JPH0520560A - Fire point detecting device - Google Patents

Abstract

PURPOSE:To correctly perform the distance measurement, etc., by specifying the same fire point which can be seen from two television cameras even at the time of the fire source caught as a surface on a screen while this is the device to detect the fire point (one point in the fire source) in a warning area by using the image of the fire image-picked up with two color television cameras. CONSTITUTION:R, G and B components of an image image-picked up by two color television cameras 10a and 10b at the time of fire occurrence are extracted and stored 46 and 48, and further, the ratio of two components of the R, G and B components is calculated 52. Continuously, with each picture element of two images as an attention picture element, a matrix is applied 54 in the sequence, the sum of the square error of the picture element included in the matrix applied for two images is calculated 56 as a parameter to show similarity, the matrix having the minimum value in the sum of the square error obtained concerning the whole combination of the matrix applied for two images is selected 58 as the strongest similarity and based on the position of the central picture element of two selected matrixes and the scanning angle of a television camera, the fire point is decided 60.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a fire point detecting device for detecting a fire point (one point in a fire source) in a warning area by using a fire image taken by two color television cameras. .

[0002]

2. Description of the Related Art Conventionally, as a fire monitoring device using a TV camera, a TV camera monitors a monitoring area,
It is known that the fire extinguishing nozzle is controlled by detecting the distance and direction to the fire source by a video signal from a television camera when a fire occurs (Japanese Patent Laid-Open No. 1-268572).

In detecting the distance to the fire source in this case, the position of the luminance signal exceeding a predetermined level is detected as a fire point, and the number of pulses from the rise of the horizontal and vertical synchronizing signals to the fire point detection position is counted, By previously determining the distance per pulse, the distance is obtained from the pulse count value. However, since detection of the distance to the fire source by such image processing is inaccurate, two TV cameras 1a and 1b are installed in a large space structure as shown in FIG. Based on the scanning angles α1 and α2 and the installation distance L when the fire source 2 is captured by the TV cameras 1a and 1b, the fire source 2 is supplied from the fire extinguishing nozzle 3 installed in the middle of the TV camera 1 according to the principle of triangulation. The distance X to is calculated.

[0004]

By the way, in the distance measurement based on the principle of triangulation using such two TV cameras, since the target is a large space structure, the TV camera 1 fires. Since the distance to the source 2 is sufficiently long and the fire source 2 can be captured as a point on the screen of the television camera 1, the distance can be accurately measured.

However, in a normal fire monitoring in a small space such as a room, since the monitoring distance from the television camera is short, the fire source 2 is caught as a surface on the screen as shown in FIG. , The position of the fire source 2 viewed from the television cameras 1a and 1b is different from P and Q, and as a result, if the distance is obtained by the principle of triangulation, the distance to the R point farther than the fire source 2 will be obtained. There was a problem that the measurement error of the distance became large.

The present invention has been made in view of the above-mentioned problems of the prior art. Even if a fire source is caught as a surface on the screen, a fire point common to two TV cameras is specified and the distance is set. It is an object of the present invention to provide a fire point detection device capable of accurate measurement and the like.

[0007]

To achieve this object, the present invention is constructed as follows. The reference numerals in the drawings of the embodiments are also shown. That is, the fire point detection device according to the present invention includes the first and second color television cameras 10 for imaging the surveillance area.
a, 10b and first and second color television cameras 10
Scanning mechanism for scanning a and 10b in the horizontal and vertical directions (1
2, 14, 24, 26) and the first and second when a fire occurs
First imaged by color television cameras 10a, 10b
And image memories 46 and 48 for extracting and storing R, G, and B components for each pixel of the second images 62 and 64, and the first image 62 and the second image 6 stored in the image memories 46 and 48.
A ratio calculation unit 52 that calculates the ratio of two components for each pixel using the R, G, and B components of 4, and the number of pixels determined in advance as each pixel of the first image 62 and the second image 64 n
A matrix scanning unit 54 that performs matrix scanning in which (for example, n = 8) matrices A and B are sequentially applied, and a component ratio by a ratio calculation unit 52 of a plurality of pixels included in the matrix A that scans the first image 62 A square error calculator 56 that calculates the sum of the square errors of the component ratios by the ratio calculator 52 of a plurality of pixels included in the matrix B that scans the two images 64.
And the squared error e of the component ratios obtained by the squared error calculator 56 for all combinations of the matrices A and B in which all the pixels of the first image 62 and the second image 64 are applied as the target pixel.
_The minimum value selection unit 58 that selects the minimum value in _ij , the scanning angle information (α, β1, β2) of the first and second color television cameras 1a and 1b, and the minimum value selected by the minimum value selection unit 58. A fire point determination unit 60 that determines a fire point based on the center pixels A _ij and B _ij of the matrix applied to the first image 62 and the second image 64 that give values is provided.

Here, the ratio calculator 54 calculates the ratio of the B component to the R component (B / R) or the ratio of the G component to the R component (G / R).
R) is calculated. In addition, the square error calculation unit 56 sets the component ratio of _n pixels included in the matrix A applied to the first screen 62 to μ _n , and the component ratio of _n pixels included in the matrix B applied to the second screen 64. When the ratio is ε _n , the total sum of squared errors e _ij between the two matrices A and B is

[0009]

[Equation 2]

It is calculated as Furthermore, in order to simplify the calculation when determining the fire point, the scanning mechanism is controlled when a fire is detected to control the first and second television cameras 10a and 10b.
After centering so that the image of the fire source comes to the center of the screen, the image data is stored in the image memories 46 and 48.

Further, in order to reduce the calculation load on and after the ratio calculation section 52, it is desirable to cut out and process a predetermined range of images including the fire source from the images stored in the image memories 46 and 48. Further, when the processing speed is prioritized, the first
The pixels of the screen 62 and the second screen 64 to be processed may be processed with a block pixel, which is a block of a predetermined number, as one unit.

[0012]

According to the fire spot detecting device of the present invention having such a structure, the following actions can be obtained. The basic principle of the present invention is
For example, a 3 × 3 matrix is applied to each of the images of the fire source captured by two TV cameras, the similarities of the pixels included in the two matrices are compared, and the pixel of interest (center of Pixel) is determined as a common fire point in the two images.

In the present invention, the ratio B / R or G / R is used as the ratio of the R, G and B components of the pixel for the comparison of the similarities by applying this matrix. As the parameter indicating the similarity, the sum of the squared errors of the component ratios of the two matrices to be compared is introduced, and the applied portion of the matrix of each image having the smallest squared error has the strongest similarity,
That is, it is determined that the flame images match each other, and the pixel of interest (center pixel) in this matrix is determined as a fire point indicating the same point on the fire source viewed by the two color television cameras.

Since the fire points viewed from the two color television cameras are the same point in this way, even if the fire source is caught as a surface on the screen, the fire point position can be accurately specified and the fire point, for example, can be fired. The distance to the source can be measured.

[0015]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a block diagram of an embodiment showing one embodiment of the present invention. In FIG. 1, 10a is a first television camera, and 10b is a second television camera. As shown in FIG. 8, the television cameras 10a and 10b are installed at a predetermined installation interval L so that the warning area can be monitored.

The color television camera 10a is provided with a scanning mechanism having a horizontal rotation motor 12 and a vertical rotation motor 14 so that the optical axis of the television camera 10a can be directed to an appropriate position in the monitoring area. There is. The horizontal rotation motor 12 is driven by a drive circuit 16, and the vertical rotation motor 14 is driven by a drive circuit 18. Further, the horizontal scanning angle of the color television camera 10a by the horizontal rotation motor 12 is detected by the horizontal scanning angle sensor 20, and the vertical scanning angle of the color television camera 10a by the vertical rotation motor 14 is detected by the vertical scanning angle sensor 22. It

Similarly, for the color television camera 10b, the scanning mechanism drive circuits 28 and 30 having the horizontal rotation motor 24 and the vertical rotation motor 26, and the horizontal scanning angle sensor 32 are provided.
And a vertical scanning angle sensor 34 are provided. Reference numeral 40 is a microprocessor (hereinafter referred to as “MPU”) that executes the fire position detection processing of the present invention, and realizes the functions of the circuit units shown inside by program control.

First, the MPU 40 is provided with a scan controller 42 and a scan angle detector 44. In the normal monitoring state, the scanning control unit 42 positions the color television cameras 10a and 10b in a predetermined direction overlooking the entire warning area,
When a fire is detected, for example, the color TV camera 10
The color television cameras 10a and 10b are centered so that the position of the fire source in the image obtained from a, 10b, that is, the brightness signal indicating the flame is located at the center of the screen.

The scanning angle detector 44 is the color television camera 1.
The angle detection signals from the horizontal scanning angle sensors 20, 32 and the vertical scanning angle sensors 22, 34 provided in 0a, 10b are input, and the intersection angle α of the television camera with respect to the fire point is determined based on the horizontal scanning angles α1, α2, for example. It is calculated and output, and the vertical scanning angles β1 and β2 are output as they are. Image signals from the color television cameras 10a and 10b are read into the image memories 46 and 48, respectively, when a fire is detected. The image memories 46 and 48 are color television cameras 10.
Since an image signal for one screen is obtained for each RGB from a, 10b, R, G, B, and luminance components 4 are stored.
It consists of one frame memory.

An image cutout unit 50 is provided following the image memories 46 and 48. The image cutout unit 50 cuts out an image in a predetermined range including the fire source in each image stored in the image memories 46 and 48 in response to the detection of a fire as a processed image for detecting a fire position, and gives the cutout image to the ratio calculation unit 52. . The ratio calculation unit 52 calculates the component ratio of the R, G, and B components for the cutout image cut out by the image cutout unit 50. In this embodiment, the ratio B / R of the B component and the R component is calculated. The ratio calculated by the ratio calculation unit 52 may be the ratio G / R of the G component and the R component other than B / R.

A matrix scanning unit 54 is provided subsequent to the ratio calculation unit 52. The matrix scanning unit 54 applies, for example, a 3 × 3 matrix to each pixel for each clipped image of the color television cameras 10a and 10b by the image clipping unit 50. When the matrix is applied in the matrix scanning unit 54 with specific pixels in the two cutout images as the target pixel, the square error calculation unit 56 expresses the similarity between the two matrices by 2 for each matrix constituent pixel. The sum of power errors is calculated using the B / R ratio.

The matrix scanning section 54 and the square error computing section 5
The processing by 6 is obtained for all combinations of matrices applied to each pixel of the two cutout images obtained from the color television cameras 10a and 10b. A minimum value selection unit 58 is provided subsequent to the squared error calculation unit 56, and the squared error of the squared error calculated by applying the matrix to all the pixels of the two images calculated by the squared error calculation unit 56 is calculated. The minimum value is selected as the one with the highest degree of similarity from the total sum.

A fire point determination unit 60 is provided subsequent to the minimum value selection unit 58, and a pixel of interest in a matrix of two images that gives the sum of squared errors that are the minimum values selected by the minimum value selection unit 58, That is, the position of the central pixel of the matrix is determined as a fire point that looks the same when the fire source is viewed from the two color television cameras 10a and 10b, and the scanning angles α, β1, β2 obtained from the scanning angle detection unit 44 are determined. , For example, according to triangulation, color television cameras 10a, 10
Calculate distances such as fire points determined from b.

FIG. 2 is a flow chart showing the fire detection processing by the MPU 40 of FIG. In FIG. 2, first, the presence or absence of a fire is monitored in step S1 (hereinafter “step” is omitted), and when a fire is detected, the process proceeds to S2 and subsequent steps. First, in S2, for example, the color television camera 10
Based on the luminance signal in the image obtained from a, 10b, centering is performed as shown in FIG. 3 (1) so that the fire source is in the center of the screen, and the image is stored in the image memories 46, 48 by the color television cameras 10a, 10b. The captured image of fire source 2 is R
The GB component is stored separately.

Next, in S3, as shown in the first image 62 and the second image 64 of FIG. 3 (2), an image of a predetermined range including the fire source 2 is cut out, and for each pixel included in the cut out image. , For example, the component ratio B / R is obtained. Then, the procedure proceeds to S4, in which the 3 × 3 matrix A is applied to the first cut-out image 62-1 and the nine components (a 0, a
The ratio B / R for each pixel A _ij corresponding to 1, ... A8) is obtained as (μ0, μ1, ... μ8).

The matrix A applied to the first cut-out image 62-1 has a total of nine matrix components a0 to a8 of length × width = 3 × 3, as shown in FIG. The matrix component a4 of the first cutout image 62-
It is applied to the target pixel A _ij indicated by the slanted line which is the processing target in 1. The pixel B corresponding to the matrix components a0 to a8 of the matrix A applied to the cut-out image 62-1 has B
The / R ratio is extracted as (μ0 to μ8).

In FIG. 3, the first cutout image 62-1 is displayed.
The state in which the matrix A is applied to an arbitrary pixel of interest A _ij is shown. The matrix A performs matrix scanning sequentially applied to each pixel as shown in FIG. Figure 4
(A) shows the first cutout image 62-1 and each pixel forming the image is represented by A _ij . Here, i is a pixel number in the row direction and has a value of 0 to n. Further, j is a pixel number in the column direction and has a value of 0 to m. The sizes of the maximum values n and m are appropriately determined according to the size of the cutout image.

The first cutout image 62 shown in FIG.
-1 is applied so that the central component a4 of the matrix A coincides with the pixel at the upper left corner where A _ij = A ₀₀ as shown in FIG. 4 (b). By applying the matrix A to such a pixel A ₀₀ , the corresponding B / R ratio (μ 0 to μ 8
) Is obtained, then the matrix A is moved to the right by 1
And the next pixel A ₁₀ is used as the pixel of interest to apply the matrix A to obtain the B / R ratio. Similarly, the matrix is applied to the last pixel A _nm to obtain the B / R ratio.

Referring again to FIG. 2, next in S5, as shown in FIG.
The same applies to the second cutout image 64-1 shown in (3).
By applying the matrix B of × 3, B / of the pixel B _ij corresponding to the constituent elements (b0, b1, ... b8) of the matrix B
The R ratio is calculated as (ε0, ε1, ... ε8). This matrix B is also composed of a total of nine components b0 to b8 with a length and width of 3 × 3, as shown in FIG. 3 (4). Also, for the second cut-out image 64-1, the matrix B for all the pixels B _{00 to} B _nm is performed in the same manner as shown in FIG.
To determine the B / R ratio.

In this way, the first and second S4 and S5 are performed.
Of the cutout images 62-1 and 64-1
If the B / R ratio is found in the application, in S6
Target pixel A to which lix A is applied_ijThe B / R ratio of
Pixel B to which Trix B is applied _ijSquare error of B / R ratio for each
Find the sum of. That is, by the processing of S4 and S5 shown in FIG.
As shown in (5), all the target pixels A_ijAnd B_ijAgainst
By applying each matrix A and B, the B / R ratio (μ0 to μ8
 ) And (ε0 to ε8) were obtained, (6)
As shown in, the components a0 to a8 and b of the matrices A and B are
Squared errors K0 to K8 are obtained for each 0 to b8.

That is, B in the component a _n of the matrix A
/ R ratio is μ _n , the same matrix component b _{n of} matrix B
Assuming that the B / R ratio at is ε _n , the square error is given by the following equation as a general equation.

[0032]

[Equation 3]

In this embodiment, the matrices A and B are 9 elements of 3 × 3, and since n = 0 to 8, the respective squared errors are obtained by the following equation.

[0034]

[Equation 4]

Then, as shown in FIG. 3 (7), the sum of squared errors e _ij is obtained according to the following equation.

[0036]

[Equation 5]

Referring again to FIG. 2, the matrix A in which the sum of squared errors is applied to the first cut-out image 62-1 in S6.
If all combinations of the matrix B applied to the second cut-out image 64-1 are calculated, the minimum value of the square error e _ij indicating the strongest similarity is obtained in S7. Finally, in S8, the scanning angles α, β1, obtained by the scanning angle detection unit 44 for the color television cameras 10a, 10b at this time are determined.
The central components a4 and b of the matrices A and B from which β2 and the total sum e _{ij of the} squared errors which are the minimum values selected in S7 are obtained.
The fire point is determined using the coordinates of the target pixels A _ij and B _ij corresponding to 4. This is the process shown in FIGS. 3 (7), (8) and (9).

The pixels A _ij and B _ij determined as the fire points in S8 are the central pixels indicated by diagonal lines in the matrices A and B of (3) in FIG. 3, and the cutout images 62-1 and 64-
Although it is in a different position on Fig. 1, it is the same point when viewed from the color television cameras 10a and 10b in the fire source 2 of Fig. 3 (1). FIG. 5 is a plan view showing a detected state of the fire position according to the present invention, in which the first and second color television cameras 10a and 10b are positioned with respect to the spread fire source 2 as shown.

When the fire position detection processing of the present invention is not performed, the fire points when the TV cameras 10a and 10b are directed to the fire source 2 are P and Q, and the fire point is farther from the fire source 2.
Judge the point as a fire point. On the other hand, in the present invention, one point S on the fire source 2 is processed by the processing shown in FIGS.
Is determined as a fire point and the distance to the fire source 2 is recognized as the same S point viewed from two TV cameras, so that the distance can be accurately detected.

To measure the distance to the determined fire point S, the optical axes of the television cameras 10a and 10b are directed to the same determined fire point S on the fire source 2, and the scanning angle α in this state is measured. And the camera installation interval L may be used for the measurement. More specifically, the color television cameras 10a and 10b are set so that the pixels A _ij and B _ij determined to be the fire points shown in FIG. 3C are located at the centers of the screens 62 and 64 of the color television cameras 10a and 10b. By scanning, it is possible to create a state in which the color television cameras 10a and 10b are directed to the fire point S shown in FIG.

FIG. 6 is an explanatory view showing another embodiment of the pixel configuration of an image to be processed according to the present invention. That is, in the above-described embodiment, one of the cutout images 62-1 and 64-1 is used.
The processing is performed by applying the matrices A and B for each pixel, but in the embodiment of FIG. 6, for example, one pixel block BL _ij is constituted by 4 × 4 16 pixels, and this pixel block BL _The matrix is similarly applied to the cutout image constituted by _ij to obtain the sum of squared errors of the component ratio.

Here, the component ratio for the pixel block B _ij constitutes, for example, one pixel block BL _ij , for example, 16
The sum or average value of the individual pixels may be used. By dividing the pixels shown in FIG. 6 into blocks, the number of pixels to be processed can be reduced, and a faster fire position determination process can be performed. Of course, it is desirable to process each pixel in order to improve the position detection accuracy.

[0043]

As described above, according to the present invention, the ratio of two components among the R, G and B components forming a color image is compared with the conventional fire position determined only by the luminance signal. Since the corresponding points on the two screens that are at the same position on the fire source are determined as the fire points based on the similarity calculated from, even if the fire source can be seen as a surface on the screen, the two TV cameras accurately It is possible to specify a fire point, accurately measure the distance to the fire point, and perform more efficient fire extinguishing control.

[Brief description of drawings]

FIG. 1 is a configuration diagram of an embodiment of the present invention.

FIG. 2 is a flowchart showing a detection process of the present invention.

FIG. 3 is an explanatory diagram showing the processing contents of the present invention.

FIG. 4 is an explanatory diagram showing application scanning of a matrix for a cutout image according to the present invention.

FIG. 5 is an explanatory view showing the determination of a fire point according to the present invention in a plan view.

FIG. 6 is an explanatory diagram in which pixels in a cutout image processed by the present invention are divided into blocks.

FIG. 7 is an explanatory diagram of fire source monitoring using two TV cameras in a conventional large space.

FIG. 8 is an explanatory view showing a problem of fire source monitoring by two TV cameras in a conventional small space.

[Explanation of symbols]

10a: First color television camera 10b: second color TV camera 12, 24: Horizontal rotation motor 14, 26: Vertical rotation motor 16, 18, 28, 30: drive circuit 20, 32: Horizontal scanning angle sensor 22, 34: Vertical scan angle sensor 40: Microprocessor (MPU) 42: Scan control unit 44: Scanning angle detector 46, 48: image memory 50: Image cutout section 52: Ratio calculation unit 54: Matrix scanning unit 56: Square error calculator 58: Minimum value selection section 60: Fire point determination unit 62: First image 64: Second image

Continued front page    (72) Inventor Taku Watanabe             2-1043 Kamiosaki, Shinagawa-ku, Tokyo Ho             Chiki Co., Ltd.

Claims (6)

[Claims] 1. A first and a second color television camera for imaging a surveillance area, a scanning mechanism for scanning the first and the second color television camera in horizontal and vertical directions, and the first and the second when a fire occurs. R, G, and B for each pixel of the first and second images captured by the second color television camera
An image memory for extracting and storing components, and R, G of the first and second images stored in the image memory,
A ratio calculation unit that calculates the ratio of two components for each pixel using the B component, and a matrix of a predetermined number (n) of pixels, in which each pixel of the first and second images is a target pixel, in order. A matrix scanning unit that performs a matrix scan to be applied, and a component calculation ratio of a plurality of pixels included in the matrix that scans a first image and the ratio calculation unit that includes a plurality of pixels included in a matrix that scans a second image The square error calculator calculates the sum of the squared errors of the component ratios by, and the square error calculator calculates all combinations of the matrix in which all the pixels of the first image and the second image are applied as the target pixels. A minimum value selecting section for selecting a minimum value in the sum of squared errors of the obtained component ratios; scan angle information of the first and second color television cameras, and a minimum value selected by the minimum value selecting section. give A fire point detection device, comprising: a fire point determination unit that determines a fire point based on a position of a central pixel of a matrix applied to the first and second images. 2. The fire point detection device according to claim 1, wherein the ratio calculation means is a ratio of B component and R component (B / R) or a ratio of G component and R component (G / R). A fire point detection device characterized by calculating. 3. The fire point detection device according to claim 1, wherein the squared error calculator calculates the component ratio of _n pixels included in the matrix applied to the first screen as μ _n , and the second screen as the second screen. The component ratio of n pixels included in the matrix applied to
_{Let n} be the sum of squared errors between the two matrices e
_ij is given by Fire point detection device characterized by calculating as. 4. The fire point detection device according to claim 1, wherein when the fire is detected, the scanning mechanism is controlled to control the first and second fire mechanisms.
The fire point detection device, which stores image data in the image memory after centering the image of the fire source at the center of the screen of the television camera. 5. The fire point detection device according to claim 1, wherein an image in a predetermined range including a fire source is cut out from the images stored in the image memory and processed. apparatus. 6. The fire point detection device according to claim 1, wherein the unit of pixels to be processed on the first screen and the second screen is a block pixel obtained by blocking every predetermined number of pixels. Fire point detection device characterized by being.