Intelligent fire-fighting fire smoke particle detection method

Abstract

The invention relates to the technical field of intelligent sensors, in particular to an intelligent fire-fighting fire smoke particle detection method, which comprises the following steps: acquiring multiple frames of target images of a target area, and determining whether the smoke area in each frame of target image is a fire smoke area according to the gray value of the smoke area in each frame of target image; determining the actual brightness of each sub-region in the fire smoke region under the condition that the fire smoke region is the fire smoke region; determining the global concentration of the fire smoke area according to the original brightness, the actual brightness of each sub-area and the number of pixels, and determining the instantaneous acceleration of the smoke concentration based on the global concentration of the fire smoke area in the target image of the current frame and the global concentration of the fire smoke area in the target image of the previous frame; and determining a fire risk grade by using the instantaneous speed increase of the smoke concentration and triggering an alarm of the corresponding grade. Therefore, the accuracy and reliability of the detection result of the fire smoke particles are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent sensor, in particular to a fire smoke particle detection method for intelligent fire fighting. BACKGROUND

[0002] Fire, as a destructive disaster, threatens people's life and property safety at any time. According to the statistical data released by the State Fire Rescue Bureau, in recent years, fire accidents have occurred frequently in China, causing significant casualties and economic losses. As the core physical feature of early warning of fire, the timeliness and accuracy of the detection of fire smoke play a decisive role in the effectiveness of the entire fire warning system. In some scenarios, the smoke particles of fire are detected based on a single sensing principle. For example, an ionic smoke sensor detects the influence of smoke particles on ion current to determine fire, and a photoelectric smoke sensor triggers an alarm by using the scattering or shielding of light by smoke. However, such technology has serious defects of high false alarm rate and insufficient reliability in complex environments, often leading to sensor misjudgment and unnecessary alarms. Therefore, the above-mentioned method for detecting fire smoke particles may result in low accuracy and reliability of the detection results of fire smoke particles. SUMMARY

[0003] In order to solve the technical problem of low accuracy and reliability of the detection results of fire smoke particles, the purpose of the present application is to provide a fire smoke particle detection method for intelligent fire fighting, and the technical solution is as follows: The present application provides a fire smoke particle detection method for intelligent fire fighting, comprising: acquiring multiple target images of a target area, determining whether the smoke area in each target image is a fire smoke area according to the gray value of the smoke area in each target image; in the case of a fire smoke area, determining the actual brightness of each sub-area in the fire smoke area according to the original brightness of the light source center in the fire smoke area without smoke, the brightness attenuation coefficient and the distance, the distance being the distance between the position of each sampling point arranged along the optical axis of the light source center and the light source center, and the sub-area being the sub-area formed by dividing the fire smoke area by each sampling point arranged along the optical axis of the light source center; determining the global concentration of the fire smoke area according to the original brightness, the actual brightness and the pixel number of each sub-area, and determining the instantaneous increase rate of the smoke concentration based on the global concentration of the fire smoke area in the current frame of target image and the global concentration of the fire smoke area in the last frame of target image; determining the fire risk level by using the instantaneous increase rate of the smoke concentration and triggering the alarm of the corresponding level.

[0004] Optionally, the acquiring the multiple target images of the target region comprises: comparing the image of the target region captured by the camera with the background image, and calculating the pixel-level difference between the current frame of the image and the background image by image difference method; if the pixel-level difference exceeds a first threshold, determining that the target region is suspected to have fire; and performing high-frame shooting on the target region by the camera to obtain the target image, and the frame rate of the high-frame shooting is a ratio between the smoke diffusion speed and the maximum allowed displacement amount of a single frame.

[0005] Optionally, the determining whether the smoke region in each target image is a fire smoke region according to the gray value of the smoke region in each target image comprises: segmenting the smoke region from each target image based on the gray difference between the target image and the background image; determining an area change trend according to the area of the core region of the smoke region of the adjacent target images, and determining a gray change trend according to the average gray of the core region of the smoke region of the adjacent target images, the core region being a high-gray-value core area where smoke particles are concentrated in the smoke region; in a case where the area change trend is greater than a second threshold and the gray change trend is greater than the second threshold, determining that the smoke region is a suspected fire smoke region; in a case where the smoke region is the suspected fire smoke region, screening feature pixel points rich in texture from the suspected fire smoke region; determining an included angle between a motion vector of the feature pixel point and a unit vector in the vertical direction according to a horizontal component and a vertical component of the feature pixel point, the motion vector being determined by comparing the position change of the same feature pixel point in the adjacent target images; in a case where the proportion of the feature pixel points with the included angle less than a third threshold is greater than or equal to a fourth threshold, determining that the suspected fire smoke region is a fire smoke region.

[0006] Optionally, the determining the area change trend according to the area of the core region of the smoke region of the adjacent target images comprises: calculating a first difference value between the area of the core region of the smoke region of the target image of the next frame and the area of the core region of the smoke region of the target image of the current frame; and determining a first ratio between the first difference value and the area of the core region of the smoke region of the target image of the current frame as the area change trend of the core region of the smoke region of the target image of the current frame.

[0007] Optionally, determining the included angle between the motion vector of the feature pixel and the unit vector of the vertical direction according to the horizontal component and the vertical component of the feature pixel comprises: calculating the square root of the sum of the square of the horizontal component and the square of the vertical component, and calculating a third ratio between the vertical component and the square root of the sum; processing the third ratio by using an inverse cosine function to obtain the included angle.

[0008] Optionally, determining the actual brightness of each sub-region in the fire smoke region according to the original brightness of the light source center in the smoke-free state, the brightness attenuation coefficient and the distance in the fire smoke region comprises: calculating a first product between the brightness attenuation coefficient and the absolute value of the distance; determining a third difference between the original brightness and the first product as the actual brightness.

[0009] Optionally, determining the global concentration of the fire smoke region according to the original brightness, the actual brightness of each sub-region and the pixel number comprises: determining the smoke concentration of each sub-region according to the actual brightness of each sub-region in the fire smoke region, the calibration constant and the original brightness; calculating a second product between the smoke concentration of each sub-region and the corresponding pixel number, and superimposing each second product to obtain a first superimposed value; superimposing the pixel numbers of each sub-region to obtain a second superimposed value; determining a fourth ratio between the first superimposed value and the second superimposed value as the global concentration of the fire smoke region.

[0010] Optionally, determining the smoke concentration of each sub-region according to the actual brightness of each sub-region in the fire smoke region, the calibration constant and the original brightness comprises: calculating the average brightness of the actual brightness of each sub-region in the fire smoke region, and calculating a fifth ratio between the average brightness and the original brightness; calculating a fourth difference between the predetermined value and the fifth ratio; determining a third product between the calibration constant and the fourth difference as the smoke concentration.

[0011] Optionally, determining the instantaneous speed-up of the smoke concentration based on the global concentration of the fire smoke region in the target image of the current frame and the global concentration of the fire smoke region in the target image of the previous frame comprises: calculating a fifth difference between the global concentration of the fire smoke region in the target image of the current frame and the global concentration of the fire smoke region in the target image of the previous frame, and the time interval between the target image of the current frame and the target image of the previous frame; determining a sixth ratio between the fifth difference and the time interval as the instantaneous speed-up.

[0012] Optionally, the step of determining the fire risk level and triggering the alarm of the corresponding level according to the instantaneous increase rate of the smoke concentration comprises: determining the fire risk level as primary and triggering the primary alarm when the instantaneous increase rate of the smoke concentration is greater than the fifth threshold value and less than the sixth threshold value, and the primary alarm informs the staff to check; determining the fire risk level as intermediate and triggering the intermediate alarm when the instantaneous increase rate of the smoke concentration is greater than or equal to the sixth threshold value and less than the seventh threshold value, and the intermediate alarm closes the ventilation system; and determining the fire risk level as high and triggering the high alarm when the instantaneous increase rate of the smoke concentration is greater than or equal to the seventh threshold value, and the high alarm starts the spraying.

[0013] The present application has the following advantages: first, a plurality of target images of the target area are acquired, and whether the smoke area in each target image is a fire smoke area is determined according to the gray value of the smoke area in each target image; then, in the case of a fire smoke area, the actual brightness of each sub-area in the fire smoke area is determined according to the original brightness of the light source center in the fire smoke area without smoke, the brightness attenuation coefficient, and the distance between the position of each sampling point arranged along the optical axis of the light source center and the light source center, the sub-area being formed by dividing the fire smoke area by each sampling point arranged along the optical axis of the light source center; second, the global concentration of the fire smoke area is determined according to the original brightness, the actual brightness, and the pixel number of each sub-area, and the instantaneous increase rate of the smoke concentration is determined based on the global concentration of the fire smoke area in the current frame of the target image and the global concentration of the fire smoke area in the last frame of the target image; and finally, the fire risk level is determined according to the instantaneous increase rate of the smoke concentration and the alarm of the corresponding level is triggered.

[0014] In this way, the present application can capture the target area by a non-contact fire smoke particle detection method, distinguish the fire smoke situation in the target area according to the difference in the gray value and the difference in the brightness information of the captured image of the target area, and construct the fire risk level of the fire smoke and trigger the alarm of the corresponding level in combination with the plurality of captured images. In this way, compared with the single sensor, the method provided by the present application is not affected by the complex environment, and the accuracy and reliability of the detection result of the fire smoke particle are improved. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, and the advantages thereof, the drawings needed to be used in the embodiments or prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0016] Figure 1A flow chart of a fire smoke particle detection method for intelligent fire protection provided by an embodiment of the present invention; Figure 2 A schematic diagram of the layout of an image capture system provided by an embodiment of the present invention; Figure 3 A schematic diagram of a change trend of a core area provided by an embodiment of the present invention; Figure 4 A schematic diagram of the distribution of sub-areas of a fire smoke area provided by an embodiment of the present invention; Figure 5 A schematic diagram of an attenuation model provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0017] To further illustrate the technical means and effectiveness of the present invention in achieving its intended objectives, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementation, structure, features, and effectiveness of a fire smoke particle detection method for intelligent firefighting proposed by the present invention. In the following description, references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0018] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0019] The specific scheme of the fire smoke particle detection method for intelligent fire protection provided by the present invention is described in detail below with reference to the accompanying drawings.

[0020] Example 1: See also Figure 1 , which shows a flow chart of a fire smoke particle detection method for intelligent fire protection provided by one embodiment of the present invention, including: Step S101 : acquiring multiple frames of target images of a target area, and determining whether the smoke area in each frame of the target image is a fire smoke area according to the grayscale value of the smoke area in each frame of the target image.

[0021] Specifically, the embodiment of the present invention realizes data collection under stable lighting conditions around the clock by deploying LED constant light sources and combining them with global monitoring. When suspected fire smoke (such as smoke characteristics generated in the early stages of a fire) is detected in the target area, the high frame rate shooting mode is automatically triggered to provide clear and continuous image data support for subsequent fire analysis. The target area can be a warehouse, etc. For example, Figure 2 As shown, Figure 2 A schematic diagram of the layout of an image capture system provided by an embodiment of the present invention is provided. Figure 2In the embodiment, the LED constant light source is arranged on the top of the target area, and the shooting range thereof covers the whole target area, so that images of any area can be shot in all directions. In the system startup, initialization is firstly performed, and dozens of static photos containing goods are shot in the period when there is no personnel and goods in the target area (such as at night), and an "average background image" is generated. The average background image records the brightness, fixed facility contour and texture of the normal state of the warehouse, and is used as a reference for subsequent detection. After the initialization is completed, the real-time monitoring stage is entered, in which the camera continuously shoots videos, and each frame is compared with the background image. The pixel level difference between the current frame and the background image is calculated by the image difference method, and the dynamic change in the target area is detected. Since the smoke area blocks light, the gray value in the current frame is lower than that in the background image. If the local brightness of a frame image decreases, the frame image is marked as a suspicious area of fire smoke, and high frame shooting is started to record the behavior of the smoke.

[0022] Further, as an optional embodiment of the present application, acquiring the plurality of target images of the target area comprises: comparing the image of the target area collected by the camera with the background image, calculating the pixel level difference between the current frame and the background image by the image difference method; if the pixel level difference exceeds the first threshold value, determining that the target area is suspected to have a fire; and shooting the target area by the camera at a high frame rate to obtain the target image, the frame rate of the high frame shooting being a ratio between the smoke diffusion speed and the maximum allowed displacement amount of a single frame.

[0023] Specifically, in the embodiment, the pixel level difference between the current frame and the background image is calculated by the image difference method, and the dynamic change in the target area is detected. Since the smoke area blocks light, the gray value in the current frame is lower than that in the background image. If the local brightness of a frame image decreases, the frame image is marked as a suspicious area of fire smoke, and high frame shooting is started to record the behavior of the smoke. The first threshold value can be determined according to actual conditions, and the embodiment is not limited thereto. The pixel level difference between different images can be determined by the image difference method according to the known technology, and the embodiment is not described herein.

[0024] Further, when the frame rate of the high frame shooting is determined, the following formula can be used for calculation: In the formula, represents the frame rate of the high frame shooting. represents the smoke diffusion speed, and the value range is usually 0.5-2 m / s. represents the maximum allowed displacement amount of a single frame, which is used to ensure the continuity of the smoke movement track, and the value range is recommended to be 0.01-0.03 m.

[0025] Further, for the multi-frame target image collected by high frame rate sequence image acquisition, in order to realize accurate discrimination of the fire smoke area, first, the accurate segmentation of the smoke area in the single-frame target image is completed by analyzing the gray difference between the smoke area and the background, such as using a neural network to realize the accurate segmentation of the smoke area, to provide basic visual information for subsequent fire identification. Among them, in the warehouse fixed light source scene, the smoke covered area presents a translucent characteristic due to the decrease of light transmittance, and its gray value is between the completely blocked shadow area and the light source direct area, resulting in that the gray value is concentrated in the medium gray interval, showing single-peak or double-peak (split into main peak and diffusion peak when the concentration is uneven) distribution. Based on this rule, the initial threshold interval is manually set by analyzing the histogram peak and valley positions, and the threshold boundary is optimized by combining the maximum inter-class variance algorithm to improve the segmentation accuracy, and then the 5*5 elliptical kernel closed operation is used to smooth the burrs or broken parts of the smoke cluster edge according to the cluster diffusion characteristics of the smoke.

[0026] Further, in order to ensure the accuracy and reliability of the determination of the fire smoke area, the embodiment of the present application also analyzes the time domain characteristics of the high frame rate image sequence, captures the diffusion behavior of the smoke in the space-time domain, combines the unique motion performance of the fire smoke, and adopts a mixed judgment strategy to analyze the change of the area (the concentration and motion law of the smoke) in the rising process of the smoke and the delay motion direction, so as to realize the fine characterization of the diffusion behavior of the fire smoke. The initial stage of the fire smoke presents a significant concentration gradient characteristic: the central area forms a high gray value core area due to high particle density, and the surrounding area presents a decreasing distribution with the diffusion process. With the continuous burning, the smoke area undergoes space-time evolution: the core area continuously expands, the gray value of the original core area significantly increases, and a new high concentration center is formed. The overall diffusion area maintains the "center-edge" decreasing mode, but the coverage range is larger than that in the initial stage, and the global maximum gray value is improved, presenting the "core brightening-edge expansion" fire smoke diffusion characteristic. The motion trajectory of the smoke follows the coordinated regulation of the environmental airflow and the thermodynamic mechanism, and the initial stage of the smoke motion presents a vertical lifting motion driven by buoyancy. Even if it encounters obstacles to produce local turbulence, the macro motion vector still maintains consistency with the direction of the main flow field.

[0027] Further, based on the dynamic diffusion characteristics of the fire smoke, the determination of the real fire smoke area is realized by gradient evolution analysis of the multi-frame target image. First, the gradient of the smoke area in each frame of the target image is calculated to identify the core area with high concentration. The core area is the connected area with the maximum gradient amplitude, and the edge area has a decreasing gradient amplitude with distance. The core area of the continuous multiple frames is dynamically analyzed, and the area and gray value of the core area should increase with time. For example, as shown in Figure 3 , it is a change trend diagram of the core area provided by the embodiment of the present application, Figure 3 ​Figure 3 In the embodiment, the smoke concentration and the area of the core region at the t+1 moment are higher than the smoke concentration of the core region at the t moment, which indicates that the fire at this moment is in the form of gradual expansion.

[0028] Further, as an optional embodiment of the present application, the method for determining whether the smoke region in each target image is a fire smoke region according to the gray value of the smoke region in each target image comprises: segmenting the smoke region from each target image based on the gray difference between each target image and the background image; determining the area change trend according to the area of the core region of the smoke region of the target image of the adjacent frame, and determining the gray change trend according to the average gray of the core region of the smoke region of the target image of the adjacent frame, the core region being a high gray value core region where smoke particles are concentrated in the smoke region; in the case that the area change trend is greater than the second threshold value and the gray change trend is greater than the second threshold value, determining that the smoke region is a suspected fire smoke region; in the case of being the suspected fire smoke region, screening feature pixel points with rich texture from the suspected fire smoke region; determining the included angle between the motion vector of the feature pixel point and the unit vector in the vertical direction according to the horizontal component and the vertical component of the feature pixel point, the motion vector being determined by comparing the position change of the same feature pixel point in the adjacent target images; in the case that the proportion of the feature pixel points with the included angle less than the third threshold value is greater than or equal to the fourth threshold value, determining that the suspected fire smoke region is a fire smoke region.

[0029] Specifically, the area change trend is calculated by the following formula in the embodiment of the present application: In the above formula, represents the area change trend of the area of the core region from the t moment to the t+1 moment. t represents the area of the core region of the t frame. represents the area of the core region of the t+1 frame.

[0030] Further, the gray change trend is calculated by the following formula in the embodiment of the present application: In the above formula, represents the gray change trend of the gray of the core region from the t moment to the t+1 moment. represents the average gray of the core region of the t+1 frame. μt represents the average gray of the core region of the t frame.

[0031] Further, the second threshold value can be 0 in the embodiment of the present application, and the smoke of the core region detected satisfies and , the smoke region is marked as a suspected fire smoke region.

[0032] Further, the embodiment of the present application extracts key points in the suspected smoke area of each target image, adopts a corner detection algorithm, screens N pixel points with rich textures as tracking targets, i.e., feature pixel points (such as corners of smoke edges), uses a Lucas-Kanade optical flow algorithm to track displacement of the feature points (feature pixel points are referred to as feature points) frame by frame. By comparing position changes of the same feature point in adjacent two target images, motion vectors in horizontal and vertical directions of the feature point are calculated, and the motion vector of the feature point is synthesized In this process, feature points that cannot be matched due to occlusion or motion blur are removed, and only points that are continuously successfully tracked are retained for subsequent analysis. By calculating motion vectors of the N feature points, combining the angle between the motion vector and the unit vector in the vertical direction, and judging whether the motion direction of the feature point conforms to the motion characteristics of fire smoke, the motion vector of the feature point i is The unit vector in the vertical direction can be expressed as Then, the angle θ between the motion vector and the unit vector can be derived from the vector dot product formula.

[0033] Further, as an optional embodiment of the present application, determining the angle between the motion vector of the feature pixel point and the unit vector in the vertical direction according to the horizontal component and the vertical component of the feature pixel point includes: calculating the square root of the sum of the square of the horizontal component and the square of the vertical component, and calculating a third ratio between the vertical component and the square root of the sum; and using an inverse sine function to process the third ratio to obtain the angle.

[0034] Specifically, the dot product of the motion vector and the unit vector in the embodiment of the present application is defined as: .

[0035] Substitute the motion vector and the vertical unit vector in the above formula: Calculate the dot product: Calculate the modulus: Further, substitute the dot product and the modulus into the above formula: Obtain .

[0036] In the above formula, x represents the displacement of the feature point i in the horizontal direction, i.e., the horizontal component, and y represents the displacement of the feature point i in the vertical direction, i.e., the vertical component.

[0037] Further, since the fire smoke is mainly driven by buoyancy in the warehouse, it rises vertically and has high consistency in movement. Therefore, compared with non-fire smoke, such as dust and water vapor, the movement of the fire smoke is more random and the direction is dispersed, and the proportion of the vertical direction is low.

[0038] Further, in the embodiment of the present application, the third threshold value can be 20°, and the fourth threshold value can be 70%. When the angle θ between the motion direction of the feature points in the suspected smoke area in two consecutive target images and the vertical direction is less than 20°, and the proportion of the feature points satisfying the angle is greater than or equal to 70%, the suspected smoke area is determined as a fire smoke area.

[0039] In step S102, in the case of a fire smoke area, the actual brightness of each sub-area in the fire smoke area is determined according to the original brightness of the light source center in the smoke-free state, the brightness attenuation coefficient, and the distance.

[0040] The distance is the distance between the position of each sampling point arranged along the optical axis of the light source center and the light source center, and the sub-area is formed by dividing the fire smoke area by each sampling point arranged along the optical axis of the light source center. For example, as shown in FIG. 4, Figure 4 Figure 4 is a distribution diagram of a sub-area of a fire smoke area provided by an embodiment of the present application, Figure 4 In the embodiment, the sampling points are uniformly arranged perpendicular to the central axis with the transverse maximum value of the fire smoke area as the boundary, and then the sampling points are connected with the light source. The area formed on the fire smoke area by the lines connecting all the sampling points with the light source and the lines between every two adjacent lines of the central axis is the sub-area. It is worth noting that sampling points can also be arranged at other positions, which are not limited in the embodiment.

[0041] Further, in the fixed light source scene such as a warehouse, the quantification of smoke concentration can be realized by analyzing the natural attenuation law of the light source. The specific method is as follows: first, the distribution characteristics of the light source center (optical axis) are calibrated by using the smoke-free background image, that is, the light source center has the highest brightness, and the brightness decreases in a conical shape to the surrounding (the farther from the center, the darker). According to the segmented smoke area, equidistant sampling is performed along the light source central axis in both left and right directions, and the gray value of each sampling point is recorded. Since the smoke will block the light (the thicker the smoke, the less the light will penetrate), the light intensity and the propagation distance are inversely proportional, and along the light source central axis, the brightness decreases linearly with the horizontal distance d. The smoke concentration of each sampling point can be deduced, and the lower the measured brightness of the light after penetrating the smoke, the thicker the smoke at that position.

[0042] ​Further, in the embodiment of the present application, first, the luminance linearly decays with the horizontal distance d along the optical axis of the light source, and the luminance decay coefficient can be determined by linear fitting, and as an optional embodiment of the present application, according to the original luminance of the light source center in the fire smoke area without smoke, the luminance decay coefficient and the distance, the actual luminance of each sub-region in the fire smoke area is determined, including: calculating a first product between the absolute value of the luminance decay coefficient and the distance; determining a third difference between the original luminance and the first product as the actual luminance.

[0043] Specifically, the actual luminance is calculated by the following formula in the embodiment of the present application: In the above formula, represents the actual luminance of the sub-region divided when the light propagates to the distance d. represents the original luminance of the light source center without smoke. k represents the luminance decay coefficient of the light source, and in the warehouse scene where the light source is fixed, k usually takes a value of 1-3. represents the distance between the position of each sampling point arranged along the optical axis of the light source center and the light source center. Wherein, represents the horizontal coordinate of the sampling point, represents the horizontal coordinate of the light source center.

[0044] Step S103, determining the global concentration of the fire smoke area according to the original luminance, the actual luminance of each sub-region and the pixel number, and determining the instantaneous acceleration of the smoke concentration based on the global concentration of the fire smoke area in the target image of the current frame and the global concentration of the fire smoke area in the target image of the last frame.

[0045] Specifically, the sampling points at each position divide the fire smoke area along the central axis into i sub-regions, and the average luminance of all sub-regions is obtained by averaging the actual luminance of each sub-region The smoke concentration C of each sub-region i is proportional to the luminance decay coefficient.

[0046] Further, as an optional embodiment of the present application, the smoke concentration of each sub-region is determined according to the actual luminance of each sub-region in the fire smoke area, the calibration constant and the original luminance, including: calculating the average luminance of the actual luminance of each sub-region in the fire smoke area, and calculating the fifth ratio between the average luminance and the original luminance; calculating the fourth difference between the predetermined value and the fifth ratio; determining the third product between the calibration constant and the fourth difference as the smoke concentration.

[0047] Specifically, in the embodiment of the present application, the predetermined value is 1, and the smoke concentration is calculated by the following formula: In the above formula, represents the smoke concentration of the i-th sub-region. K represents a calibration constant, which is obtained by the ratio of the measured brightness in the absence of smoke to the theoretical brightness of the attenuation model. represents the actual brightness of the i-th sub-region, i.e. the original brightness. represents the average brightness of all sub-regions.

[0048] As shown in the example, Figure 5 Figure 5 is a schematic diagram of an attenuation model provided by an embodiment of the present application, in which the light intensity is highest along the central axis, and the light intensity decreases horizontally along the central axis. In each sub-region, the theoretical brightness of each sub-region is determined according to this decreasing rule.

[0049] Further, an embodiment of the present application specifically calculates the global concentration of the fire smoke region by the following formula: In the above formula, represents the global concentration of the fire smoke region. represents the number of sub-regions in the fire smoke region. represents the number of pixels in the i-th sub-region. represents the smoke concentration of the i-th sub-region.

[0050] In step S104, the instantaneous rate of increase of the smoke concentration is used to determine the fire risk level and trigger the alarm of the corresponding level.

[0051] Specifically, after obtaining the quantitative index of the fire smoke in an embodiment of the present application, the smoke concentration of each frame in the multiple frames of target images is quantitatively analyzed, the fire risk level of the warehouse is classified based on the instantaneous rate of increase of the fire smoke concentration between the multiple frames of target images, and the manual intervention reminder is triggered in time.

[0052] Further, as an optional embodiment of the present application, the instantaneous rate of increase of the smoke concentration is determined based on the global concentration of the fire smoke region in the current frame of target image and the global concentration of the fire smoke region in the previous frame of target image, which includes: calculating a fifth difference value between the global concentration of the fire smoke region in the current frame of target image and the global concentration of the fire smoke region in the previous frame of target image, and a time interval between the current frame of target image and the previous frame of target image; and determining a sixth ratio between the fifth difference value and the time interval as the instantaneous rate of increase.

[0053] Specifically, in an embodiment of the present application, the smoke concentration of the current t-th frame of target image is first obtained If is 0, it indicates that there is no fire smoke in the target region, i.e. no fire. When ​is not 0, the instantaneous increase rate of the smoke concentration is calculated , and the calculation formula is as follows: In the above formula, represents the instantaneous increase rate of the smoke concentration of the target image of the tth frame. represents the smoke concentration of the target image of the tth frame. represents the smoke concentration of the target image of the (t-1)th frame.

[0054] Further, as an optional embodiment of the present application, the step of determining the fire risk level and triggering the alarm of the corresponding level by using the instantaneous increase rate of the smoke concentration comprises: determining the fire risk level as primary and triggering the primary alarm when the instantaneous increase rate of the smoke concentration is greater than a fifth threshold value and less than a sixth threshold value, and notifying the staff to check during the primary alarm; determining the fire risk level as intermediate and triggering the intermediate alarm when the instantaneous increase rate of the smoke concentration is greater than or equal to the sixth threshold value and less than a seventh threshold value, and closing the ventilation system during the intermediate alarm; and determining the fire risk level as high-level and triggering the high-level alarm when the instantaneous increase rate of the smoke concentration is greater than or equal to the seventh threshold value, and starting the sprinkling during the high-level alarm.

[0055] Specifically, the fifth threshold value, the sixth threshold value and the seventh threshold value in the embodiment of the present application can be determined according to actual conditions, which are not limited herein. The fifth threshold value can be 0.2β crit , the sixth threshold value can be β crit , and the seventh threshold value can be 2β crit . If β t > 0.2β crit , the primary alarm is triggered, the sound and light alarm is triggered, the on-duty staff is notified to check the camera screen, β crit is a critical increase rate threshold value of the smoke concentration change, which is determined by the burning material (natural wood 0.8-1.5, textile 1.2-1.8, foam 1.8-2.5). If β t > β crit , the intermediate alarm is triggered, the ventilation system is closed, and the data is uploaded to the fire platform. β t > 2β crit , the high-level alarm is triggered, the sprinkling is started, and the fire department is notified.

[0056] The embodiment of the present application can capture the target area through the non-contact fire smoke particle detection mode, distinguish the fire smoke condition in the target area according to the difference of the gray value of the captured image of the target area and the difference of the brightness information, construct the fire risk level of the fire smoke in combination with the multiple captured images, and trigger the alarm of the corresponding level. Thus, compared with the single sensor, the method provided by the embodiment of the present application is not affected by the complex environment, and the accuracy and reliability of the detection result of the fire smoke particle are improved.

Claims

1. A fire smoke particle detection method for intelligent firefighting, characterized in that: The fire smoke particle detection method of the intelligent fire protection comprises: Acquire multiple frames of target images of the target area, and determine whether the smoke area in each frame of the target image is a fire smoke area according to the grayscale value of the smoke area in each frame of the target image; In the case of the fire smoke region, the actual brightness of each subregion in the fire smoke region is determined based on the original brightness of the light source center in the fire smoke region when there is no smoke, the brightness attenuation coefficient, and the distance, where the distance is the distance between the position of each sampling point arranged along the optical axis of the light source center and the light source center, and the subregion is formed by dividing the fire smoke region by each sampling point arranged along the optical axis of the light source center; determining a global density of the fire smoke region based on the original brightness, the actual brightness of each subregion, and the number of pixels, and determining an instantaneous growth rate of the smoke density based on the global density of the fire smoke region in the target image of the current frame and the global density of the fire smoke region in the target image of the previous frame; The instantaneous increase in smoke concentration is used to determine the fire risk level and trigger an alarm of the corresponding level.

2. The fire smoke particle detection method for intelligent fire protection according to claim 1, characterized in that: The acquiring of multiple frames of target images of the target area comprises: Comparing the image of the target area captured by the camera with the background image, and calculating the pixel-level difference between the image of the current frame and the background image by image difference method; If the pixel-level difference exceeds a first threshold, determining that a fire is suspected to have occurred in the target area; The target area is photographed at a high frame rate by the camera to obtain the target image. The frame rate of the high frame rate photographing is the ratio between the smoke diffusion speed and the maximum allowable displacement of a single frame.

3. The fire smoke particle detection method for intelligent fire protection according to claim 1, characterized in that: The step of determining whether the smoke region in each frame of the target image is a fire smoke region according to the grayscale value of the smoke region in each frame of the target image comprises: Segmenting a smoke region from each target image based on a grayscale difference between the target image and the background image in each frame; determining an area change trend based on an area of ​​a core region of a smoke region of a target image of adjacent frames, and determining a grayscale change trend based on an average grayscale of a core region of a smoke region of a target image of adjacent frames, wherein the core region is a core region with a high grayscale value where smoke particles are concentrated in the smoke region; When the area change trend is greater than a second threshold and the grayscale change trend is greater than the second threshold, determining that the smoke area is a suspected fire smoke area; In the case of the suspected fire smoke area, screening feature pixels with rich texture from the suspected fire smoke area; Determining an angle between a motion vector of the feature pixel and a unit vector in a vertical direction based on a horizontal component and a vertical component of the feature pixel, wherein the motion vector is determined by comparing position changes of the same feature pixel in target images of adjacent frames; When the proportion of the characteristic pixel points whose angle is smaller than the third threshold is greater than or equal to a fourth threshold, the suspected fire smoke area is determined to be a fire smoke area.

4. The fire smoke particle detection method for intelligent fire protection according to claim 3, characterized in that: Determining the area change trend according to the area of ​​the core area of ​​the smoke area of ​​the target image of the adjacent frames includes: Calculating a first difference between an area of ​​a core region of a smoke region of a target image in a next frame of a current frame and an area of ​​a core region of a smoke region of a target image in a current frame; determining a first ratio between the first difference and the area of ​​the core area of ​​the smoke region of the target image of the current frame as an area change trend of the core area of ​​the smoke region of the target image of the current frame; Determining the grayscale change trend according to the average grayscale of the core area of ​​the smoke area of ​​the target image of adjacent frames includes: Calculating a second difference between the average grayscale of the core area of ​​the smoke region of the target image of the next frame of the current frame and the average grayscale of the core area of ​​the smoke region of the target image of the current frame; A second ratio between the second difference and the average grayscale of the core area of ​​the smoke area of ​​the target image of the current frame is determined as a grayscale change trend of the core area of ​​the smoke area of ​​the target image of the current frame.

5. The fire smoke particle detection method for intelligent fire protection according to claim 3, characterized in that: The determining of the angle between the motion vector of the feature pixel point and the unit vector in the vertical direction according to the horizontal component and the vertical component of the feature pixel point includes: calculating a square root of a sum of the square of the horizontal component and the square of the vertical component, and calculating a third ratio of the vertical component to the square root of the sum; The third ratio is processed using an arc cosine function to obtain the included angle.

6. The fire smoke particle detection method for intelligent fire protection according to claim 1, characterized in that: The determining of the actual brightness of each sub-area in the fire smoke area according to the original brightness, brightness attenuation coefficient, and distance of the light source center in the fire smoke area when there is no smoke includes: Calculating a first product between the brightness attenuation coefficient and the absolute value of the distance; A third difference between the original brightness and the first product is determined as the actual brightness.

7. The fire smoke particle detection method for intelligent fire protection according to claim 1, characterized in that: Determining the global density of the fire smoke area according to the original brightness, the actual brightness of each sub-area, and the number of pixels includes: Determine the smoke density of each sub-area according to the actual brightness of each sub-area in the fire smoke area, the calibration constant and the original brightness; Calculating a second product between the smoke density of each sub-region and the corresponding number of pixels, and superimposing each of the second products to obtain a first superposition value; Superimposing the number of pixels in each of the sub-regions to obtain a second superimposed value; A fourth ratio between the first superposition value and the second superposition value is determined as the global concentration of the fire smoke area.

8. The fire smoke particle detection method for intelligent fire protection according to claim 7, characterized in that: Determining the smoke density of each sub-area in the fire smoke area according to the actual brightness of each sub-area, the calibration constant, and the original brightness includes: calculating an average brightness of actual brightness of each sub-area in the fire smoke area, and calculating a fifth ratio between the average brightness and the original brightness; calculating a fourth difference between the predetermined value and the fifth ratio; A third product between the calibration constant and the fourth difference is determined to be the smoke concentration.

9. The fire smoke particle detection method for intelligent fire protection according to claim 1, characterized in that: The determining of the instantaneous increase rate of smoke density based on the global density of the fire smoke region in the target image of the current frame and the global density of the fire smoke region in the target image of the previous frame includes: Calculating a fifth difference between the global density of the fire smoke region in the target image of the current frame and the global density of the fire smoke region in the target image of the previous frame, and a time interval between the target image of the current frame and the target image of the previous frame; A sixth ratio between the fifth difference and the time interval is determined as the instantaneous speed increase.

10. The fire smoke particle detection method for intelligent fire protection according to claim 1, characterized in that: The method of determining the fire risk level by using the instantaneous increase rate of the smoke concentration and triggering an alarm of a corresponding level includes: When the instantaneous increase rate of the smoke concentration is greater than the fifth threshold value and less than the sixth threshold value, the fire risk level is determined to be primary and a primary alarm is triggered, and a staff member is notified to conduct an inspection during the primary alarm; When the instantaneous increase rate of the smoke concentration is greater than or equal to the sixth threshold and less than the seventh threshold, the fire risk level is determined to be medium and a medium alarm is triggered. The ventilation system is shut down during the medium alarm. When the instantaneous increase rate of the smoke concentration is greater than or equal to the seventh threshold, the fire risk level is determined to be high and a high-level alarm is triggered, and the sprinkler is started in the high-level alarm.

Images