Fire source identification and positioning method based on infrared imaging and single-line laser radar

Abstract

An embodiment of the present invention discloses a fire source identification and positioning method based on infrared imaging and a single-line laser radar, relating to the field of fire protection technology and capable of improving the stability and accuracy of fire source identification at a relatively low cost. The present invention comprises: jointly calibrating an infrared imager and a single-line laser radar, and establishing a correspondence between infrared image data and laser radar point cloud data; extracting static features and dynamic characteristics from infrared images captured by the infrared imager, and comprehensively utilizing the static features and dynamic characteristics to identify the fire source; the pan-tilt head rotates at a constant speed at a preset angular velocity and in a preset direction, starting from an initial angle, in each scanning section of the single-line laser radar, until the fire source is first identified in the scanning section and recorded as the root of the fire source.

Description

Technical Field

[0001] The present invention relates to the field of fire protection technology, and in particular to a fire source identification and positioning method based on infrared imaging and single-line laser radar. Background Art

[0002] With my country's rapid economic development, urban buildings are becoming densely populated. Complex structures such as large supermarkets, sports centers, and industrial workshops are densely populated, making fires easily escalating into major blazes in a short period of time. Detecting fires early in their development effectively provides firefighting robots with accurate fire locations, allowing them to control water cannons and extinguish fires promptly. This effectively ensures personnel safety and prevents further property damage. However, traditional water cannons rely on firefighters' experience and observation of the fire scene to locate the fire point, significantly reducing firefighting efficiency and increasing the risks faced by firefighters.

[0003] Traditional methods rely solely on high-temperature signatures to locate fire sources, but they cannot eliminate interference from other high-temperature objects, such as high-temperature boilers. While multi-line lidar offers high performance and can meet current needs, it is expensive, has high implementation costs, and has low upper limits on angular frequency and sensitivity. Compared to using multi-line lidar to locate fire sources, single-line lidar is less expensive and has a faster response in terms of angular frequency and sensitivity. However, single-line lidar can only scan in a plane, generating a sparse point cloud and unable to determine the target's height, thus being quite limited in practical use. Summary of the Invention

[0004] The embodiments of the present invention provide a fire source identification and positioning method based on infrared imaging and single-line laser radar, which can improve the stability and accuracy of fire source identification at a relatively low cost.

[0005] To achieve the above objectives, the embodiments of the present invention adopt the following technical solutions:

[0006] The method is used for a firefighting robot, wherein an infrared imager and a single-line laser radar are installed on the firefighting robot, and the firefighting robot is provided with a pan-tilt platform, and the single-line laser radar is installed on the pan-tilt platform; the method comprises:

[0007] S1. Jointly calibrate the infrared imager and single-line lidar, and establish the corresponding relationship between the infrared image data and the lidar point cloud data;

[0008] S2. Extract static features and dynamic characteristics from the infrared image captured by the infrared imager, and use the static features and dynamic characteristics to identify the fire source;

[0009] S3. The gimbal starts from an initial angle and rotates at a preset angular velocity and in a preset direction at a constant speed. In each scanning section of the single-line laser radar, the process of S2 is repeated until the fire source is identified in the scanning section for the first time and recorded as the root of the fire source.

[0010] The fire source identification and positioning method based on infrared imaging and single-line laser radar provided by the embodiment of the present invention reduces the probability of misjudgment of fire source identification by fusing the data of the active imaging single-line laser radar and the passive imaging infrared imager, and using the dynamic characteristics of optical flow to compensate for the static characteristics of a single temperature. The traditional method of determining the fire source by relying solely on high-temperature characteristics cannot rule out the interference of other high-temperature objects, such as high-temperature boilers. The fire source will fluctuate with the airflow. According to an important feature of the flame optical flow: the intensity of the irregular movement of the flame gradually decreases from top to bottom, by calculating the optical flow direction of the pixels in a fixed-size rectangular area, if the variance of the upper half is greater than the variance of the lower half, it means that the movement direction of the lower half of the target is more consistent, while the movement of the upper half is more random. In layman's terms, the outer flame of a flame always sways with a relatively large amplitude, while the base of the flame is relatively stable. This embodiment, based on this characteristic, uses the optical flow method to represent the intensity of pixel motion, thereby identifying the base of the flame. The laser radar in this embodiment locates the fire source using a single-line laser radar and a pan-tilt system (PTZ) instead of a multi-line laser radar, which is lower in cost, has higher resolution, and is more accurate in obtaining the distance and precision of the target. Currently, multi-line laser radars on the market, such as the commonly used 16-line and 32-line models, are much more expensive than single-line models, and are primarily used for scenarios such as environmental mapping. Their resolution is higher for closer objects and lower for more distant objects (the point cloud is sparse at distant locations). In this embodiment, the single-line laser radar is mounted on a pan-tilt system, and by adjusting the level of detail in the pan-tilt motion, such as by reducing the pan-tilt step size, a higher resolution than a multi-line laser radar can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0012] Figure 1 A flow chart of a fire source identification and location method based on an infrared imager and a laser radar provided in an embodiment of the present invention;

[0013] Figure 2 A schematic diagram of the joint calibration of an infrared imager and a lidar provided in an embodiment of the present invention;

[0014] Figure 3 Schematic diagram of coordinate conversion between infrared imager and laser radar provided by an embodiment of the present invention;

[0015] Figure 4 A schematic diagram of the positioning principle of a single-line laser radar provided in an embodiment of the present invention;

[0016] Figure 5 A schematic diagram of a method flow chart provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0017] To enable those skilled in the art to better understand the technical solutions of the present invention, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. The embodiments of the present invention will be described in detail below, with examples of the embodiments illustrated in the accompanying drawings. Throughout, identical or similar reference numerals represent identical or similar elements or elements having identical or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and intended only to explain the present invention and are not to be construed as limiting the present invention. Those skilled in the art will appreciate that, unless otherwise stated, the singular forms "a," "an," "said," and "the" used herein may also include the plural forms. It should be further understood that the term "comprising" as used in the description of the present invention refers to the presence of the stated features, integers, steps, operations, elements, and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when an element is referred to as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or intervening elements may be present. Furthermore, "connected" or "coupled" as used herein may include wireless connections or couplings. The term "and / or" as used herein includes any and all combinations of one or more associated listed items. It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those skilled in the art in the art to which the present invention belongs. It should also be understood that terms such as those defined in general dictionaries should be understood to have meanings consistent with their meanings in the context of the prior art, and will not be interpreted in an idealized or overly formal sense unless defined as such herein.

[0018] The embodiment of the present invention provides a fire source identification and positioning method based on infrared imaging and single-line laser radar, such as Figure 5 As shown, the method is used for a firefighting robot, such as Figure 2As shown, the firefighting robot is equipped with an infrared imager and a single-line laser radar. The firefighting robot is equipped with a pan-tilt platform, and the single-line laser radar is mounted on the pan-tilt platform. The infrared imager can be mounted on a separate platform, while the single-line laser radar can be mounted on the platform. The relative position of the pan-tilt platform and infrared imager is determined by the rotation matrix R and translation vector t obtained through joint calibration. Each sensor is connected to a host computer, which controls the pan-tilt platform and processes data.

[0019] The method comprises:

[0020] S1. Jointly calibrate the infrared imager and single-line lidar, and establish the corresponding relationship between the infrared image data and the lidar point cloud data;

[0021] S2. Extract static features and dynamic characteristics from the infrared image captured by the infrared imager, and use the static features and dynamic characteristics to identify the fire source;

[0022] S3. The gimbal starts from an initial angle and rotates at a preset angular velocity and in a preset direction at a constant speed. In each scanning section of the single-line laser radar, the process of S2 is repeated until the fire source is identified in the scanning section for the first time and recorded as the root of the fire source.

[0023] This embodiment combines a single-line LiDAR with a gimbal to achieve scanning and positioning in 3D space. This replaces the expensive multi-line LiDAR, which can only obtain distance information from a single plane to the center of the LiDAR. The combination of a single-line LiDAR and a gimbal allows for LiDAR scanning in 3D space, while the gimbal, which can move up, down, left, right, up, left, down, and right, enables LiDAR to scan 3D space.

[0024] Specifically, sensor calibration is the prerequisite for sensor data fusion. The infrared imager and the single-line laser radar are jointly calibrated so that the infrared image and the laser radar point cloud data are synchronized in space and time, preparing for the fusion of the two data. The purpose of joint calibration is to let the two sensors know their relative positions, so as to facilitate the subsequent matching of the infrared imager's image data and the laser radar's point cloud data. The temperature of a certain point and the distance of a certain point from the center of the entire mechanical structure can be known, so as to realize the identification and positioning of the fire source. The specific implementation process of this embodiment is as follows. Figure 1As shown in the figure, the infrared imager and the laser radar are jointly calibrated so that the infrared image and the point cloud data are synchronized in space and time, preparing for the fusion of the two data. Specifically, since the infrared imager and the laser radar cannot be in the same position, the collected data has a certain deviation in space, so the infrared imager and the laser radar are first jointly calibrated to establish the correspondence between the laser radar point cloud and the infrared image pixels. The infrared image data (u, v) and the laser radar point cloud data (x, y, z) are mapped to the two-dimensional image data through the joint calibration matrix M. In S1, it includes:

[0025] Through the joint calibration matrix M, the infrared image data (u, v) and the lidar point cloud data (x, y, z) are mapped to the two-dimensional image data, where the mapping relationship is expressed as:

[0026]

[0027]

[0028] (u0, v0) is the coordinate of the center of the infrared image, f u and f v are the scale factors of the infrared imager in the horizontal and vertical directions, are the internal parameters of the camera, R represents the rotation matrix, t represents the translation vector, are the external parameters of the camera, v and u represent the horizontal and vertical coordinates of the infrared image, respectively, and x, y, z represent the three-dimensional coordinates of the lidar point cloud data. Specifically, the elements m11 to m34 in the matrix in the formula represent the product of the previous two matrices, that is, they can be understood in the general way of matrix multiplication. M establishes the correspondence between the two-dimensional image and the three-dimensional lidar point cloud, so it can be called a joint calibration matrix. It can be simply understood that M is essentially obtained by the camera parameters (fu, fv, u0, v0) and the rotation matrix R and the translation matrix t.

[0029] After a series of matrix transformations, the conversion relationship between infrared image data (u, v) and lidar point cloud data (x, y, z) can be obtained. The conversion relationship between infrared image data (u, v) and lidar point cloud data (x, y, z) is expressed as:

[0030]

[0031]

[0032] After joint calibration, the relative position of the infrared imager and camera remains unchanged. In practical applications, a series of linear equations can be derived based on the plane of the calibration plate in different poses, and the calibration matrix M can be solved to map the lidar data onto the infrared image. After joint calibration, the relative position of the infrared imager and camera cannot be changed; that is, the infrared imager and lidar rotate together.

[0033] In this embodiment, S2 includes: capturing an infrared image using an infrared imager; extracting infrared image features from the captured infrared image; and identifying the fire source based on the infrared image features. The extracted infrared image features include: static temperature distribution features within a single frame, and dynamic optical flow features between adjacent frames. The infrared imager captures infrared images in real time, extracting both static temperature distribution features within a single frame and dynamic optical flow features between adjacent frames. After image processing, an appropriate threshold is determined to obtain the minimum bounding rectangle target area of ​​the infrared image, thereby identifying the fire source. The innovative feature of the infrared imager in identifying the fire source is that a temperature threshold is generally set. If the temperature exceeds this threshold, the fire source is identified. This is referred to as the static temperature distribution feature. This embodiment further integrates and utilizes dynamic features based on static features, thereby incorporating information between frames. In practical applications, high temperatures are not necessarily fire sources, but flames are not static and fluctuate during combustion. Therefore, this embodiment utilizes this characteristic of flames to improve the identification method and enhance the accuracy of fire source identification.

[0034] The step of extracting infrared image characteristics from the collected infrared image includes:

[0035] The acquired infrared image is converted into a grayscale image and decomposed into single frame images.

[0036] According to the preset threshold, the image of a single frame is converted into a binary image, and the binary image is further segmented to obtain the target area. Wherein, the image of a single frame is converted into a binary image, for example, the grayscale value of the part greater than the threshold can be set to 255, and the grayscale value of the part less than the threshold can be set to 0, including: binary image T(u, v) is the temperature corresponding to the image, -T th Indicates the empirical value of the temperature threshold. Specifically, the target area is only a portion of the pixel area of ​​the binary image, and segmentation has not yet been performed here. The subsequent operation on the target area is equivalent to the segmentation of the binary image.

[0037] For a pixel point in the target area, obtain the velocity components of the pixel point in the x and y directions of the horizontal and vertical coordinate axes, then obtain the optical flow corresponding to the pixel point based on the velocity components, and further obtain the variance of the optical flow direction of the pixel point. Wherein, obtaining the optical flow corresponding to the pixel point based on the velocity components includes:

[0038]

[0039]

[0040] Among them, U, V are the velocity components of the pixel point (x, y) in the x and y directions, I i represents the pixel value of the i-th pixel, x i Indicates the horizontal coordinate of the i-th pixel, y i represents the vertical coordinate of the i-th pixel, t represents time, n represents the number of pixels in the window centered on the pixel point (x, y), and i represents the pixel number, which is usually a positive integer.

[0041] The method of obtaining the variance of the pixel optical flow direction includes: obtaining the pixel optical flow direction θ i ∈[-π,π]. Get the variance of the optical flow direction Among them, ρ 2 is the variance of the pixel optical flow direction, Represents the average value of the pixel optical flow direction, σ 2 The value of is negatively correlated with the consistency of the motion direction of the pixels in the target area.

[0042] The target area is divided into an upper half and a lower half. If the variance of the optical flow direction of the pixels in the upper half is greater than the variance of the optical flow direction of the pixels in the lower half, it is determined that there is a fire source in the target area.

[0043] For example, an infrared imager can identify the temperature information of an object and obtain an infrared image of the environment. The higher the temperature, the brighter the image will be, and the lower the temperature, the darker the image will be. Based on this feature, a single frame of the image is taken and the image is first converted into a grayscale image. According to different environments, the appropriate threshold is set through testing to convert the image into a binary image. Specifically, the grayscale value of the part greater than the threshold is set to 255, and the grayscale value of the part less than the threshold is set to 0. Then, the image processing functions provided by the opencv library, such as the dilation operation (dilate()), are used to filter the interference information inside the grayscale binary image, so that the flame in the grayscale binary image is connected. Let A be the infrared image, T(u, v) be the temperature corresponding to the image, then the binary image B(u, v) obtained after segmentation is segmented, and the target area C(u, v) is obtained by segmentation;

[0044]

[0045] However, in complex environments, using temperature features alone to determine the location of a fire source can be easily interfered with by other high-temperature objects in the environment, affecting the judgment. Since the fire source fluctuates with the airflow, the optical flow method is used to extract the dynamic characteristics of the fire source. Optical flow is the pattern of visual motion of an image object between two consecutive frames caused by the movement of an object or camera. A window function is constructed and translated across the image to observe the degree of consistency in the direction of motion of the pixels within the window. Specifically, according to the basic equation of the optical flow method:

[0046]

[0047] Where U and V are the velocity components of the pixel point (x, y) in the x and y directions, and I i represents the pixel value of the i-th pixel, x i Indicates the horizontal coordinate of the i-th pixel, y i The Lucas-Kanade algorithm assumes that the optical flow equation is complete within a small window around the pixel and satisfies the spatial consistency assumption. The least squares method is used to solve the overdetermined equation:

[0048]

[0049] Among them, I x1 to I xn Respectively represent the differential of n pixel values ​​in the window in the x-axis direction, I y1 to I yn Represents the differential of n pixel values ​​in the window in the y-axis direction, I t1 to I tn Respectively represent the differential of n pixel values ​​in the window with respect to time,

[0050] The optical flow (U, V) corresponding to the pixel point (x, y) can be obtained, where n represents the number of pixels in the window centered on the pixel point (x, y).

[0051]

[0052] Then calculate the direction of the pixel optical flow in the target area C(x, y) as:

[0053]

[0054] The variance is calculated by the following method, represents the average value of the optical flow direction:

[0055]

[0056] Where ρ 2 is the variance of the pixel optical flow direction, ρ 2 The closer it is to 0, the more consistent the movement direction of the pixels in the target area C(u, v) is. 2 The larger the variance, the greater the change in the pixel's motion direction. Based on a key characteristic of flame optical flow: the intensity of irregular flame motion decreases gradually from top to bottom, C(u, v) is split into two parts. If the variance of the upper part is greater than that of the lower part, the region is considered a fire source. Otherwise, the region is considered a non-fire source. The target region, after eliminating other high-temperature interference through dynamic characteristics, is set as D(u, v).

[0057] In this embodiment, the uniform rotation at a preset angular velocity and in a preset direction includes: the pan / tilt head uniformly rotating upward at a preset angular velocity, wherein the initial angle is 45° when the pan / tilt head is looking down. In practical applications, the angular velocity is set via the pan / tilt head protocol and is determined by a two-digit hexadecimal number, with 00 representing the lowest speed and 3f representing the highest speed. It can be set to be positively correlated with the distance to the fire source, i.e., the speed can be lower for closer distances and higher for farther distances. In a preferred embodiment, the angular velocity can be set to 3° / second to 6° / second.

[0058] Furthermore, before S3, the gimbal is first rotated horizontally by 360°, and during the horizontal rotation, the direction of the suspected fire source is identified by the infrared imager, wherein the infrared imager obtains the temperature in the shooting field of view in real time during the rotation with the gimbal, and takes the direction with the highest temperature as the direction of the suspected fire source.

[0059] Among them, you can first cruise horizontally, that is, turn horizontally, and use the infrared imager to first determine the direction where the fire source is most likely to appear, because the infrared imager has a relatively large field of view. Then, at a fixed angular velocity in the vertical direction, let the gimbal scan again in the vertical direction until it scans the root of the fire source and stops. Specifically, a single-line laser radar is fixed on the gimbal, and the gimbal is given an initial downward angle of 45°. By giving the gimbal a fixed angular velocity, the gimbal is controlled to tilt up at a uniform speed. Scan each section, and use the method of step two to determine whether there is a fire source on each section. The gimbal is tilted up at a uniform speed until any point on the section appears in the rectangular area obtained in step two. This point is considered to be the root of the fire source, and the fire source is located. For example, a single-line LiDAR is fixedly mounted on a gimbal that can rotate 360° infinitely. Considering that fire source location is often far from firefighters, relying solely on human position determination is insufficiently accurate. Therefore, LiDAR is used to locate the fire source. Therefore, the initial gimbal attitude is set to a 45° downward angle. By setting a fixed angular velocity, the gimbal tilts upward at a constant rate. A single laser emitter rotates uniformly within the LiDAR, emitting a laser beam once for each small rotation. After a certain rotation angle, a complete frame of data is generated. Therefore, the data from the single-line LiDAR can be viewed as a row of dots at the same height. Step 1 already fuses the single-line LiDAR data with the infrared image. This allows us to obtain the position (u, v) of each dot in the row at the same height in the infrared image, as well as the corresponding temperature T for each dot. Identify the fire source area using the method in step 2. If there is no fire source target area at present, that is, any array point (u, v) is not within D(u, v), continue scanning until (u, v) appears within D(u, v), and this point is considered to be the root point of the fire source.

[0060] The main advantages of this embodiment are: by fusing the data of an active imaging single-line laser radar and a passive imaging infrared imager, due to the different working principles of the sensors, different noises can be suppressed, thereby enhancing the stability and accuracy of fire source identification and positioning. The single-line laser radar transmits infrared lasers and calculates the time of flight (TOF) to obtain the distance information of the target, thereby achieving positioning. At the same time, it is different from traditional temperature and smoke sensors that are suitable for locating fire sources in small spaces and cannot provide accurate internal information of the fire scene. The method proposed in this embodiment still has good fire source identification and positioning capabilities in complex environments such as outdoors, large buildings, and poor lighting. The infrared image processing part of this embodiment uses the dynamic characteristics of optical flow to compensate for the static characteristics of a single temperature, reducing the probability of misjudgment of fire source identification. The traditional method of relying solely on high temperature characteristics to determine the fire source cannot rule out interference from other high-temperature objects, such as high-temperature boilers. Fire sources fluctuate with airflow. Based on a key characteristic of flame optical flow: the intensity of the irregular flame motion gradually decreases from top to bottom, the optical flow direction of pixels in a fixed-size rectangular area is calculated. If the variance of the upper half is greater than that of the lower half, the motion of the lower half of the target is more consistent, while the motion of the upper half is more random. This can be understood as the outer flame of a flame always fluctuating with a relatively large amplitude, while the base of the flame is relatively stable. This embodiment leverages this characteristic by using optical flow to represent the intensity of pixel motion, thereby identifying the base of the flame. This embodiment uses a single-line laser radar and pan-tilt system to locate the fire source, replacing a multi-line laser radar. This results in lower cost, higher resolution, and more accurate target range and accuracy. Currently available multi-line laser radars, such as the commonly used 16- and 32-line models, are much more expensive than single-line models. Multi-line laser radars are primarily used for environmental mapping, resulting in higher resolution for nearby objects and lower resolution for distant objects (the point cloud is sparse at long distances). In this embodiment, a single-line laser radar is installed on a gimbal. By setting the level of detail of the gimbal movement, such as reducing the step size of the gimbal, a higher resolution than that of a multi-line laser radar can be achieved.

[0061] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited to this. Any changes or replacements that can be easily thought of by any technician familiar with this technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A fire source identification and positioning method based on infrared imaging and single-line laser radar, characterized in that: The method is used for a firefighting robot, which is equipped with an infrared imager and a single-line laser radar. The firefighting robot is provided with a pan-tilt platform, and the single-line laser radar is installed on the pan-tilt platform. The single-line laser radar is used to obtain distance information from a single plane to the center of the laser radar. The method comprises: S1. Jointly calibrate the infrared imager and single-line lidar, and establish the corresponding relationship between the infrared image data and the lidar point cloud data; S2. Extract static features and dynamic characteristics from the infrared image captured by the infrared imager, and use the static features and dynamic characteristics to identify the fire source; S3, the pan / tilt head rotates at a preset angular velocity and in a preset direction at a constant speed starting from the initial angle, wherein the process of S2 is repeated in each scanning section of the single-line laser radar until the fire source is identified in the scanning section for the first time and recorded as the root of the fire source; In S1, including: Through the joint calibration matrix M, the infrared image data (u, v) and the lidar point cloud data (x, y, z) are mapped to the two-dimensional image data, where the mapping relationship is expressed as: (u0,v0) is the coordinate of the center of the infrared image, f u and f v are the scale factors of the infrared imager in the horizontal and vertical directions, R represents the rotation matrix, t represents the translation vector, v and u represent the horizontal and vertical coordinates of the infrared image, and x, y, and z represent the three-dimensional coordinates of the lidar point cloud data; The conversion relationship between infrared image data (u, v) and lidar point cloud data (x, y, z) is expressed as: Among them, after joint calibration, the relative positions of the infrared imager and the camera remain unchanged; In S2, including: Collect infrared images through an infrared imager; extracting infrared image characteristics from the collected infrared image; identifying a fire source based on the infrared image characteristics; The extracted infrared image features include: static features of temperature distribution of a single frame, and dynamic characteristics of optical flow between adjacent frames; The step of extracting infrared image characteristics from the collected infrared image includes: Converting the acquired infrared image into a grayscale image and decomposing it into single frame images; According to the preset threshold, the single frame image is converted into a binary image, and the binary image is further segmented to obtain the target area; For the pixel points in the target area, the velocity components of the pixel points in the x and y directions of the horizontal and vertical coordinate axes are obtained, and then the optical flow corresponding to the pixel points is obtained based on the velocity components, and the variance of the optical flow direction of the pixel points is further obtained; The target area is divided into an upper half and a lower half. If the variance of the optical flow direction of the pixels in the upper half is greater than the variance of the optical flow direction of the pixels in the lower half, it is determined that there is a fire source in the target area. The obtaining of the optical flow corresponding to the pixel point according to the velocity component includes: Among them, U, V are the velocity components of the pixel point (x, y) in the x and y directions, I i represents the pixel value of the i-th pixel, x i Indicates the horizontal coordinate of the i-th pixel, y i represents the vertical coordinate of the i-th pixel, t represents time, n represents the number of pixels in the window centered on the pixel point (x, y), and i represents a positive integer; The variance of the optical flow direction is: Among them, σ 2 is the variance of the pixel optical flow direction, Represents the average value of the pixel optical flow direction, σ 2 The value of is negatively correlated with the consistency of the movement direction of the pixels in the target area. The direction of the optical flow of the pixels is θ i ∈[-π,π]; The target area C(u,v) is divided into two parts, upper and lower. If the variance of the upper part is greater than that of the lower part, the area is determined to be a fire source. Otherwise, the area is determined to be a non-fire source area. The target area after excluding other high-temperature interference through dynamic characteristics is D(u,v); If any array point (u, v) is not within D(u, v), it means that there is no fire source target area. Then continue scanning until (u, v) appears within D(u, v), and this point is considered to be the root point of the fire source.

2. The method according to claim 1, characterized in that The step of converting a single frame image into a binary image comprises: binary image T(u,v) is the temperature corresponding to the image, T th Indicates the empirical value of the temperature threshold.

3. The method according to claim 1, characterized in that The uniform rotation at a preset angular velocity and in a preset direction includes: The gimbal rotates upward at a preset angular velocity, wherein the initial angle is 45° when the gimbal looks down.

4. The method according to claim 1, wherein Also includes: Before S3, the gimbal is first rotated horizontally by 360°, and during the horizontal rotation, the direction of the suspected fire source is identified by the infrared imager, wherein the infrared imager obtains the temperature in the shooting field of view in real time during the rotation of the gimbal, and takes the direction with the highest temperature as the direction of the suspected fire source.

Images