KR101076734B1 - Device for monitoring forest fire of information analysis type and method therefor - Google Patents

Abstract

The present invention relates to an information analysis type forest fire monitoring device and method for analyzing and diagnosing forest fire more accurately, and to periodically collect the weather observation detection data of at least one of the temperature, humidity and wind speed of the forest fire monitoring point. A sensor data collection unit, a peripheral data collection unit for periodically collecting the weather observation surrounding data for the periphery of the monitoring point from an external database, and a weather observation forecast data for the monitoring point periodically collected from an external database Forecast data collection unit, accumulation data collection unit for collecting accumulated multi-year meteorological observation accumulated data for the monitoring point from an external database, and the sensor data collection unit and peripheral data collection unit, forecast data collection unit and accumulated data Compare the weather observation data collected through the collection unit, respectively, When the detected data collected through the collecting unit exceeds the value of the weather observation data collected through at least one of the surrounding data collecting unit, the forecasting data collecting unit, and the accumulated data collecting unit, the wildfire occurrence state according to the set excess range It provides a central processing unit for determining step by step.

Description

Information analysis type forest fire monitoring device and its method {DEVICE FOR MONITORING FOREST FIRE OF INFORMATION ANALYSIS TYPE AND METHOD THEREFOR}

The present invention relates to forest fire monitoring, and more particularly, to an information analysis type forest fire monitoring apparatus and method for analyzing and diagnosing forest fire more accurately.

In the conventional forest fire monitoring system, a forest monitoring post is installed on the elements of a mountain where humans frequently enter and leave, and a forest monitoring agent is located at the post, visually monitoring the occurrence of a forest fire, and when a forest fire occurs, communication is installed on the post. It was in the form of informing the central control station by means.

Such a forest fire monitoring system may be the most reliable monitoring method within the scope of the vision while the surveillance personnel are monitoring the forest, but since the forest fire depends on the means of monitoring the forest fire, the monitoring range may be limited. Due to the limitation of the number of watchmen, there is a problem that 24 hours monitoring is impossible unless the watchman resides in the watchtower.

In order to improve such a problem, another method has been proposed to install an unmanned surveillance camera in an element of a mountain and to centrally manage it at the central control station to monitor and cope with the occurrence of a forest fire.

However, in the conventional unmanned surveillance system as shown in FIG. 1, when the system operator directly operates the camera controller 14 to transmit a control command to drive the camera 10, the control receiver 11 may receive a control command. After receiving the command through the central processing unit 13 drives the rotating machine 12 for driving the monitoring camera 10 to drive the camera in the desired direction, the situation of the field input from the camera 10 and The related information was outputted to the display device 15 through the central processing unit 13.

However, such a system inputs arbitrary virtual position information using a control system and a matrix switcher system through a control signal, and a preset function and a keyboard (operator) which calls the position. By using the key board (manual) to manually control the camera control device 14 has a function of tracking the scene of the fire occurs there is a problem that can not escape the elementary forest fire monitoring system.

According to the conventional system, since the camera control device was operated manually by the system operator to monitor the site of the forest fire, the site of the forest fire was increased due to the increased weight on the skilled functions and senses of the skilled operator for a long time. Not only does it take a long time to track, but there is a problem in that no countermeasures can be found at all in the event of the absence of surveillance personnel.

In addition, even if the camera is controlled automatically and unattended monitoring, it is difficult to distinguish whether the captured video is caused by forest fire or cloud or fog, so the reliability of reading data is low. There is a difficulty in automating.

The present invention is an information analysis type forest fire monitoring device that can analyze and predict forest fire more accurately by analyzing whether or not a fire occurs by using the monitoring data in the field as well as monitoring data and surrounding data, accumulated data and satellite data, etc. It is to provide a method.

The technical problems to be achieved by the present invention are not limited to the technical problems mentioned above.

Forest fire monitoring apparatus of the present invention for achieving the above object, the sensor data collection unit for periodically collecting at least one of the weather detection data of the temperature, humidity and wind speed for the forest fire monitoring point; Peripheral data collection unit for collecting the weather observation peripheral data for the surroundings of the monitoring point periodically from an external database; Forecast data collection unit for collecting the weather observation forecast data for the monitoring point periodically from an external database; An accumulation data collection unit for collecting accumulated years of meteorological observation accumulated data for the monitoring point from an external database; And the weather observation data collected through the sensor data collector, the peripheral data collector, the forecast data collector, and the accumulated data collector, respectively, and the sensed data collected through the sensor data collector is predicted from the peripheral data collector. It may include a central processing unit for judging the state of occurrence of fire in accordance with the set excess range when the value of the weather observation data collected through at least one of the data collection unit and the accumulated data collection unit.

The peripheral data collection unit may geometrically average by collecting peripheral data of at least three surroundings around the monitoring point.

The central processing unit generates prediction data by geometrically averaging forecast data and accumulated data collected through the forecast data collection unit and the accumulated data collection unit, and compares the generated prediction data with the sensed data to determine whether a forest fire occurs. Can be determined.

The forest fire monitoring method of the present invention for achieving the above object, the first step of periodically collecting at least one of the weather observation detection data of the temperature, humidity and wind speed for the forest fire monitoring point; A second step of periodically collecting meteorological observation data on the periphery of the monitoring point from an external database; A third step of periodically collecting meteorological forecast data for the monitoring point from an external database; A fourth step of collecting accumulated multi-year meteorological observation accumulated data for the monitoring point from an external database; A fifth step of comparing the detected data with the sensed data by using the collected ambient data, forecast data and accumulated data; And a sixth step of judging a state of occurrence of wildfire according to the degree of excess when the value of the collected sensed data exceeds at least one of the surrounding data, the forecasted data and the accumulated data as a result of the comparison. It may include;

In the second step, the surrounding data may be a value obtained by geometrically averaging weather observation data of at least three surroundings around the monitoring point.

The fifth step may include generating prediction data by geometrically averaging the collected forecast data and accumulated data, and comparing the generated prediction data with the sensed data; And comparing the sensed data collected above with the surrounding data.

Another forest fire monitoring apparatus of the present invention for achieving the above object, the external image collecting unit for periodically collecting at least one or more external images for the forest fire monitoring point from an external database; Peripheral image collecting unit for periodically collecting the peripheral image photographed by each of the observation equipment located in the vicinity of the monitoring point from an external database; A learning image generation unit for processing and combining a plurality of cumulative images of the same time zone with respect to the monitoring point to generate a learning image for the still image; A surveillance image receiver configured to periodically receive at least one surveillance image of the surveillance point through a camera rotated at a set angle; And a central processing unit configured to compare the frequency image and the histogram region of the color with at least one of the external image, the peripheral image, and the learning image collected in the surveillance image, to determine a forest fire state step by step. .

The external image may include at least one of a satellite image, a radar image, and a satellite image, a radar image, and a cloud photograph obtained from an aircraft.

The central processing unit compares the surveillance image and the learning image, the color of the edge of the image area that is different, the color of the image that is changed in the frequency domain of the surveillance image and the learning image, and the surveillance image, the external image, and the surroundings. The wildfire occurrence state may be determined step by step using the change in the histogram for the image.

Another forest fire monitoring method of the present invention for achieving the above object, the first step of periodically collecting at least one or more external images for the forest fire monitoring point from an external database; A second step of periodically collecting the surrounding images photographed by the observation equipment located around the monitoring point from an external database; Generating a learning image of a still image by processing and synthesizing a plurality of cumulative images of the same time zone with respect to the monitoring point; A fourth step of periodically receiving at least one surveillance image of the surveillance point through the camera rotated at a set angle; And comparing the surveillance image with colors acquired through edge analysis, frequency domain analysis, and histogram analysis of at least one of the collected external image, surrounding image, and learning image, to determine a forest fire occurrence step by step. The fifth step may include.

As described above, the present invention can analyze and predict forest fires more accurately by comprehensively using the monitoring data in the field as well as monitoring data, surrounding data, accumulated data, and satellite data. In addition, by showing the image of the forest fire location and at the same time automatically generates an alarm sound so that administrators can accurately recognize the forest fire location and whether the forest fire can minimize the damage caused by forest fire.

In addition, the present invention has the advantage of being able to monitor forest fires even unattended because of excellent ability to read forest fire by comprehensively analyzing the various information collected.

1 is a view showing a forest fire monitoring apparatus according to the prior art.
2 is a view showing a forest fire monitoring apparatus according to an embodiment of the present invention.
3 is a flowchart illustrating a forest fire monitoring process of FIG. 2.
4 is a view showing a forest fire monitoring apparatus according to another embodiment of the present invention.
FIG. 5 is a flowchart illustrating a forest fire monitoring process of FIG. 4.
6 and 7 are views illustrating a satellite image and a radar image, respectively.
8 is a diagram illustrating a method for obtaining a surveillance image of a surveillance camera.
9A and 9B are diagrams showing a histogram of a captured image.
10 is a view showing an example of data transmitted to the control station in the forest fire monitoring apparatus according to the present invention.

Hereinafter, with reference to the accompanying drawings will be described in detail a preferred embodiment of the present invention. Like elements in the figures are denoted by the same reference numerals wherever possible. In addition, detailed descriptions of well-known functions and configurations that may unnecessarily obscure the subject matter of the present invention will be omitted.

2 is a view showing a forest fire monitoring apparatus according to an embodiment of the present invention, forest fire monitoring device 100 is a monitoring sensor 110, sensor data collection unit 120, peripheral data collection unit 131, forecast data It comprises a collecting unit 133, the accumulated data collecting unit 135, the storage means 170, and the central processing unit 180.

Surveillance sensor 110 is installed in the monitoring point consists of a temperature sensor, a humidity sensor and anemometer to sense the temperature, humidity and wind speed of the monitoring area.

The sensor data collection unit 120 periodically monitors at least one of meteorological observation data of temperature, humidity, and wind speed from a monitoring sensor 110 installed at a monitoring point to monitor a forest fire, for example, a temperature sensor, a humidity sensor, and an anemometer. Collect.

The peripheral data collection unit 131 periodically collects meteorological observation peripheral data about the periphery of the monitoring point from an external database 200 through a network.

The forecast data collection unit 133 periodically collects weather observation forecast data for a monitoring point from an external database 200 through a network.

The accumulated data collection unit 135 collects accumulated multi-year meteorological observation accumulated data for the monitoring point from the external database 200 through the network.

Here, the external database 200 may be a database of the Korea Meteorological Agency or its equivalent.

The central processing unit 180 compares the meteorological observation data collected through the sensor data collector 120, the peripheral data collector 131, the forecast data collector 133, and the accumulated data collector 135, respectively. The sensor data collected through the sensor data collector 120 is collected by at least one of the peripheral data collector 131, the forecast data collector 133, and the accumulated data collector 135. When the value of the observed data is exceeded, the fire occurrence state is determined step by step according to the set excess range.

In the above, the peripheral data collection unit 131 collects the peripheral data of at least three surroundings around the monitoring point and geometrically averages them.

In addition, the central processing unit 180 generates the prediction data by geometrically averaging the forecast data and the accumulated data collected through the forecast data collection unit 133 and the accumulated data collection unit 135, and generating the predicted data and The detection data may be compared with each other to determine whether a wildfire has occurred.

In the storage means 170, each of the weather observation data collected through the sensor data collection unit 120, the peripheral data collection unit 131, the forecast data collection unit 133, and the accumulated data collection unit 135 are coordinate positions ( GPS information) can be stored for each time zone.

The operation of the forest fire monitoring device configured as shown in FIG. 2 will be described with reference to the flowchart of FIG. 3.

First, the sensor data collection unit 120 periodically collects at least one weather observation detection data of temperature, humidity, and wind speed from the monitoring sensor 110 installed at the forest fire monitoring point (S11). Here, the collected sensing data is shown in Table 1 below.

TABLE 1

The surrounding data collection unit 131 may periodically collect meteorological observation surrounding data of the surroundings of the monitoring point from the external database 200 through the network (S12). The surrounding data may be meteorological observation data for at least three surroundings around the monitoring point, and the surrounding data may be a geometrically averaged value as shown in Equation 1 below.

Equation 1

Here, a, b and c are ambient data (temperature, humidity and wind speed data) measured at each observation point located around the monitoring point, and t is a geometric mean value of a, b and c. The geometric mean is more error resistant than the arithmetic mean, and has a feature to grasp the center of gravity of the data.

Subsequently, the forecast data collection unit 133 may periodically collect meteorological forecast data for the monitoring point from the external database 200 (S13). In the present invention, when collecting data from the external database 200, etc., the web is connected to the DB of the Meteorological Agency, which is set in advance, to send a web query based on the HTTP protocol. Subsequently, when the web data is received from the external database 200, the web data is converted into a system analysis data format as shown in Table 1 and stored in the storage means 170. If the data received through the network is compressed, the collecting unit will decode the data.

In addition, the accumulated data collection unit 135 may collect the accumulated multi-year meteorological observation accumulated data for the monitoring point from the external database 200 (S14).

The central processing unit 180 compares the sensed data using the collected ambient data, the forecast data and the accumulated data, respectively, and the sensed data collected as a result of the comparison is selected from the surrounding data, the forecast data, and the accumulated data. When at least one value is exceeded, the wildfire occurrence state may be determined step by step according to the excess degree (S15, S16).

For example, the central processing unit 180 collects the temperature, humidity, and wind speed data accumulated for the past 10 years for the monitoring point, and then calculates an average value of the accumulated data through the geometric mean, and forecasts the current data for the monitoring point. Predicted data can be obtained by geometric mean of the accumulated data.

The predicted data and the sensed data obtained as described above may be compared with each other, and the sensed fire state may be determined step by step by comparing the collected sensed data with the surrounding data. For example, the central processing unit 180 may diagnose the heat wave caution alarm when the sensing data (temperature) of the monitoring point is 10% or more larger than the predicted data (temperature) and the sensing data increases by 10% or more for 1 minute. Can be diagnosed with heat wave warning alarm and fire warning alarm when the detected data is more than 20% larger than the predicted data, and the detected data is increased by more than 20% for 1 minute, the detected data is more than 30% larger than the predicted data, When the detected data is increased by more than 30% for 1 minute, it can be diagnosed with heat wave alarm or fire alarm, and when the detected data is more than 30% larger than the surrounding data, it can be diagnosed with fire alarm or equipment error caution alarm.

In addition, when the sensed data (wind speed) is 100% or more larger than the predicted data (wind speed), it can be diagnosed with a strong wind warning, and when the sensed data is 200% or more larger than the predicted data, it can be diagnosed with a strong wind alarm. In situations where the detected data (humidity) is less than 10%, a fire caution alarm can be diagnosed. Here, the caution alarm stage is when the probability of occurrence is 33% or more, and the alert stage is when the probability of occurrence is 66% or more.

The CPU 180 may display various collected data on the display unit 191 or display the analysis result on the display unit 191.

In addition, the central processing unit 180 controls to generate an alarm signal through the alarm generator 195 or to transmit a fire emergency message to a mobile communication terminal of a set manager when the analyzed fire occurrence corresponds to a caution alarm or an alarm. You may.

4 is a view showing a forest fire monitoring apparatus according to another embodiment of the present invention, the forest fire monitoring apparatus 100 is a camera 140, surveillance image receiver 150, external image collector 161, peripheral image collector 163, the learning image generation unit 165, the storage unit 170, and the central processing unit 180 are included.

The camera 140 is installed at the top of the mountain to be monitored, and is set with the downward direction of the mountain as the viewing angle. The camera 140 may be an infrared camera or a heat sensing camera configured to rotate at an angle by the angle adjuster 145.

The surveillance image receiver 150 periodically receives at least one surveillance image of the surveillance point through the camera 140 rotated at a set angle.

The external image collecting unit 161 periodically collects at least one external image of the forest fire monitoring point from the external database 200 through the network. Here, the external image may include at least one of a satellite image, a radar image, and a cloud image obtained from an aircraft.

The surrounding image collecting unit 163 periodically collects the surrounding images photographed by the observation equipment located around the monitoring point from the external database 200. Here, the surrounding image collecting unit 163 may collect surrounding images of at least three surroundings based on the monitoring point by using the position of the monitoring point and the azimuth information with the monitoring point.

The learning image generator 165 generates a learning image for the still image by processing and combining a plurality of cumulative images of the same time zone with respect to the monitoring point.

The central processing unit 180 acquires the surveillance image through an edge analysis of at least one of the collected external image, peripheral image, and learning image, analysis of frequency domain, and histogram analysis. Compare the color to determine the wildfire occurrence step by step.

In addition, the central processing unit 180 compares the surveillance image and the learning image, the color of the edge of the image area that is different, the color of the image is changed in the frequency domain for the surveillance image and the learning image, and the surveillance image and Forest fire occurrence can be determined step by step using the histogram of the external and surrounding images.

The storage unit 170 may store each weather observation image collected through the external image collecting unit 161, the peripheral image collecting unit 163, and the surveillance image receiving unit 150 by time zone according to the coordinate position (GPS information). Can be.

The operation of the forest fire monitoring device configured as shown in FIG. 4 will be described below with reference to the flowchart of FIG. 5 and the accompanying drawings.

First, the external image collecting unit 161 may periodically collect at least one or more external images of the forest fire monitoring point from the external database 200 through the network (S21). Here, the external image may include at least one of a satellite image, a radar image, and a cloud image obtained from an aircraft. The external database 200 may be a database of the Korea Meteorological Agency or its equivalent. The external image may be a current image, but may be a forecast image after 15 or 30 minutes.

In the case of satellite images, as shown in FIG. 6, clouds, fog, water vapor, and the like can be measured. Satellite images may include satellite images and infrared images. In the case of the radar image, as shown in FIG. 7, the amount of water vapor can be measured. For example, in the case of yellow or more, it means the rainfall (10 mm / h), thereby invalidating the forest fire judgment.

When collecting the external image through the external database 200, the location information (GPS information) of the monitoring point and the GPS origin of the external image is collected and collected, and the reference point of the monitoring point from the external image reference point, for example, 1km * 1km Obtained by converting the size to image coordinates.

In addition, the peripheral image collecting unit 163 may periodically collect the peripheral image photographed by each of the observation equipment located around the sensing point from the external database 200 through the network (S22). Here, the surrounding image collecting unit 163 may collect surrounding images of at least three surroundings based on the monitoring point by using the position of the monitoring point and the azimuth information with the monitoring point. That is, the azimuth angle is calculated from the position information (GPS) of the current monitoring point and the position information (GPS) of the first surrounding station, and the captured image of the first surrounding station is received. Calculate the azimuth angle from the position information of the station to receive the captured image of the second neighboring station, and calculate the azimuth from the position information of the current monitoring point and the position information of the third neighboring station to receive the captured image of the third neighboring station. Can be.

In addition, a learning image of a still image may be generated by processing and synthesizing a plurality of cumulative images of the same time zone with respect to the monitoring point (S23). For example, a spatial sharpening filter is applied to an image, and then an edge detection filter of the image is applied to synthesize, for example, 10-day cumulative images of the same time zone (only when passing the same data band during synthesis). In addition, by applying a spatial sharpening filter on the image and then applying a noise canceling filter on the image, for example, a 10-day cumulative image of the same time period is synthesized (only when passing through the same data band). Subsequently, the edge detection cumulative composite image and the noise reduction cumulative composite image may be synthesized again to generate a learning image for the still image. This is a process of acquiring the outlines of fixed subjects (mountains, trees, rocks, etc.) except clouds and fog.

Subsequently, as shown in FIG. 8, the central processing unit 180 controls the angle adjusting unit 145 to rotate the camera 140 installed at the monitoring point at a set angle (90 °), and the surveillance image receiving unit 150. At least one surveillance image for the monitoring point is periodically received through S24. Accordingly, the central processing unit 180 obtains four images (0 to 90 degrees, 90 to 180 degrees, 180 to 270 degrees, 270 to 360 degrees) according to the viewing angle of the camera 140, and thus the satellite of the GPS reference point. Or you get a full image that compares to aerial images.

The central processing unit 180 compares the received surveillance image as described above with at least one of the collected external image, the surrounding image, and the learning image and the frequency region and the histogram region of the color in order to determine a wildfire occurrence state step by step. It is made (S25).

In S25, the central processing unit 180 detects image edges of the surveillance image and the training image, and compares each detected edge to obtain a difference region, and obtains a color of the edge represented by the acquired difference region. Detect. The fire probability may be calculated according to the number of pixels for each color detected. That is, the edge detection filter is applied to the training image and the current surveillance image to detect the edge, and the detected edge image is differentiated to obtain a difference region for different edges, and to detect the color of the edge represented by the difference region. do. For detected edge colors, for example, it is possible to add + 1% fire probability per 6 pixels of red edge, + 0.5% fire probability per 4 pixels of yellow edge, and + 2% fire probability per 4 pixels of black edge. .

In addition, the central processing unit 180 converts each of the learning image and the surrounding image into a frequency domain through fast Fourier transform, compares each signal in the frequency domain, obtains a differential image, and performs a fast Fourier with respect to the difference image. The color of the difference image may be obtained through inverse transformation. And, the fire probability can be calculated according to the number of pixels for each color detected above. In other words, it acquires image data in the frequency domain by fast Fourier transforming the learning image, and performs the fast Fourier transform on the surrounding images (three peripheral images) for the current monitoring point, and then separates the surrounding image and the learning image and shows fog or smoke. Acquire images such as. Then, by performing fast Fourier inverse transformation on the difference image to obtain an image of fog or smoke, add + 1% fire probability per 6 red pixels, + 0.5% fire probability per 4 yellow pixels, and black 4 You can add + 2% fire probability per pixel.

In addition, the central processing unit 180 extracts histograms indicating cumulative quantities of colors for luminance components from the surveillance image, the external image, and the surrounding image, respectively, as shown in FIG. 9A, and extracts the histograms of the extracted surveillance image from the external image and the peripheral image. Compared with each other, the probability of fire occurrence is calculated according to the white distribution as shown in FIG. 9B. Here, the histogram represents the pixel distribution from the dark part to the bright part of the image, and the left end is 0 (black) and the right end is 255 (white). If the image is bright due to white clouds or the like, the pixel distribution (black portion) will be focused to the right as shown in FIG. 9B.

Therefore, when the white histogram obtained from the current surveillance image exceeds 10% or more of the surrounding image occurrence probability and the external image occurrence probability, the central processing unit 180 adds the fire occurrence probability by + 1%, and the gray obtained from the current surveillance image. When the histogram exceeds the probability of surrounding images and the occurrence of external images by more than 10%, the probability of fire is added + 2%. If the result of the fire occurrence obtained through color and histogram analysis in the color and frequency domain for the edge exceeds 25%, a fire warning alarm (33% probability of fire occurrence) is predicted, and the fire occurrence obtained through the above process is generated. If the reading for more than 50%, a fire alarm (66% probability of fire) is forecast. The central processing unit 180 may generate an alarm signal through the alarm generating unit 195 or control the fire emergency message to be transmitted to a mobile communication terminal of a set manager when the analyzed fire occurrence corresponds to a caution alarm or an alarm. have.

That is, in the present invention, the weather forecast data collected by the Korea Meteorological Administration and its corresponding institutions, the database accumulated by meteorological observations, and sensor data obtained from the current monitoring point are combined to compare the predicted weather and the collected information of the current point, By analyzing the real-time ground image obtained from satellites or aerial photographs and the image data of the current point obtained through the image acquisition system such as a camera, it is possible to more accurately observe and forecast the forest fires to be measured.

On the other hand, after the forest fire monitoring device 100 compresses all the read data, it can be sent to the control station 300 along with the data on whether or not the fire occurred through the network as shown in FIG. To this end, the forest fire monitoring apparatus 100 overlays the monitoring data (temperature, humidity, wind speed, time, etc.) collected through a predetermined image processing unit into the image data, and determines whether natural disaster has occurred in the image data. After recording and compressing the recorded data, the compressed data is transmitted to the control station 300 through the network.

The control station 300 generates an alarm signal when the fire occurrence determination result is more than the caution alarm stage, and also enables the administrator to easily know the emergency situation through different display (screen, alarm, etc.) at each step. In the control station, for example, in the alarm phase, text, email, and voice can be controlled to be transmitted to the set administrator 10 seconds after the occurrence of the situation. Can be controlled.

As described above, the present invention installs an imaging device (camera) on the mountain top to be managed and reports the captured image and the angle of the camera to the central processing unit 180, and the central processing unit 180 captures the captured image. If it is determined to be a wildfire by analyzing the forest fire location through the display unit 191 and at the same time automatically generates an alarm sound through the alarm generating unit 195 to allow the administrator to correctly recognize the forest fire location and whether the forest fire to the forest fire Damage caused by this can be minimized. In addition, as a result of comprehensive analysis of the collected information, the forest fire reading ability is excellent, so that the forest fire can be monitored unattended.

Forest fire monitoring according to the present invention may be analyzed by combining both the weather observation data and the results of the observation image, it is also possible to predict whether a fire occurs using the weather observation data and each observation image. In the embodiment of the present invention, a method of analyzing the occurrence of a fire by using meteorological observation data and the observed image was described. Of course, the analysis of fire occurrence monitoring results using meteorological observation data and fire occurrence monitoring results using observation images may be more accurate.

Such forest fire monitoring is not limited to the configuration and manner of operation of the embodiments described above. The above embodiments may be configured such that various modifications may be made by selectively combining all or part of the embodiments.

100: forest fire monitoring device 110: monitoring sensor
120: sensor data collector 131: peripheral data collector
133: forecast data collector 135: accumulated data collector
140: camera 150: surveillance video receiver
161: external image collecting unit 163: peripheral image collecting unit
165: learning image generation unit 170: storage means
180: central processing unit

Claims (16)

A sensor data collection unit for periodically collecting at least one meteorological observation data of a temperature, humidity, and wind speed of a forest fire monitoring point;
Peripheral data collection unit for collecting the weather observation peripheral data for the surroundings of the monitoring point periodically from an external database;
Forecast data collection unit for collecting the weather observation forecast data for the monitoring point periodically from an external database;
An accumulation data collection unit for collecting accumulated years of meteorological observation accumulated data for the monitoring point from an external database; And
The weather data collected through the sensor data collector, the peripheral data collector, the forecast data collector, and the accumulated data collector are compared, respectively, and the sensed data collected by the sensor data collector is the peripheral data collector and the forecast data. Information processing type forest monitoring including; central processing unit for judging the state of occurrence of fire in accordance with the set excess range when the value of the meteorological observation data collected through at least one of the collection unit and the accumulated data collection unit; Device.
The method according to claim 1,
And the peripheral data collection unit collects peripheral data of at least three surroundings around the monitoring point and geometrically averages them.
The method according to claim 1,
The central processing unit generates prediction data by geometrically averaging forecast data and accumulated data collected through the forecast data collection unit and the accumulated data collection unit, and compares the generated prediction data with the sensed data to determine whether a forest fire occurs. Information analysis forest fire monitoring device to determine the.
The method according to claim 1,
And said external database is a meteorological observation database of the Meteorological Administration or its equivalent.
Periodically collecting weather detection data of at least one of temperature, humidity, and wind speed at a forest fire monitoring point;
A second step of periodically collecting meteorological observation data on the periphery of the monitoring point from an external database;
A third step of periodically collecting meteorological forecast data for the monitoring point from an external database;
A fourth step of collecting accumulated multi-year meteorological observation accumulated data for the monitoring point from an external database;
A fifth step of comparing the detected data with the sensed data by using the collected ambient data, forecast data and accumulated data; And
A sixth step of judging a wildfire occurrence state according to the degree of excess when the value of the collected sensing data exceeds at least one of the surrounding data, the forecast data, and the accumulated data as a result of the comparison; Information analysis forest fire monitoring method comprising a.
The method according to claim 5,
In the second step, the surrounding data is an information analysis forest fire monitoring method of the geometrically averaged value of the weather observation data for at least three places around the monitoring point.
The method according to claim 5,
The fifth step may include generating prediction data by geometrically averaging the collected forecast data and accumulated data, and comparing the generated prediction data with the sensed data; And comparing the sensed data collected from the collected data and surrounding data with each other.
An external image collecting unit periodically collecting at least one external image of a forest fire monitoring point from an external database;
Peripheral image collecting unit for periodically collecting the peripheral image photographed by each of the observation equipment located in the vicinity of the monitoring point from an external database;
A learning image generation unit for processing and combining a plurality of cumulative images of the same time zone with respect to the monitoring point to generate a learning image for the still image;
A surveillance image receiver configured to periodically receive at least one surveillance image of the surveillance point through a camera rotated at a set angle; And
And a central processing unit for judging a state of occurrence of forest fire by comparing the surveillance image with at least one of the collected external image, surrounding image, and learning image with a frequency region and a histogram region of color. Forest fire monitoring device.
The method according to claim 8,
And the surrounding image collecting unit collects surrounding images of at least three surroundings based on the monitoring point by using the position of the monitoring point and azimuth information with the monitoring point.
The method according to claim 8,
The external image is an information analysis forest fire monitoring device comprising at least one of a satellite image, radar, and satellite image obtained from the aircraft, radar image and cloud image.
The method according to claim 8,
The central processing unit compares the surveillance image and the learning image, the color of the edge of the image area that is different, the color of the image that is changed in the frequency domain of the surveillance image and the learning image, and the surveillance image, the external image, and the surroundings. An information analysis forest fire monitoring device that uses a change in the histogram of an image to determine a forest fire step by step.
Periodically collecting at least one external image of a forest fire monitoring point from an external database;
A second step of periodically collecting the surrounding images photographed by the observation equipment located around the monitoring point from an external database;
Generating a learning image of a still image by processing and synthesizing a plurality of cumulative images of the same time zone with respect to the monitoring point;
A fourth step of periodically receiving at least one surveillance image of the surveillance point through the camera rotated at a set angle; And
A step of judging a wildfire occurrence state by comparing the surveillance image with colors acquired through edge analysis, frequency domain analysis, and histogram analysis of at least one of the collected external image, surrounding image, and learning image; Information analysis type fire monitoring method comprising;
The method of claim 12,
The external image in the first step is an information analysis forest fire monitoring method comprising at least one of satellite, radar, and satellite image, radar image and cloud photograph obtained from the aircraft.
The method of claim 12,
The fifth step may include detecting image edges of the surveillance image and the training image, respectively; Comparing each detected edge to obtain a difference region; Detecting a color for the edge represented by the difference region; And calculating a fire probability according to the number of pixels for each color detected.
The method of claim 12,
The fifth step may include converting the learning image and the surrounding image into a frequency domain through fast Fourier transform; Comparing the respective signals in the frequency domain to obtain a differential image; Acquiring a color of the difference image through a fast Fourier inverse transform on the difference image; And calculating a fire probability according to the number of pixels for each color detected in the above.
The method of claim 12,
The fifth step may include extracting histograms representing cumulative quantities of colors of luminance components in the surveillance image, the external image, and the surrounding image, respectively; And comparing the histogram of the surveillance image with at least one histogram of an external image and an ambient image, respectively, and calculating a fire occurrence probability according to a white distribution amount.

Images