CN101574566A - Monocular vision technique based fire monitor control method for adjusting relative positions of fire point and water-drop point - Google Patents

Abstract

The invention relates to a fire monitor control method for adjusting the relative positions of a fire point and a water drop point based on monocular vision technology, and belongs to the technical field of fire monitoring and automatic fire extinguishing. This method solves the problem of using two-dimensional images to realize the relative position adjustment of the fire point and the falling water point in the three-dimensional space by controlling the pitch angle and the horizontal angle multiple times in sequence, so that the fire monitor's falling point can track the position of the fire point in real time. Changes provide conditions for automatic fire extinguishing in large spaces.

Description

Fire monitor control method based on relative position adjustment of fire point and falling water point based on monocular vision technology

technical field

The fire monitor control method for adjusting the relative positions of a fire point and a water drop point based on monocular vision technology of the present invention belongs to the technical field of fire monitoring and automatic fire extinguishing.

Background technique

Fire is one of the most common serious disasters in human society. According to the fire statistics of the United States, Japan and the European Union in recent years, the direct economic loss of fire in developed countries accounts for 0.2% to 0.3% of the gross national product. Fire fatalities in the US are 0.0016%, in the EU 0.0013%, and in Japan 0.0012%, including many major fires. For example, the fire caused by the Great Hanshin Earthquake in Japan in 1995 killed more than 5,200 people. In 1997, a fire broke out in the holy city of Mecca in Saudi Arabia during Ramadan, killing more than 1,000 people. In June 2002, the forest fire in Arizona State, USA was out of control, which shocked the astronauts on the space shuttle. In recent years, with the advancement of science and technology and the development of society, our country has made gratifying achievements in fighting fires. But at the same time, with the acceleration of urbanization and the rapid growth of population, the number of fires in our country and the losses caused by them are on the rise. According to statistics, in 1997, more than 140,000 fires broke out across the country, killing 2,722 people, injuring 4,930 people, and causing direct property losses of 1.54 billion yuan. Among them, there were 88 extraordinarily large fires in which more than 10 people were killed or more than 50 households were affected, or the direct property loss was more than 1 million yuan.

Numerous examples prove that fire is one of the most destructive disaster phenomena in modern civilized society. Therefore, one of the most effective ways to minimize the damage caused by fire is to monitor it in real time and extinguish it before it spreads. Therefore, researches on intelligent fire protection systems have been carried out at home and abroad. However, the current research in this area is mainly the automatic monitoring of fire and is mainly aimed at indoor areas, see patent 200710044936. There are not many studies on the automatic extinguishing of large outdoor fires after they are discovered. In fact, in many outdoor large space environments, such as chemical units, oil tanks and some fire prevention points, there is a considerable demand for real-time fire monitoring and automatic extinguishing. One of the difficulties of an outdoor large-space automatic fire-fighting system is: the precise positioning of the water column trajectory (hereinafter referred to as the water channel) sprayed by the fire monitor, especially the position of the end of the water channel (hereinafter referred to as the water drop point). At present, many processing methods are to first find the fire point, and then use the formula calculation method to obtain the required water column trajectory, and then control the fire monitor to carry out the fire extinguishing work. In a large space environment, the distance between the fire point and the fire monitor is often tens of meters or even more than 100 meters. In the case of long-distance shooting, it is difficult to determine the spatial position of the fire monitor's water drop point, especially it is also easily affected by environmental factors such as wind direction. Point.

Contents of the invention

The purpose of the present invention is to provide a fire monitor control method based on monocular vision technology for adjusting the relative positions of the fire point and the water drop point, so that the water point of the fire monitor can track the change of the fire point position in real time, and provide conditions for automatic fire extinguishing in large spaces.

A method for intelligent fire detection and extinguishment based on computer vision technology, characterized in that it comprises the following steps:

(1) Use the substation control module for fire monitoring:

Control the movement of the pan-tilt, drive the infrared camera to move, and scan its visible area;

Control the video acquisition chip for real-time acquisition and analysis of infrared images, and quickly extract suspected fire points through threshold segmentation;

For suspected fire points, the flame is further judged by the color distribution characteristics, flame change characteristics, flame area spread growth characteristics, flame shape change characteristics, and flame edge change characteristics;

(2) In the case of confirming the existence of the flame, the spatial position of the flame is estimated through the binocular parallax technology;

(3) According to the opening rules of the fire monitor, determine the fire water monitor to be opened;

(4), using computer vision technology or image processing technology to obtain the position of the end of the sprayed water column of the fire monitor in the CCD image, that is, the position of the water drop point;

(5) According to the relative position of the falling water point and the flame on the CCD image, adjust the fire monitor to realize the closed-loop control of the fire monitor to extinguish the fire;

(6) According to the monitoring image, it is judged that the fire extinguishing is completed, and the fire monitor is turned off.

An intelligent fire detection and extinguishing system based on computer vision technology is characterized in that it includes: a server and several sub-stations connected to the server through a bus to form a network. The relative position deviation of the flame adjusts the embedded control module of the fire monitor's water outlet angle, the controllable pan-tilt, and a pair of dual-band CCD cameras installed on the controllable pan-tilt for flame and fire water channel monitoring and positioning.

A method for identifying fire monitor waterways and waterway ends, characterized in that it comprises the following steps:

(1) Starting from the starting point of the water channel, which is the position of the fire monitor, along the direction of the fire monitor's water outlet, scan the pixel columns of the water channel image column by column to find the center point of the water channel track of each pixel column, and the gray value is used to find the center point of the water channel track of each pixel column The method of combining the statistical method and the trajectory prediction method, the specific method is as follows:

Predict the position range of the center point of the next pixel row of water channel track from the determined track of the water channel;

At the same time, find the center point of the water channel track of the pixel column through the gray value statistical method;

When the center point found by the gray value statistical method is within the error range of the center point obtained by the trajectory prediction method, the center point found by the gray value statistical method is used as the center position of the pixel column water channel track;

When the center point found by the gray value statistical method is not within the error range of the center point obtained by the trajectory prediction method, discard the search result and continue to search for the center point of the next pixel column water channel trajectory.

(2) Connect the center points of the water channel trajectories of all the pixel columns into a water channel curve. When the track center points are not found for 10 consecutive columns, it is determined that the last found pixel column water channel track center point is the end position of the water channel.

A fire monitor control method based on the relative position adjustment of the fire point and the water drop point based on monocular vision technology, which is characterized in that it includes the following steps:

(1) When the fire point is found, use the principle of binocular parallax to process the image of the dual-band camera to obtain the depth and distance information of the flame, and then combine the location information of the fire monitor to determine the initial horizontal angle of the fire monitor by using the preset rules and pitch angle, and start the fire monitor to spray water;

(2), using a common CCD camera to obtain real-time two-dimensional images with both flames and water channels;

(3) In the case that the two-dimensional image can only reflect the deviation position relationship between the fire monitor's water drop point and the flame, but cannot clearly reflect the left-right relationship between the water channel's implementation point and the flame, the distance between the fire monitor's water outlet position reflected in the two-dimensional image The far and near position relationship between the falling water point of the water channel and the flame is firstly controlled to make the water falling point close to the fire point by controlling the pitch angle of the fire monitor. The distance difference meets the preset error allowable range;

(4) After the adjustment of the pitch angle is completed, the water channel falling point and the position of the flame can be regarded as equidistant from the starting point of the water channel, and at the same time, the left and right position relationship between the water channel falling point and the flame is clearly reflected in the two-dimensional image; then further Using the left and right positional relationship between the waterway falling point and the flame reflected in the two-dimensional image, continue to control the horizontal angle of the fire monitor to make the falling point closer to the fire point. The angle between the starting point of the water channel and the line connecting the fire point meets the preset error allowable range, and the adjustment of the fire monitor is completed;

(5) With the change of the position of the flame center point, repeatedly adjust the pitch angle and horizontal angle of the fire monitor, so that the falling water point can track and fall on the flame until the flame is extinguished.

The present invention has the following technical effects:

1. A networked fire monitoring and extinguishing system is formed through the server and multiple sub-stations, which greatly increases the area of fire monitoring and extinguishing, and can realize fire monitoring and extinguishing in large outdoor areas;

2. The dual-band CCD camera utilizes binocular parallax to realize the positioning of the flame, which improves the accuracy of the spatial positioning of the flame and provides conditions for rapid fire extinguishing;

3. Fire monitoring and flame extinguishing are integrated in one system, which improves the real-time performance of fire extinguishing and can effectively prevent the spread of flames;

4. Real-time monitoring of the relative position of the fire monitor waterway (falling point) and the flame through computer vision to realize real-time closed-loop control of the fire monitor and improve the accuracy of the water monitor fire extinguishing;

5. Using the combination of grayscale statistics method and trajectory prediction method, starting from the water outlet of the fire monitor, the image is processed by trajectory tracking method to find the water falling point, which overcomes the shortcomings of the usual pattern recognition method that can only recognize objects with relatively fixed shapes . It provides conditions for the closed-loop control of fire monitors required for automatic fire extinguishing of large space fire monitors.

6. The fire monitor control method based on the relative position adjustment of the fire point and the falling water point based on monocular vision technology solves the problem of using two-dimensional images to realize the fire point and falling water point in three-dimensional space by controlling the pitch angle and horizontal angle multiple times in sequence. The problem of adjusting the relative position of the fire monitor enables the fire monitor to track the change of the fire position in real time, providing conditions for automatic fire extinguishing in large spaces.

Description of drawings

Figure 1 The overall flow chart of the intelligent fire detection and extinguishing system.

Figure 2 Schematic diagram of binocular imaging and x-z plane projection. Figure 2(a) is a schematic diagram of binocular imaging, and Figure 2(b) is an x-z plane projection.

Figure 3 is a schematic diagram of the relative positions of the water channel and the fire point.

Figure 4 is a schematic diagram of the water channel and the position of the fire point after the pitch angle control of the water cannon is completed.

Label names in the figure: 1. Simulated fire point; 2. Simulated fire monitor water channel curve; 3. Initial water channel falling point; 4. Straight line distance of water channel; 5. Distance from water channel starting point to fire point; 6. Adjusted pitch angle of fire monitor Waterway curve; 7. The water drop point after the pitch angle of the fire monitor is adjusted; 8. The direction of the horizontal angle adjustment of the fire monitor; 9. The distance from the starting point to the fire point after the pitch angle of the fire monitor is adjusted.

Detailed ways

The two parts of fire monitoring and fire extinguishing belong to the category of fire protection and are related to each other. Therefore, the present invention utilizes computer vision technology to integrate the two into one system to realize closed-loop control of fire fighting and fire extinguishing, and to realize intelligent fire monitoring and fire extinguishing.

The system consists of a computer and several substations, and the computer realizes centralized management and distributed control. The computer and the substation are connected through the bus (or network) to realize information sharing. The number and location of sub-stations are based on the area of the fire monitoring area and the characteristics of the objects (such as equipment, etc.) in the area, as well as the location of the key fire prevention areas that are prone to fire. For key fire prevention areas, sub-stations can also be added as needed. Each substation takes the MPU embedded module as the control center and has the ability of autonomous control. Each substation can be composed of a pair of dual-band CCD cameras, a controllable fire monitor, a controllable pan-tilt and an embedded control module. The dual-band CCD camera pair includes an infrared camera (composed of ordinary CCD camera + infrared filter) and an ordinary CCD camera. Dual-band cameras are mounted on controllable pan/tilts. The MPU-based embedded substation control module (hereinafter referred to as the substation control module) is mainly responsible for controlling two cameras, controlling the PTZ, video acquisition, image analysis, fire monitor control, communication with the server, and executing instructions from the host computer server, etc. Work.

The fire monitoring is mainly completed under the autonomous control of the substation control module. The steps of the substation control module for fire monitoring are: 1. Control the movement of the pan/tilt, and the infrared camera scans its visible area; 2. Control the video acquisition chip to collect and analyze the infrared images in real time, and quickly extract them through the method of threshold segmentation Suspected fire point; 3. Once a suspected fire point is found, further determine the flame through the color distribution characteristics of the flame, flame change characteristics, flame area spread growth characteristics, flame shape change characteristics, and flame edge change characteristics; 4. After confirmation In the presence of flames, start another CCD ordinary camera. The dual-band camera pair uses binocular parallax technology to estimate the spatial position of the flame.

The control module uploads the detected flame alarm information, flame area and spatial location information to the server through the bus, and the server automatically formulates a fire protection plan after analysis. And realize the following operations: 1. Start the linkage telephone alarm; 2. Send the fire information and the location information of the fire point to all sub-stations, so that each sub-station can increase the flame scanning density and frequency; 3. According to the formulated fire protection plan, the fire will be found The sub-stations and the sub-stations controlled by the fire monitors around the flame form a fire-fighting team to realize centralized management and distributed control.

Under the centralized management mode of the server, the fire extinguishing team performs closed-loop control on the water outlet angle of the fire monitor. Each substation presets the water discharge angle of the fire monitor according to the spatial position information of the flame, and starts the fire monitor to spray water. The sub-station analyzes the relative position of the falling water point of each water channel and the flame in the monitoring image, and uses the monocular vision positioning fire water channel tracking flame control method to adjust the control angle of the fire monitor in real time, so as to realize the tracking and extinguishing of the flame by the fire monitor.

The overall flow chart of the system is shown in Figure 1. Each substation realizes real-time fire monitoring. If a fire is found, the server will formulate a fire extinguishing plan and designate a fire extinguishing sub-station. The fire extinguishing sub-station performs closed-loop control on the fire monitors, and realizes the tracking of the flames at the fire monitor waterway drop point, and realizes automatic fire extinguishing. The key technologies are:

1. Fire monitoring and positioning

The fire monitoring mainly controls the pan-tilt to drive the infrared camera to move through the substation control module, and the infrared camera scans its visible area. The substation control module then controls the video acquisition chip to perform real-time acquisition and analysis of infrared images. The method of threshold segmentation is used to quickly extract suspected fire points. Once a suspected fire point is found, further determine whether it is a flame, (see patent 200810124425), the specific steps are: (1) according to the principle that the red component of the flame infrared image is prominent, use a computer system to carry out the infrared image based on the red component Grayscale processing, and take the reference gray value of the flame as the threshold, perform binary threshold segmentation on the image, extract the suspected image and perform filtering processing; (2) use the computer system to further analyze the suspected image to obtain the color distribution characteristics of the flame Criterion, flame image change characteristic criterion, flame area spread growth characteristic criterion, flame image circularity criterion, flame shape change characteristic criterion five criteria; (3) use neural network to judge According to 5 as input, comprehensive judgment is made to obtain the final judgment of whether there is a fire. After confirming that there is a flame, start another ordinary CCD camera. A pair of dual-band cameras consisting of an infrared camera and an ordinary CCD camera estimate the spatial position of the flame through binocular stereo vision ranging technology.

The binocular stereo vision distance measurement technology is to simulate the depth or distance of the object when the creature observes the object with two eyes at the same time, so as to realize the measurement of the depth of the three-dimensional space. According to the depth perception calculation hypothesis, the human depth perception ability is generated by the comparison calculation of disparity, and the calculation of disparity is implemented based on the two-dimensional projection image information of the two retinas obtained by the left and right hemibrains. Stereo vision is based on the hypothesis of depth perception computing, observe the same scene from two or more viewpoints, obtain a set of images under different perspectives, and then infer the target object in the scene through the parallax between corresponding pixels in different images s position.

The schematic diagram of binocular imaging is shown in Figure 2. In Figure 2(a), the projection points of the spatial point P(x _w , y _w , z _w ) on the two images acquired from different positions are P _l and P _r respectively , the line connecting the focus centerlines of the left and right cameras is set as the x-axis, and the projections of P _l and P _r on the x-axis are x _pl and x _pr respectively. The optical axes of the two cameras are parallel and located on the xz plane. In this condition, the cameras are said to be in a parallel alignment state. The z axis is parallel to the optical axis of the two cameras, the focal length of the two cameras is f, and the distance between them is d.

Figure 2(b) is the projection diagram of point P on the xz plane. In the coordinate system shown in the figure, the projection coordinates of the spatial point P(x _w , y _w , z _w ) on the xz plane are (x _w , z _w ), its image coordinates on the left and right images are (x _pl , y _pl ), (x _pr , y _pr ), it can be seen that the relationship is as follows:

${\mathrm{<span\; class="notranslate">\; z</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; df</span>}}{{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; pl</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; pr</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

Then the depth distance of the object from the camera measured by binocular stereo vision is: z _w -f

The projections of the object on the x and y axes are:

${\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; d</span>}\frac{{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; pl</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="pr"\; filenames="A20091003303200151.png"\; state="\{\{state\}\}">pr</figure-callout></span>}}}{\mathrm{<span\; class="notranslate">\; 2</span>}}\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; pl</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; pr</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

${\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; d</span>}\frac{{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; pl</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; the\; <figure-callout\; id="2"\; label="y\; pr"\; filenames="A20091003303200151.png"\; state="\{\{state\}\}">y</figure-callout></span><figure-callout\; id="2"\; label="y\; pr"\; filenames="A20091003303200151.png"\; state="\{\{state\}\}">\; </figure-callout>}}_{\mathrm{<span\; class="notranslate">\; </span>}}}{\mathrm{<span\; class="notranslate">\; 2</span>}}\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; pl</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; pr</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

The spatial location of the fire can be realized by using formulas (1)-(3).

2. Confirmation of fire monitor trajectory and water falling point

The automatic fire extinguishing system of the present invention performs real-time closed-loop control on the fire monitor based on the relative position of the fire monitor's water drop point and the flame. Therefore, the positioning accuracy of the falling water point directly affects the accuracy of fire extinguishing. However, the image of the falling point of the fire monitor obtained by the CCD camera is affected by the background image, and the splashing water is affected by various uncertain factors. Therefore, it is difficult to recognize and judge the falling water point image by the usual image pattern recognition method.

Since the installation position of the fire monitor is fixed, the position of the fire monitor in the image obtained by the CCD camera is known. The invention proposes to start from the starting point of the waterway, that is, the position of the fire monitor, and use the waterway tracking method to identify the water falling point. Then the determination of the trajectory of the fire monitor is essentially a problem of finding the trajectory of the waterway and the end point of the trajectory, that is, the water drop point, based on the starting point of the waterway.

Similarly, the water channel image is affected by factors such as wind direction, fire monitor spray angle, water monitor pressure, etc., and the shape of the water channel image also has great uncertainty. At the same time, the gray value of the water channel will be affected by the light and shooting angle, and the change is very large. Therefore, it is also difficult to obtain waterways by simply using threshold segmentation. However, it can be seen visually from the obtained CCD image that there is a clear boundary between the waterway part and the background image.

The water channel tracking method starts from the position of the fire monitor and scans the image pixel column by column along the water outlet direction to find the center point of the water channel trajectory of each pixel column. Then connect the water channel center points of all columns to form a water channel curve, and determine the position of the water channel falling point with the end position of the water channel curve. Therefore, the key to waterway recognition lies in the search for waterway images in each pixel column of the image. The present invention uses gray value statistical method and trajectory prediction method to make judgment in water channel search. Firstly, the position of the water channel in the next column is found by using the determined water channel trajectory through the method of gray value statistics, and then the predicted position of the water channel in the column is predicted by the determined water channel trajectory. If the predicted position of the row of waterways and the searched position are within a certain range of error, then the row of waterways is determined; otherwise, the position of the row of waterways is discarded, and the track position of the next row of waterways is searched in the same way. If the waterway cannot be determined for 10 consecutive columns, the waterway track is considered to be terminated.

Assuming that i track points of the water channel have been determined, the pixel coordinates of the center point of the track are (x ₁ , y ₁ )～( _xi , y _i ), the gray level of the center point of the i-th pixel column is G _i , and the width of the water channel is k _i . Then the method of finding the center point of the next trajectory by the gray-level statistical method and the trajectory prediction method is as follows:

1. Grayscale statistical method

Scan the i+1th column according to the information of the track center point of the ith column, and read the coordinates of the pixel point as (x _i +1, y _i -k _i /2) ~ (x _i +1, y _i +k _i /2) are G(x _i +1, y _i -k _i /2), G(x _i +1, y _i -k _i /2+1), ...G(x _i +1 , y _i +k _i /2).

Find the maximum value of the gray value of the i+1th column:

Max(G(x _i +1, y _i -k _i /2), G(x _i +1, y _i -k _i /2+1),..., G(x _i +1, y _i +k _i /2-1), G(x _i +1, y _i +k _i /2)), and the coordinates corresponding to the maximum point are (x _i+1 , y′ _i+1 ). In order to determine the width of the channel on column i+1, take (x _i+1 , y′ _i+1 ) as the center, and move in two directions, namely (x _i+1 , y _i+1 ) to (x _{i+ 1} , y _i+1 -k _i /2-10) and (x _i+1 , y _i+1 ) or (x _i+1 , y _i+1 +k _i /2+10) to find gray-scale mutation points , the boundary of the waterway. Since the grayscale of the water channel is gradual, the grayscale in the two directions from (x _i+1 , y _i+1 ) to y on the i+1th column is decreasing, and the grayscale value when reaching the boundary of the waterway There will be a sudden change, which is the waterway boundary.

Take the positive direction of y as an example, that is, 3m1, m1∈(y _i+1 , y _i+1 +k _i +10], m1∈Z, from (x _i+1 , y _i+1 +1)～(x _i+1 , y _m1-1 ) have judged non-boundary points one by one, when y _m1 satisfies:

G(x _i+1 , y _m )-G(x _i+1 , y _m1-1 )＞0, |G(x _i+1 , y _m1 )-G(x _i+1 , y _m1-1 ) |＞|G(x _i+1 , y _m1-1 )-G(x _i+1 , y _m1-2 )|or G(x _i+1 , y _m )-G(x _i+1 , y _{m1 -1} )＜0，|G(x _i+1 ，y _m1 )-G(x _i+1 ，y _m1-1 )|＞3|G(x _i+1 ，y _m1-1 )-G(x _i+1 , y _m1-2 )| Then determine (xi ₊₁ , y _m1 ) as the upper boundary of the i+1 pixel column, and similarly determine the lower boundary as (xi ₊₁ , y _m2 ). Then the width on the i+1th column is m2-m1.

2. Trajectory prediction method

i waterway trajectory points have been determined, and the pixel coordinates of the trajectory center points are (x ₁ , y ₁ )～( _xi , y _i ), respectively. Using this i point, the curve fitting of the waterway is carried out by the least square method, and x _{i +1} is brought into the fitting curve to obtain the predicted value y _i+ 1 of the channel trajectory of the i+1th point.

3. Control of fire monitor based on monocular image

When the system detects a fire and activates the fire monitor, whether the trajectory of the water channel sprayed by the water monitor can hit the fire point is crucial to the efficiency of fire extinguishing. The present invention adjusts the fire monitor according to the relative position of the flame obtained from the image and the water drop point until the two overlap.

When the fire point is found, according to the image of the dual-band camera and the depth and distance information of the flame obtained by using the principle of binocular parallax, the initial parameters of the fire monitor are set to spray. The image acquired by ordinary CCD camera has the image of flame and waterway at the same time. However, the images obtained by ordinary CCD cameras are two-dimensional images, but what it reflects is three-dimensional space. The definite relative positions of water fall point and flame in three-dimensional space cannot be clearly judged from two-dimensional images. Judging from the position of the water channel implementation point and the flame reflected in the image, the control of the pitch angle of the fire monitor is clear. However, the control of its horizontal angle is difficult to determine, as shown in Figure 3.

Therefore, the present invention proposes to complete the control of the fire monitor in two steps, first to complete the control of the pitch angle, and then to complete the control of the horizontal angle. As shown in Figure 3, the adjustment of the pitch angle is realized according to 4. The linear distance of the water channel and 5. The distance from the starting point of the water channel to the fire point. Set the allowable error range δ _d , when d ₁ -d ₂ >0 and |d ₁ -d ₂ |＞δ _d , then the pitch angle control will go up, when d ₁ -d ₂ <0 and |d ₁ -d ₂ | ＞δ _d , the pitch angle is controlled downward.

After the adjustment of the pitch angle is completed, the falling water point, the position of the flame, and the installation point of the water monitor can be regarded as equidistant, that is, the distance d ₂ from the starting point of the water channel to the fire point in Figure 4 and 9. After the pitch angle of the fire bubble is adjusted The distance d ₁ ′ from the starting point to the fire point is approximately equal. The relative positions of the two in the horizontal direction are clarified from the two-dimensional picture, as shown in FIG. 4 . Control the horizontal direction of the fire monitor according to the size of the θ angle in Figure 4. Set the allowable range of angle error δ _θ , when θ>0, |θ|>δ _θ , the horizontal angle is controlled counterclockwise; when θ<0, |θ|>δ _θ , the horizontal angle is controlled clockwise.

Claims (1)

1, a kind of based on the fire point of monocular vision technique and the fire monitor control method of water-drop point relative position adjustment, it is characterized in that may further comprise the steps:

(1) after finding the fire point, utilize the binocular parallax principle that the image of two waveband video camera is handled, obtain the depth distance information of flame, again in conjunction with fire monitor installation position information, utilize preset rules to determine the fire monitor water outlet initial water straight angle and the angle of pitch, and start the fire monitor water spray;

(2) utilize the common CCD camera to obtain the two dimensional image that has flame and water channel simultaneously concurrently in real time;

(3) on two dimensional image, can only reflect the differential location relation of fire monitor water channel water-drop point and flame, and can not reflect clearly that water channel implements under the situation about concerning about point and flame, concern according to far and near position, the fire monitor water outlet position of two dimensional image reflection apart from water channel water-drop point and flame, at first make water-drop point connect the periareon by the control fire monitor angle of pitch, concrete mode is: adjust the fire monitor angle of pitch and make the water channel starting point satisfy default error allowed band to water-drop point air line distance and water channel starting point to fire point air line distance difference;

(4) after angle of pitch adjustment is finished, the relative water channel starting point with flame location of water channel water-drop point can be considered equidistant, and the while also makes the position, the left and right sides of water channel water-drop point and flame close to tie up in the two dimensional image and clearly reflects; Then further utilize position, the left and right sides relation of the water channel water-drop point and the flame of two dimensional image reflection, continuation makes water-drop point further connect the periareon by control fire monitor horizontal angle, and concrete mode is: adjust the fire monitor horizontal angle and make the water channel starting point satisfy default error allowed band to the angle of water-drop point line and water channel starting point and fire point line.Then the fire monitor adjustment is finished;

(5) with the change of flame kernel point position, repeat to adjust the angle of pitch and the horizontal angle of fire monitor, water-drop point can be followed the tracks of drop on the flame, put out until flame.

Images