoring area image. The processing here is arbitrary, but specifically, the monitoring area image which is the image of the photographing area A11 in the monitoring area of FIG. 1 is continuously photographed. Here, for example, the monitoring area image, which is an image of the photographing area A11 including the smoke 91 of FIG. 1, is continuously photographed at 30 frames per second.

Returning to FIG. 3, in SA2, the generation unit 141 preprocesses the monitoring area image. The processing here is optional, but at least it is a processing for making it possible to distinguish the smoke reflected in the surveillance area image from the background of the smoke. Specifically, the images were continuously photographed by SA1. Preprocessing is performed by acquiring monitoring area images for a plurality of consecutive frames and performing noise removal and color correction on the acquired monitoring area images. As for the noise removal here, any method including a known method can be used (the same applies to color correction, binarization described later, superimposition, expansion / contraction processing). Here, for example, the monitoring area image, which is the image of the photographing area A11 including the smoke 91 of FIG. 1, is preprocessed by performing noise removal and color correction.

Returning to FIG. 3, in SA3, the generation unit 141 extracts a candidate region and the like. Here, the "candidate area" is an area in the monitoring area image, specifically, an area used for detecting a fire, for example, a part of the area in the monitoring area image. , It is an area where possible smoke due to the occurrence of a fire is reflected, and as an example, it is an area where smoke 91 of FIG. 1 is reflected.

The processing here is arbitrary, but at least it is a processing for making it possible to distinguish the smoke reflected in the surveillance area image from the background of the smoke. Specifically, first, the preprocessing is performed by SA2. With reference to each of the plurality of monitored area images performed, a predetermined area including an area that changes rapidly with the passage of time in consideration of moving average and the like (for example, a region having a size smaller than the size of the entire monitored area image). ) As a candidate area, and then binarizing the area corresponding to the extracted candidate area in each of the plurality of monitoring area images preprocessed by SA2 to obtain a plurality of binarization candidate area images. Generate. The "binarized candidate area image" is a binarized image, and specifically, is an image obtained by binarizing the area corresponding to the candidate area in the monitoring area image, for example. It includes a white image corresponding to the region of smoke 91 in FIG. 1 and a black image corresponding to a region other than the region of smoke 91. Next, the superposed binarized candidate region images are generated by superimposing, expanding / contracting, and the like on the generated plurality of binarized candidate region images. The "superimposed binarization candidate region image" is an image that has been superposed, and specifically, a plurality of binarization candidate region images have been superposed and expanded / contracted. It is an image, for example, one image. FIG. 4 is a diagram illustrating an superimposed binarization candidate region image. In FIG. 4, coordinate axes are shown for convenience of explanation, and the coordinate axes will be described below assuming that they correspond to the coordinate axes of FIG. 1 (the same applies to FIGS. 5 to 8 described later).

Here, for example, since the smoke 91 in FIG. 1 rises with the passage of time, the area of the smoke 91 suddenly changes in the plurality of monitoring area images preprocessed by SA2, and the change is made. After extracting a predetermined area including the area as a candidate area, a plurality of binarization candidate area images corresponding to the extracted candidate area are generated, and based on the generated plurality of binarization candidate area images, FIG. After determining that the smoke as a product exists in the region corresponding to the smoke image Im91 of FIG. 4, as shown in FIG. 4, the white smoke image Im91 corresponding to the region of the smoke 91 of FIG. 1 and the smoke 91 A superposed binarization candidate region image including the black background image Im92 corresponding to the region other than the above region is generated. In addition, for example, this smoke image Im91 corresponds to the “product existence region”.

Returning to FIG. 3, in SA4, the generation unit 141 identifies and extracts the feature amount. Here, the "feature quantity" is a characteristic physical quantity in the monitoring area image, specifically, a characteristic physical quantity related to the detection of fire, for example, the moving direction of smoke and the presence of products. Of the area. The method here is arbitrary, but specifically, the movement direction of smoke is based on a plurality of monitoring area images preprocessed by SA2 and a superposed binarization candidate area image generated by SA3. And the scale of the product existence area is specified and extracted.

First, in detail about the moving direction of smoke, a plurality of monitoring area images preprocessed by SA2 are acquired, and a plurality of randomly selected areas corresponding to candidate areas in each of the acquired plurality of monitoring area images are obtained. Identify the points, identify the color change of each of the identified points over time, calculate the optical flow based on the color change of each of the identified points, and use the calculated optical flow as the smoke. Specify as the moving direction and extract. The "optical flow" is a concept representing the movement of an object in an image, and is, for example, a concept representing a moving direction of smoke. Here, for example, a plurality of monitoring area images preprocessed by SA2 are acquired, and 20 to 30 points are randomly selected from the areas corresponding to the candidate areas in each of the acquired plurality of monitoring area images. Identify, identify the color change of each of the 20 to 30 identified points over time, and perform an operation on each change (eg, when the color changes from the first point to the second point). Calculates a vector from the first point to the second point, performs such a calculation on the specified plurality of points described above to obtain a plurality of vectors, and then synthesizes these plurality of vectors. The direction of the arrow Ar1 whose angle formed with the Y-axis in FIG. 4 is the angle An1 is calculated as an optical flow, and the direction indicated by the arrow Ar1 is used as the direction of smoke movement. Identify and extract.

Further, in detail about the scale of the product existence region, the superimposed binarization candidate region image generated by SA3 is acquired, and in the acquired superimposed binarization candidate region image, the circumscribing rectangle with respect to the smoke image is formed. The length of each side of the generated circumscribing rectangle is measured, and the measured length of the long side is specified as the scale of the product existence area and extracted. As the circumscribing rectangle here, basically, a rectangle composed of two line segments parallel to the Y axis and two line segments parallel to the Z axis of the superimposed binarization candidate region image is used. It should be noted that a rectangle may be used for the binarization candidate region image due to the inclination of the photographing unit 11 to the horizontal plane or the like. FIG. 5 is a diagram illustrating a state in which an circumscribing rectangle is illustrated with respect to the image of FIG. Here, for example, the circumscribing rectangle of FIG. 5, which is a rectangle in contact with the outer peripheral portion of the smoke image Im91 of the binarization candidate region image of FIG. 4 generated by SA3, is generated, and each of the generated circumscribing rectangles of FIG. The length of the side is measured, and the length of each measured side is specified and extracted as the scale of the product existence area. Here, when the extrinsic rectangle is a square, the length of one side may be specified and extracted, and when the extrinsic rectangle is a rectangle, either the long side or the short side or these may be specified. You may specify and extract both of them. Here, for example, the lengths of the long side and the short side of the circumscribed rectangle of FIG. 5 are extracted.

Returning to FIG. 3, in SA5, the generation unit 141 generates a detection reference template. FIG. 6 is a diagram exemplifying the outline template, FIG. 7 is a diagram exemplifying the outline template and the product corresponding region, and FIG. 8 is a diagram exemplifying the detection reference template. Although the method here is arbitrary, it generates a detection reference template corresponding to fire smoke. Specifically, based on the processing result of SA4, the outline templates Im921 and FIGS. By generating the product-corresponding region Im911 of No. 7, the hatched detection reference template Im900 of FIG. 8 is generated.

First, in detail about the outer shape template, the scale of the product existence area specified and extracted by SA4 is acquired, and the outer shape template corresponding to the acquired scale of the product existence area is generated. Here, for example, as the scale of the product existence region specified and extracted by SA4, the lengths of the long side and the short side of the circumscribing rectangle of FIG. 5 are acquired, and based on the acquired lengths, FIG. Generate the outline template Im921 of FIG. 6 corresponding to the extrinsic rectangle (ie, including the smoke image Im91). The specific size and shape of the outer shape template Im921 here are arbitrary as long as they follow the above definitions, but for example, a case where the same shape and size as the circumscribing rectangle of FIG. 5 is used will be described.

Next, in detail about the product-corresponding region, the moving direction of the smoke specified and extracted by SA4 (for example, the direction indicated by the arrow Ar1 in FIG. 4) is acquired and tilted according to the acquired moving direction. Generate a product-corresponding area. Here, for example, as shown in FIG. 7, a reference line B2 extending in the direction indicated by the arrow Ar1 in FIG. 4, which is the movement direction of the acquired smoke, is generated with reference to the reference point B1 in the outer shape template Im921. , The reference line B2 is opened at the reference point B1 by the same angle B3 (for example, a predetermined angle of about 15 to 25 degrees), and two half straight lines B4 having the reference point B1 as an end. The product-corresponding area Im911, which is an area including the reference line B2, is generated by superimposing it on the outer shape template Im921 among the areas separated by and surrounded by. The "reference point" B1 is a point in the outer shape template Im921, specifically, a predetermined point, for example, determined in consideration of the position of an object that can be a fire source in the monitoring area. As an example, the external template Im921 is located at a position closer to the lower side (-Z direction) in the vertical direction (Z-axis direction) and closer to the center in the horizontal direction (Y-axis direction). Is. The position of the reference point B1 shown in FIG. 7 is described for convenience of explanation, and is actually not limited to the position shown in the drawing, and is configured to process other positions as the reference point B1. You may.

Next, with respect to the detection reference template, in detail, the region where the above-mentioned generated outer shape template Im921 and the product corresponding region Im911 overlap each other is specified, and the identified region is generated as the detection reference template Im900 of FIG. Here, for example, in FIG. 8, a detection reference template Im900 having the same color (here, white) as the color of the smoke image Im91 of FIG. 4 is generated.

Returning to FIG. 3, the fire detection unit 142 performs correlation matching in SA6. Here, "correlation matching" is a process of calculating the degree of matching between two pieces of information. The processing here is arbitrary, but specifically, the generation of the detection reference template generated by SA5 and the plurality of binarization candidate region images generated by SA3 (that is, the generation of the detection reference template generated by SA5). The latest binarization candidate region image of the multiple underlying binarization candidate region images) is acquired, and the smoke in FIG. 1 in the acquired latest binarization candidate region image for the acquired detection reference template. The degree of matching of the white images corresponding to the 91 regions is calculated. Here, for example, the detection reference template Im900 of FIG. 8 and the white image corresponding to the region of smoke 91 of FIG. 1 in the acquired latest binarization candidate region image are substantially mutual (for example, 90 (%)). ) When they match, the case of calculating 90 (%) will be described. In the following, the unit “%” will be omitted (the same applies to the fire detection threshold value described later).

Returning to FIG. 3, the fire detection unit 142 detects a fire in SA7. The processing here is arbitrary, but specifically, it will be described as assuming that the fire detection threshold value is stored in the recording unit 13 of FIG. The "fire detection threshold" is a threshold for detecting a fire, specifically, a degree of agreement regarding correlation matching, which is determined based on a simulation or the like and input by the user. It is assumed that "85" is stored. The processing of the SA7 is detected by acquiring the degree of agreement calculated by the SA6 and the fire detection threshold value of the recording unit 13 and comparing the acquired information. More specifically, when the degree of matching calculated by SA6 is equal to or higher than the fire detection threshold value, the fire is detected, an alarm is output via the alarm unit 12 of FIG. 2, and then the process is terminated. Further, when the degree of matching calculated by SA6 is less than the fire detection threshold value, the process is terminated without detecting the fire. Here, for example, when "90" is calculated by SA6, the "90" is equal to or higher than the fire detection threshold value of "85". Therefore, a fire is detected and the alarm unit 12 does not show the speaker by voice. An alarm is output by outputting an alarm message or turning on an indicator lamp (not shown) of the alarm unit 12. This completes the fire detection process.

(About fire detection and non-detection)
As described above, when the fire smoke 91 is generated in the monitoring area of FIG. 1, the detector 100 detects the fire by performing correlation matching with an extremely high degree of matching (for example, 90). On the other hand, on the other hand, even when a person or an object moves in the monitoring area of FIG. 1, each step of the fire detection process of FIG. 3 is performed. In this case, especially in SA5. While the generated detection reference template Im900 corresponds to smoke, the monitoring area image does not reflect smoke, only moving people and objects, so correlation matching The degree of concordance will be reduced and no fire will be detected. In this way, it is possible to accurately detect a fire when a fire actually occurs.

(Effect of embodiment)
As described above, according to the present embodiment, by detecting a fire in the monitoring area based on the detection reference template Im900 for detecting the fire and the captured monitoring area image, for example, smoke generated by the fire. Since the fire can be detected without waiting for the fire to arrive, the fire can be detected quickly. In particular, since the fire can be detected by using the detection reference template Im900, the fire can be detected by a relatively simple process such as comparison between the detection reference template Im900 and the monitoring area image, and the fire can be detected. It can be detected more quickly.

Further, by generating the detection reference template Im900 based on the scale and the moving direction of the product existence area due to the occurrence of a fire, for example, when the product due to the occurrence of a fire is smoke, the smoke is predetermined. It is assumed that the smoke exists in the range of the area and that the inclination is affected by the wind that appears as the moving direction. In such a case, the smoke exists based on the characteristics of the smoke, the detected scale, and the moving direction. Then, the predicted area is generated as the detection reference template Im900, and the monitoring area image can be compared with the detection reference template Im900. Therefore, the area detected as a product due to the occurrence of a fire is a fire such as smoke. Since it is compared whether or not the fire is caused by the product caused by the occurrence of the fire, including whether or not the region has the same characteristics as the region generated by the product caused by the occurrence of the fire, the fire can be detected quickly and surely.

Further, by determining a region determined by two half-line lines opened by a predetermined angle with respect to the moving direction with a predetermined point in the outline template as an end point as a product-corresponding region, for example, in a product. A fire can be reflected in the detection reference template Im900 because certain characteristics of smoke (for example, characteristics such as spreading horizontally as it rises and being affected by the influence of wind and the like in the flow direction of the entire smoke) can be reflected in the detection reference template Im900. Can be detected quickly and reliably.

[Modified example with respect to the embodiment]
Although the embodiments according to the present invention have been described above, the specific configuration and means of the present invention may be arbitrarily modified and improved within the scope of the technical idea of each invention described in the claims. Can be done. Hereinafter, such a modification will be described.

(About the problem to be solved and the effect of the invention)
First, the problem to be solved by the invention and the effect of the invention are not limited to the above-mentioned contents, and may differ depending on the implementation environment and the details of the configuration of the invention, and only a part of the above-mentioned problems. May be resolved or only some of the above effects may be achieved.

(About distribution and integration)
Further, the above-described configuration is a functional concept, and does not necessarily have to be physically configured as shown in the figure. That is, the specific form of dispersion or integration of each part is not limited to the one shown in the figure, and all or a part thereof can be functionally or physically dispersed or integrated in any unit. The term "system" in the present application is not limited to a system composed of a plurality of devices, but includes a system composed of a single device. Further, the "device" in the present application is not limited to a device composed of a single device, but includes a device composed of a plurality of devices.

(About image color)
Further, in SA3 of FIG. 3 of the above-described embodiment, regarding the binarization candidate region image and the superimposed binarization candidate region image, a white image corresponding to the region of smoke 91 in FIG. 1 and a region of the smoke 91. Although the black image corresponds to the area other than the above, for example, these colors may be replaced. Further, for example, the colors of the background of the smoke 91 in FIG. 1 and the monitoring area are acquired, and based on these acquired colors, the color of the image corresponding to the area of smoke 91 in FIG. 1 and the area other than the area of the smoke 91 are obtained. The color of the image corresponding to the area of may be automatically determined. Further, for example, the entire process may be performed by performing multi-value increase of ternary value or more.

(About detection criteria template)
Further, in SA4 of FIG. 3 of the above embodiment, the moving direction of smoke is extracted using a plurality of monitoring region images preprocessed by SA2. For example, a plurality of binarization candidate regions generated by SA3 are extracted. The moving direction of smoke may be extracted using an image.

Further, in SA4 of FIG. 3 of the above embodiment, 20 to 30 points were randomly specified and processed in order to calculate the optical flow. For example, the area of the smoke image Im91 of FIG. 4 is calculated. Then, the optical flow may be calculated after randomly specifying the number of points according to the calculated area. Further, for example, the optical flow may be calculated after specifying one or more points with a point where the color change is larger than a predetermined size as a feature point.

Further, in SA4 of FIG. 3 of the above embodiment, the length of the side of the circumscribing rectangle of FIG. 5 is extracted as the scale of the product existence region, but for example, the area of the circumscribing rectangle or the length of the long side and The ratio of the lengths of the short sides or the area of the smoke image Im91 may be extracted as the scale of the product presence area.

Further, in SA4 of FIG. 3 of the above-described embodiment, a predetermined angle B3 of the product-corresponding region Im911 of FIG. 7 is used. For example, the spread of smoke is based on the captured monitoring region image. An angle may be detected and the detected angle may be used as the angle B3.

Alternatively, as another method, a temperature sensor may be arranged in the monitoring area, and the angle B3 may be changed according to the detection result of the temperature sensor. Since the smoke image becomes an acute angle as the calorific value rises, for example, if the temperature detected by the temperature sensor or the temperature rise rate within a predetermined time exceeds the predetermined angle, the angle B3 is set to an acute angle, and the temperature detected by the temperature sensor The angle B3 may be changed based on the degree of temperature rise.

Further, in SA4 of FIG. 3 of the above embodiment, the circumscribing rectangle is determined as the outer shape template, but for example, the circumscribing rectangle may be reduced to a predetermined scale.

(About correlation matching (1))
Further, in SA6 of FIG. 3 of the above embodiment, correlation matching was performed between the detection reference template generated in SA5 and the latest binarization candidate region image. For example, other than the detection reference template Im900 of the outer shape template Im921. Correlation matching with the latest binarization candidate region image may be performed for the region of.

(About correlation matching)
Further, in SA6 of FIG. 3 of the above embodiment, correlation matching was performed between the detection reference template generated in SA5 and the latest binarization candidate region image. For example, the detection reference template and two other than the latest are performed. A valued candidate area image, a monitored area image acquired by SA1, a monitored area image preprocessed by SA2, a superimposed binarized candidate area image generated by SA3, or a monitored area taken after generating a detection reference template. Correlation matching with the image may be performed.

(About shape)
Further, the shape of each image (including the circumscribing rectangle) or the template may be changed to an arbitrary shape (for example, a circle, an ellipse, a pentagon, etc.).

(About non-generation of detection criteria template)
Further, without generating the detection reference template, the detection reference template is recorded in advance in the recording unit 13 of FIG. 2, and a fire is caused based on the detection reference template and the monitoring area image captured by the photographing unit 11. It may be configured to detect. In this case, the generation unit 141 may be omitted.

(About features)
Further, the features of the above-described embodiment and the features of the modified example may be arbitrarily combined.

(Additional note)
The fire detection system of Appendix 1 is a fire detection system that detects a fire in the monitoring area based on a monitoring area image which is an image obtained by photographing the monitoring area, and is a photographing means for photographing the monitoring area image and the fire. A detection reference template for detecting a fire in the monitoring area and a fire detecting means for detecting a fire in the monitoring area based on the monitoring area image taken by the photographing means are provided.

In the fire detection system according to Appendix 1, the fire detection system of Appendix 2 is a product existence region, which is a region where a product is present due to the occurrence of the fire, based on the monitoring region image taken by the photographing means. The scale of the product and the moving direction of the product are specified, and based on the scale of the product existing region, an outer shape template showing the region including the product existing region is determined, and based on the moving direction of the product. The fire detection means is provided with a generation means for determining a product-corresponding region which is a region corresponding to the product-existing region and generating an overlapping region of the outer shape template and the product-corresponding region as the detection reference template. Detects a fire in the monitoring area based on the detection reference template generated by the generation means and the monitoring area image taken by the photographing means.

The fire detection system of Appendix 3 is the fire detection system of Appendix 2, wherein the generation means has two halves opened by a predetermined angle with respect to the movement direction with a predetermined point in the outline template as an end point. The region determined by the straight line is determined as the product-corresponding region.

The fire detection program of Appendix 4 is a fire detection program that detects a fire in the monitoring area based on the monitoring area image which is an image of the monitoring area, and is a detection reference template for detecting the fire in the computer. And, based on the monitoring area image taken by the photographing means for photographing the monitoring area image, it functions as a fire detecting means for detecting a fire in the monitoring area.

(Effect of appendix)
According to the fire detection system described in Appendix 1, for example, smoke generated by a fire is detected by detecting a fire in the monitoring area based on a detection reference template for detecting a fire and a captured monitoring area image. Since the fire can be detected without waiting for the arrival of the fire, the fire can be detected quickly. In particular, since the fire can be detected using the detection reference template, the fire can be detected by a relatively simple process such as comparison between the detection reference template and the monitoring area image, and the fire can be detected more quickly. Can be detected.

According to the fire detection system described in Appendix 2, for example, by generating a detection reference template based on the scale of the product existence area, which is the area where the product exists due to the occurrence of a fire, and the movement direction of the product. , When the product of a fire is smoke, it is assumed that the smoke exists in a predetermined area and that the inclination is affected by the wind that appears as the direction of movement. In such a case, it is possible to generate a region where smoke is predicted to exist based on the characteristics of the smoke, the detected scale, and the moving direction as a detection reference template, and to compare this detection reference template with the monitoring region image. Therefore, is it really due to the product of the fire, including whether the area detected as the product of the fire has the same characteristics as the area of the product of the fire such as smoke? Since the comparison is made, the fire can be detected quickly and reliably.

According to the fire detection system described in Appendix 3, a region determined by two half lines opened by a predetermined angle with respect to the moving direction is defined as a product-corresponding region with a predetermined point in the outline template as an end point. By determining, for example, the characteristics of the smoke that is the product (for example, the characteristics that spread horizontally as it rises and the way it spreads in the flow direction of the entire smoke due to the influence of wind, etc.) are detected as a detection reference template. Since it can be reflected in, the fire can be detected quickly and surely.

According to the fire detection program described in Appendix 4, for example, smoke generated by a fire is detected by detecting a fire in the monitoring area based on a detection reference template for detecting the fire and a captured monitoring area image. Since the fire can be detected without waiting for the arrival of the fire, the fire can be detected quickly. In particular, since the fire can be detected using the detection reference template, the fire can be detected by a relatively simple process such as comparison between the detection reference template and the monitoring area image, and the fire can be detected more quickly. Can be detected.

11 Imaging unit 12 Alarm unit 13 Recording unit 14 Control unit 91 Smoke 100 Detector 141 Generator unit 142 Fire detection unit 900 Installation surface Ar1 Arrow A11 Imaging area An1 Angle B1 Reference point B2 Reference line B3 Angle B4 Semi-straight line Im91 Background Image Im900 Detection Criteria Template Im911 Product Correspondence Area Im921 External Template

Claims (4)

A fire detection system that detects a fire in the monitoring area based on the monitoring area image, which is an image of the monitoring area.
An imaging means for capturing the surveillance area image and
A fire detection means for detecting a fire in the monitoring area based on a detection reference template for detecting the fire and the monitoring area image taken by the photographing means.
A fire detection system equipped with. Based on the monitoring area image taken by the photographing means,
The scale of the product existence area, which is the area where the product exists due to the occurrence of the fire, and the moving direction of the product are specified.
Based on the size of the product presence area, an outline template showing the area including the product presence area is determined.
Based on the moving direction of the product, the product-corresponding region, which is the region corresponding to the product-existing region, is determined.
A generation means for generating an overlapping region of the outline template and the product corresponding region as the detection reference template is provided.
The fire detecting means detects a fire in the monitoring area based on the detection reference template generated by the generating means and the monitoring area image taken by the photographing means.
The fire detection system according to claim 1. The generation means determines as the product-corresponding region a region determined by two half straight lines opened by a predetermined angle with respect to the movement direction, with a predetermined point in the outline template as an end point.
The fire detection system according to claim 2. A fire detection program that detects a fire in the monitoring area based on the monitoring area image, which is an image of the monitoring area.
Computer,
A fire detecting means for detecting a fire in the monitoring area based on a detection reference template for detecting the fire and the monitoring area image taken by the photographing means for capturing the monitoring area image.
A fire detection program that functions as.

Images