KR100858140B1 - Method and system for detecting a fire by image processing - Google Patents

Abstract

The present invention relates to a fire detection method and system using image processing, to accurately detect the occurrence of smoke and flames by photographing the monitoring target area, it is possible to quickly notify the occurrence of the fire caused by the smoke and flames.

In particular, the smoke and flame are separated from the background based on the RGB color values measured by processing the image captured by the camera, and the smoke and flame generation area is set, and the motion vector of the area is calculated to calculate the motion of the set area. By analyzing the degree and determining whether or not a fire occurred based on this, there is an effect that can inform the accurate occurrence of the fire.

Therefore, it is possible to improve the reliability of the fire detection system by accurately determining whether a fire occurs, as well as a rapid fire alarm.

Description

Fire detection method and system using image processing {Method and system for detecting a fire by image processing}

1 is a flowchart illustrating an example of a fire detection method using image processing according to the present invention.

2 is a block diagram showing an example of a fire detection system using the image processing according to the present invention.

3 is a diagram illustrating RGB values of feature points extracted from an example of an image photographed by the present invention.

4 is a diagram of extracting a motion vector for a flame from another example of an image processed by the present invention.

5 is a diagram extracting a motion vector for smoke from another example of an image processed by the present invention.

<Description of the symbols for the main parts of the drawings>

100: video input device 200: fire monitoring device

210: image processor 220: motion vector detector

230: smoke and fire judgment unit 300: administrator terminal

400: alarm device

The present invention relates to a fire detection method and system using image processing, to detect the occurrence of smoke and flames by photographing the monitoring target area, and to quickly notify the occurrence of fire according to the smoke and flames.

In particular, by setting smoke and flame prediction area by separating smoke and flame from the image taken by the camera, and determining the occurrence of fire by checking the movement of the set area, it is possible to inform the occurrence of fire quickly and accurately. It is to provide a fire detection method and system using the image processing.

In general, the fire detection system is installed in the monitored area using a sensor for detecting smoke or temperature, and there is a method using a camera to photograph the monitored area.

The method using the sensor for detecting the smoke or temperature is to determine whether a fire occurs by detecting a certain amount of smoke or a temperature above a certain temperature, or after the fire has been widespread, after a fire spreads a large amount of smoke To detect the occurrence of a fire in the state or the temperature is sufficiently raised, there was a problem that can not be quickly responded to the fire.

On the other hand, the method using a camera belongs to a characteristic of a general flame, that is, a series of red components in the RGB color coordinate system, and a value set by counting the number of pixels having red color in a color image by using such characteristics If it is abnormal, it is possible to solve the problem of the fire detection system using the sensor as a method of judging that a fire has occurred. However, if a fire alarm is generated due to a fire occurring for an object having a color similar to artificial lighting or flame Occurred, there was a problem that the reliability of the fire detection system is lowered.

Therefore, there is a need for a fire monitoring system capable of quickly and accurately detecting and informing a fire occurrence.

The present invention has been devised by the above-mentioned demands, and the method and system for detecting fire using image processing that can image the target area and accurately determine whether smoke and flames are generated and whether or not the fire occurs according to the result. To provide.

In particular, the present invention extracts the smoke and flame occurrence prediction region from the image data photographing the surveillance region, calculates the motion vector of the extracted prediction region, and determines whether smoke or flame has occurred in the corresponding smoke and flame occurrence prediction region. The present invention provides a fire detection method and system using image processing that can quickly and accurately determine the occurrence of a fire.

Fire detection method using the image processing according to the present invention for achieving the above object comprises the steps of: a) measuring the RGB value by image processing the image data photographed the area to be monitored; b) extracting a smoke generation prediction region and a flame generation prediction region based on the measured RGB values; c) calculating a motion vector of the extracted smoke generating region and the flame generating region; And d) issuing a smoke occurrence alarm or a fire occurrence alarm when the calculated motion vector is greater than the threshold.

In addition, the fire detection system using the image processing according to the present invention for achieving the above object, an image input apparatus for photographing the area to be monitored and output the image data, and receiving the image data output from the image input apparatus RGB Measure a value, extract a smoke generation prediction region and a flame generation prediction region based on the measured RGB value, calculate a motion vector of the extracted smoke generation prediction region and the flame generation prediction region, and calculate the calculated motion vector and threshold It is characterized in that made by including a fire detection value to output the alarm signal by determining the occurrence of smoke and fire by comparing the values.

Therefore, it is possible to respond quickly and accurately to fires occurring in the monitored area (indoor space, mountains, etc.).

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

First, briefly look at the principle of the present invention, the RGB color image characteristics of the smoke when a fire occurs is not much difference between the color values of red (R, red), green (G, green), blue (B, blue). In the flame region, red (R, red) is significantly higher in color than green (G, green).

In general, the flame properties belong to the family of red components in the RGB color coordinate system, red to yellow at low temperatures, and similar to white at high temperatures.

Accordingly, in the present invention, the smoke generation prediction region and / or the flame generation prediction region is first set by using the characteristics of the fire.

On the other hand, in Figure 3, the artifact area (X: 628, Y: 522) of the artifact color (RGB: 255, 254, 253) of the red, green, blue color appears similar to the smoke and also has a color value distribution similar to the flame region Therefore, it is necessary to distinguish between the artifact zone, the flame zone and the smoke zone.

Therefore, in the present invention, the motion vector of the corresponding region is measured to distinguish the artifact region from the smoke generation prediction region and / or the flame occurrence prediction region. / Or judging by the area of flame generation.

Hereinafter, a fire detection method using the image processing of FIG. 1 will be described with reference to the fire detection system using the image processing of FIG. 2.

When the image data of the monitoring target region captured by the image input apparatus 100 such as a camera is input to the fire monitoring apparatus 200 (S101), the image processing unit 210 of the fire monitoring apparatus 200 receives the input image. The RGB value of the data is measured (S102).

Hereinafter, a method of extracting a smoke generation prediction region and / or a flame generation prediction region from the image processor 210 will be described.

The measured RGB values are compared with the first set condition and the second set condition (S103) (S108) to extract the smoke generation prediction region and / or the flame generation prediction region (S104) (S109).

For example, the determination of the smoke region from the input color image value may be performed when the intensity of the pixel of the image is greater than 120 (the first reference value when the maximum brightness is 255) as shown in Equation 1 below. ), And the distribution of the red, green, and blue values of each pixel of the image satisfies the same condition as [Equation 2] below (when the maximum color value of each pixel is 255, which is smaller than the second reference value of 15). If so, the region is extracted as the smoke generation prediction region.

In other words, when the first set condition set as shown in [Equation 1] and [Equation 2] is satisfied (S103), the corresponding area is extracted as the smoke generation predicted area (S104).

The fire zone is determined from the input color image value by setting the pixels of which red (R) is larger than green (G) with respect to the input color image value as shown in [Equations 3] to [Equation 5] below. When the second set condition is satisfied (S108), the flame prediction region is extracted from the corresponding region (S109).

In other words, red (R) satisfies the same condition as [Equation 3] (when the maximum color value is 255, larger than the third reference value 75), red (R) is larger than green (G), [ When the condition as shown in Equation 5 (when the maximum brightness is 255, greater than the fourth reference value 175) is satisfied, the flame prediction region is extracted. Here, the first to fourth reference values may be applied to various values according to the needs of those skilled in the art.

The extracted smoke generation region and / or the flame generation region is extracted to the motion vector detection unit 220 of the fire monitoring apparatus 200, the motion vector detection unit 220 for the expected region The motion vector is detected and transmitted to the smoke and fire determination unit 230. The smoke and fire determination unit 230 distinguishes from an artificial region including an artificial structure based on the transmitted motion vector, and is not an artificial region. The smoke generation prediction area and / or the flame generation prediction area is determined as the smoke generation area and / or the flame generation area.

Hereinafter, a method of calculating the motion vector of the expected region by the motion vector detector 220 will be described in detail.

In calculating the motion vector, Equation 6 below is defined to detect the motion vector of the image data.

와

는 각각 영상에 대하여 공간적으로 수평, 수직성분에 대한 영상밝기의 미분값이며,

는 영상의 각 프레임 간 화소의 밝기 변화인 미분값이며,

와

는 각각 수평, 수직성분에 대한 광학적 흐름(optical flow)인 움직임이다. In Equation 6 above

Wow

Is the derivative of image brightness with respect to the horizontal and vertical components spatially with respect to each image.

Is the derivative that is the change in brightness of the pixels between each frame of the image,

Wow

Are motions that are optical flows for horizontal and vertical components, respectively.

,

라 하면, 하기의 [수학식 7] 및 [수학식 8]로부터 이를 반복적으로 구할 수 있다.Also, the motion for any pixel at position (x, y)

,

If it is, it can be obtained repeatedly from the following [Equation 7] and [Equation 8].

는 (x,y) 위치에 있는 임의의 화소에 주변화소 8개 화소의 평균 움직임이며,

는 0으로 나누어지는 것을 방지하기 위한 값이다.here,

Is the average movement of eight pixels around any pixel at position (x, y),

Is a value to prevent division by zero.

와

를 반복적으로 구하기 위하여

,

그리고

를 구하는 방법은 다음과 같다.In [Equation 7] and [Equation 8] above

Wow

To get it repeatedly

,

And

How to obtain is as follows.

는 (x,y) 위치에 있는 임의의 화소에 대한 [-1 -2 -1; 0 0 0; 1 2 1] 행렬을 곱하여 구하고

는 [-1 0 1; -2 0 2; -1 0 1] 행렬과 곱하여 구한다.

Is [-1 -2 -1 for any pixel at position (x, y); 0 0 0; 1 2 1] by multiplying the matrix

Is [-1 0 1; -2 0 2; -1 0 1] Find by multiplying by matrix.

는 현재 프레임의 각 화소에 대하여 임의의 앞 프레임의 각 화소와 [-1 1] 곱하여 구한다. 상기의 계산을 통하여 각각 공간상의 (x,y) 위치의 화소에 대한 미분값과 시간상의 미분값을 구할 수 있다.

Is obtained by multiplying each pixel of an arbitrary previous frame by [-1 1] with respect to each pixel of the current frame. Through the above calculation, the derivative value and the temporal derivative value for the pixel at the position (x, y) in space can be obtained, respectively.

는 각각 [0 1 0; 1 0 1; 0 1 0] 행렬과 그 전치행렬로 구하며,

,

,

와

를 구하는 절차에 따라서 각 프레임의 화소들에 대하여 계산을 하고, 움직임벡터

와

를 상기한 순서에 따라서 구한다.And, the average of the motion value of the surrounding pixels

Are each [0 1 0; 1 0 1; 0 1 0] Find the matrix and its transpose matrix.

,

,

Wow

Calculate the pixels of each frame according to the

Wow

Is obtained in the above-described order.

When the motion vector detection unit 220 of the fire monitoring apparatus 200 calculates a motion vector as described above with respect to the smoke occurrence region (S105) and transmits the motion vector to the smoke and fire determination unit 230, the smoke and fire determination unit ( If the calculated motion vector is larger than the first threshold, it is determined that the smoke is detected in the corresponding smoke generation prediction region (S106) and issues a smoke generation alarm (S107). Here, the first threshold value is a minimum value that can determine that the movement detected in the corresponding prediction area is smoke, and can be variously set according to the needs of those skilled in the art.

When the motion vector detection unit 220 of the fire monitoring apparatus 200 calculates a motion vector as described above with respect to the flame occurrence region (S110) and transmits the motion vector to the smoke and fire determination unit 230, the smoke and fire determination is performed. If the calculated motion vector is larger than the second threshold, the unit 230 determines that the flame is detected in the corresponding flame occurrence region (S111) and issues a fire occurrence alarm (S112). Here, the second threshold value is a minimum value that can determine that the motion detected in the corresponding prediction area is a flame, and can be variously set according to the needs of those skilled in the art.

In addition, the image data input in the step S101 is output through a separate display configured in the administrator terminal (for example, PC, mobile phone, PDA, etc.) 300 (S113), the smoke generation alarm and / or fire occurrence When an alarm is issued (S107) (S112), an alarm message or an alarm sound may be output through the manager terminal 300 or an alarm device (for example, a speaker or an LED) 400.

FIG. 3 is a diagram illustrating a motion vector measurement result and a flame region detection for the flame of the present invention, and FIG. 4 is a diagram showing a motion vector measurement result and a smoke region detection with respect to the smoke of the present invention.

As shown in FIG. 3, the left rectangular area is the flame area detected by measuring the color and motion vector of the detected flame, and the right figure shows the motion vector of the rectangular area.

In addition, the left side of FIG. 4 is a result of detecting smoke and measuring a motion vector with respect to a fire in which smoke occurs first, and a fire area detected by the inside of a rectangle, and the right side is a result of measuring a motion vector of a left rectangular area.

The fire detection method and system using the image processing according to the present invention has been described above. The technical configuration of the present invention may be implemented in other specific forms by those skilled in the art without changing the technical spirit or essential features of the present invention.

Therefore, the above-described embodiments are to be understood in all respects as illustrative and not restrictive, and the scope of the present invention is indicated by the appended claims rather than the foregoing description, and the meanings of the claims and All changes or modifications derived from the scope and the equivalent concept should be construed as being included in the scope of the present invention.

As described above, the present invention can accurately detect the occurrence of smoke and flames by photographing a monitoring target area, and can promptly notify the occurrence of a fire due to smoke and flames.

In particular, by setting the smoke and flame prediction area by separating the smoke and flame from the image taken by the camera, calculating the motion vector of the corresponding area, and determining the degree of movement of the set area to determine whether the fire occurred, It is effective to inform the exact occurrence of fire.

Therefore, it is possible to improve the reliability of the fire detection system by accurately determining whether a fire occurs, as well as a rapid fire alarm.

Claims (9)

a) measuring brightness values of red (R), green (G), and blue (B) by image processing the image data photographing the surveillance region; b-1) setting the smoke generation prediction region when the following Equations 1 and 2 are satisfied; [Equation 1]

[Equation 2]

b-2) setting the flame occurrence prediction region when the following Equations 3 to 5 are satisfied; [Equation 3]

[Equation 4]

[Equation 5]

c) calculating a motion vector of the extracted smoke generating region and the flame generating region; And and d) issuing a smoke generation alarm or a fire occurrence alarm when the calculated motion vector is greater than a threshold value. delete delete delete The method of claim 1, Step c) is c-1) calculating derivatives of image brightness, horizontal component motion, and vertical component motion with respect to the horizontal and vertical components of the corresponding prediction area; c-2) calculating a motion vector of the corresponding pixel based on the result of the calculation and the average motion of different pixels positioned around the corresponding pixel. An image input apparatus for photographing a surveillance region and outputting image data; Receives image data output from the image input device and measures brightness values of red (R), green (G), and blue (B), and predicts a smoke generation region and a flame occurrence based on the measured brightness values. After extracting the area, the motion vector of the smoke prediction area and the flame prediction area is calculated, and the smoke detection value which outputs an alarm signal by judging the occurrence of smoke and fire by comparing the calculated motion vector with a threshold value. Includes; The fire alarm market value, If the following equations (1) and (2) are satisfied, it is set as an expected smoke generation area. [Equation 1]

[Equation 2]

An image processor configured to set a flame generation predicted area when the following Equations 3 to 5 are satisfied; [Equation 3]

[Equation 4]

[Equation 5]

The derivative of the image brightness, the horizontal component motion and the vertical component motion with respect to the horizontal component and the vertical component of the corresponding prediction region extracted by the image processor are calculated, and based on the average motion of different pixels located around the pixel. A motion vector detector for calculating a motion vector of the corresponding pixel; And A smoke and fire determination unit outputting a smoke generation alarm signal when the motion vector calculated by the motion vector detection unit is greater than a first threshold value, and outputting a fire occurrence alarm signal when the motion vector is larger than a second threshold value; Fire detection system using the image processing, characterized in that configured. delete delete delete

Images