Flame detection method and device, server and storage medium

Abstract

The invention discloses a flame detection method and device, a server and a storage medium. The method comprises the steps of obtaining a suspected flame image sequence in response to triggering of a flame detection event; wherein the suspected flame image sequence is composed of at least two suspected flame images; flame edge features and local area brightness uniformity of each suspected flame image in the suspected flame image sequence are determined respectively; determining the edge vibration amplitude change uniformity and the brightness amplitude change uniformity of the suspected flame image sequence; and detecting whether a suspected flame region in the suspected flame image sequence is a real flame region or not according to the flame edge features, the local region brightness uniformity, the edge vibration amplitude change uniformity and the brightness amplitude change uniformity. According to the scheme, the flame false detection problem caused by various light sources such as lamplight can be solved, and the real flame detection accuracy is effectively improved.

Description

Technical Field

[0001] This invention relates to the field of flame detection technology, and in particular to a flame detection method, apparatus, server, and storage medium. Background Technology

[0002] Fire is a highly dangerous safety event, posing a significant threat to people's normal lives and property. Fire detection technology is constantly evolving with the development of science and technology, especially in the field of production safety, where various smoke and fire detection technologies are emerging, involving different disciplines, each with its own advantages and disadvantages, accuracy, and cost. With the development of AI monitoring technology, vision-based flame detection technology integrated into cameras is gradually becoming the mainstream direction in the industry.

[0003] Fire detection technology based on computer vision has many subcategories, such as pure visible light solutions, thermal imaging solutions, and multispectral fusion solutions combining infrared or ultraviolet light. Currently, in engineering, considering ease of use and cost, pure visible light solutions offer significantly more room for technological development, especially in the backend or cloud technology field. They can directly connect to a large number of older, non-intelligent or smoke-detecting front-end cameras, using the ordinary visible light or infrared video streams captured by these cameras to perform smoke analysis on the backend server.

[0004] Currently, in backend products, flame detection technology based on pure visible light solutions does not yet offer particularly high accuracy and reliability, lagging significantly behind technologies such as thermal imaging. False detection is a relatively prominent issue with visible light solutions, especially in low-light environments like nighttime when flames are highly similar to various lights, leading to severe false detections. For example, Figure 1a This is a schematic diagram of a common flame in the prior art. Figure 1b This diagram illustrates a common misidentification of flames in existing technologies. Furthermore, current backend solutions do not effectively support flame recognition in infrared scenes. Summary of the Invention

[0005] This invention provides a flame detection method, device, server, and storage medium, which can solve the problem of false flame detection caused by various light sources and effectively improve the detection accuracy of real flames.

[0006] According to one aspect of the present invention, a flame detection method is provided, comprising:

[0007] In response to a flame detection event being triggered, a sequence of suspected flame images is acquired; wherein the suspected flame image sequence consists of at least two suspected flame images;

[0008] The flame edge features and local brightness uniformity of each suspected flame image in the suspected flame image sequence are determined respectively;

[0009] Determine the uniformity of edge vibration amplitude variation and brightness amplitude variation in the suspected flame image sequence;

[0010] Based on the flame edge features, the local area brightness uniformity, the edge vibration amplitude variation uniformity, and the brightness amplitude variation uniformity, it is determined whether the suspected flame area in the suspected flame image sequence is a real flame area.

[0011] According to another aspect of the present invention, a flame detection device is provided, comprising:

[0012] A suspected flame image sequence acquisition module is used to acquire a suspected flame image sequence in response to a flame detection event being triggered; wherein, the suspected flame image sequence consists of at least two suspected flame images;

[0013] The flame edge feature determination module is used to determine the flame edge features and local brightness uniformity of each suspected flame image in the suspected flame image sequence.

[0014] The uniformity of change module is used to determine the uniformity of edge vibration amplitude change and brightness amplitude change of the suspected flame image sequence.

[0015] The flame detection module is used to detect whether a suspected flame region in the suspected flame image sequence is a real flame region based on the flame edge features, the brightness uniformity of the local area, the edge vibration amplitude variation uniformity, and the brightness amplitude variation uniformity.

[0016] According to another aspect of the present invention, a server is provided, the server comprising:

[0017] At least one processor; and

[0018] A memory communicatively connected to the at least one processor; wherein,

[0019] The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the flame detection method according to any embodiment of the present invention.

[0020] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the flame detection method according to any embodiment of the present invention.

[0021] The flame detection scheme of this invention, in response to a flame detection event, acquires a sequence of suspected flame images; wherein the suspected flame image sequence consists of at least two suspected flame images; the flame edge features and local area brightness uniformity of each suspected flame image in the suspected flame image sequence are determined; the edge vibration amplitude variation uniformity and brightness amplitude variation uniformity of the suspected flame image sequence are determined; based on the flame edge features, the local area brightness uniformity, the edge vibration amplitude variation uniformity, and the brightness amplitude variation uniformity, it is detected whether the suspected flame area in the suspected flame image sequence is a real flame area. Through the technical solution provided by this invention, the dynamic features of flame combustion are mined, which not only solves the problem of false flame detection caused by various light sources and effectively improves the detection accuracy of real flames, but also simultaneously supports flame recognition in visible light mode, nighttime scenes, and infrared scenes.

[0022] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1a A schematic diagram of a common flame in existing technology;

[0025] Figure 1b This is a diagram illustrating a common misidentification of flames in existing technologies.

[0026] Figure 2 A flowchart of a flame detection method provided in an embodiment of the present invention;

[0027] Figure 3 A schematic diagram illustrating the process of extracting the edge contour of a suspected flame region in a suspected flame image, provided as an embodiment of the present invention;

[0028] Figure 4 A schematic diagram illustrating the distribution of maximum and minimum key points on the edge contour corresponding to each suspected flame sub-region in a suspected flame image, provided in an embodiment of the present invention.

[0029] Figure 5 A schematic diagram of the distribution of key brightness points provided in an embodiment of the present invention;

[0030] Figure 6 This is a schematic diagram of the structure of a flame detection device provided in an embodiment of the present invention;

[0031] Figure 7 A schematic diagram of the server structure for implementing the flame detection method of this invention. Detailed Implementation

[0032] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0033] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0034] Figure 2 This is a flowchart illustrating a flame detection method provided in an embodiment of the present invention. This embodiment is applicable to flame detection situations. The method can be executed by a flame detection device, which can be implemented in hardware and / or software and can be configured in a server. Figure 2 As shown, the method includes:

[0035] S210. In response to a flame detection event being triggered, a sequence of suspected flame images is acquired; wherein the suspected flame image sequence consists of at least two suspected flame images.

[0036] In this embodiment of the invention, when a flame detection command is detected, a flame detection event is determined to be triggered. In response to the triggering of the flame detection event, the video stream corresponding to the target detection region is acquired, and flame detection is performed on each video frame of the video stream based on a preset flame detection algorithm to determine a suspected flame detection box in each video frame. The suspected flame region contained in the corresponding suspected flame detection box is extracted from each video frame of the video stream to generate a suspected flame image. A suspected flame image sequence is formed based on the suspected flame images extracted from each video frame of the video stream. Optionally, flame detection can be performed on the video frames based on a pre-built flame detection model, wherein the flame detection model is a deep neural network model, such as the YOLO-v5 model.

[0037] Optionally, to improve flame detection efficiency, flame detection can be performed on the video stream using a flame detection model based on a preset detection frequency. That is, the frequency of the video stream input to the flame detection model is the preset detection frequency, for example, 3 frames / second and the video stream frame rate is 25 frames / second. Then, the flame detection model detects one video frame every 8 frames in the video stream. For example, the flame detection model performs flame detection on the last video frame in every 8 frames of the video stream, identifies a suspected flame detection box in the last video frame, and extracts the suspected flame region contained within the suspected flame detection box from the last video frame. The regions in the first 7 video frames that are at the same position as the suspected flame detection region in the last video frame are taken as the suspected flame detection regions in the first 7 video frames. In this embodiment of the invention, a suspected flame image sequence can be formed based on the suspected flame detection regions extracted from every 8 video frames in the video stream, or a suspected flame image sequence can be generated based on the suspected flame regions extracted from the video frames in the video stream where flame detection is performed.

[0038] Since the target of a fire typically does not exhibit significant displacement in a fire scene, the suspected flame detection bounding box in the last video frame of every 8 frames can be appropriately expanded (e.g., increasing the width and height of the suspected flame detection box by 1 / 4 of their original width and height). Then, based on the coordinates of the expanded suspected flame detection box, the corresponding regions in the previous 7 video frames can be extracted to obtain 8 suspected flame images. To reduce missed detections and ensure a high flame detection rate, the flame detection strategy for the video stream can be set relatively leniently.

[0039] S220. Determine the flame edge features and local brightness uniformity of each suspected flame image in the suspected flame image sequence.

[0040] Analysis of the basic shape features of a single-frame image of a burning flame reveals that the flame has irregular, convex and concave boundaries, with the protruding parts often being sharp and extending upwards. Therefore, in this embodiment of the invention, each suspected flame image in the suspected flame image sequence is analyzed to determine the edge contour of the suspected flame region in each image, and the flame edge features of the edge contour of the suspected flame region are determined. These flame edge features reflect the distribution characteristics of the edge contour of the suspected flame region in the suspected flame image. For example, multiple contour key points are determined for the edge contour of the suspected flame region in the suspected flame image, and the flame edge features of the suspected flame image are determined based on these key points. Since the brightness and saturation of different areas of the flame are not uniform and vary considerably, and the color state of the flame differs under various lighting, fuel, and temperature conditions, and in order to identify flames in nighttime and infrared scenes, each suspected flame image in the suspected flame image sequence is analyzed to determine the local brightness uniformity of each suspected flame image. This local brightness uniformity reflects the brightness distribution characteristics of the suspected flame region in the suspected flame image. For example, multiple brightness key points are determined in the suspected flame area of ​​the suspected flame image, and the brightness uniformity of the local area of ​​the suspected flame image is determined based on the multiple brightness key points.

[0041] Optionally, determining the flame edge features of each suspected flame image in the suspected flame image sequence includes: for each suspected flame image in the suspected flame image sequence, determining the centroid of the suspected flame region in the suspected flame image, and dividing the suspected flame region in the suspected flame image into equal-angle regions centered on the centroid to generate at least two suspected flame sub-regions; determining the maximum and minimum key points on the edge contour corresponding to each suspected flame sub-region; and determining the flame edge features of the suspected flame image based on the maximum and minimum key points corresponding to the at least two suspected flame sub-regions.

[0042] In this embodiment of the invention, a suspected flame region is determined in a suspected flame image, wherein the suspected flame region is an irregularly shaped region, and the centroid of the suspected flame region is obtained. For example, the centroid of the suspected flame region in the i-th suspected flame image in a suspected flame image sequence can be represented as O. i (x i y iThe suspected flame region in the suspected flame image is divided into N sub-regions at equal angles, centered on the centroid, where N is an integer greater than or equal to 2. For example, the suspected flame region in the suspected flame image is divided into 8 sub-regions centered on the centroid, where the two boundary lines emanating from the centroid in each sub-region have an angle of 45°. The suspected flame regions in the suspected flame image are then processed through binarization, denoising, and connectivity adjustments to extract their edge contours. For example... Figure 3 This is a schematic diagram illustrating the process of extracting the edge contour of a suspected flame region in a suspected flame image, as provided in an embodiment of the present invention.

[0043] For each suspected flame sub-region, determine the maximum and minimum keypoints on the edge contour. The maximum keypoint is the point on the edge contour farthest from the centroid, and the minimum keypoint is the point on the edge contour closest to the centroid. For example, for each suspected flame sub-region, determine the peaks and troughs on the corresponding edge contour, using the peak farthest from the centroid as the maximum keypoint and the trough closest to the centroid as the minimum keypoint. Another example is determining the contour keypoints on the edge contour of the suspected flame region in the suspected flame image, where the contour keypoints include all peaks and troughs on the edge contour of the suspected flame region. For instance, a CNN deep model can be used to perform regression calculations on the contour keypoints on the edge contour of the suspected flame region. Select the maximum and minimum keypoints from the contour keypoints on the edge contour of the suspected flame sub-region. The edge contour corresponding to the suspected flame sub-region may contain one or more contour key points (peaks and / or troughs), or it may not contain any contour key points. When multiple contour key points exist on the edge contour corresponding to the suspected flame sub-region, the contour key point farthest from the centroid is designated as the maximum key point, and the contour key point closest to the centroid is designated as the minimum key point. When only one contour key point exists on the edge contour corresponding to the suspected flame sub-region, if the contour key point is a peak, it is designated as the maximum key point, and the intersection point of the two boundary lines drawn from the centroid with the edge contour corresponding to the suspected flame sub-region, the intersection point closest to the centroid, is designated as the minimum key point; if the contour key point is a trough, it is designated as the minimum key point, and the intersection point of the two boundary lines drawn from the centroid with the edge contour corresponding to the suspected flame sub-region, the intersection point farthest from the centroid, is designated as the maximum key point. When there are no contour keypoints on the edge contour corresponding to the suspected flame sub-region, the intersection point between the two boundary lines drawn from the centroid and the edge contour corresponding to the suspected flame sub-region is designated as the maximum keypoint, and the intersection point closest to the centroid is designated as the minimum keypoint. For example, Figure 4 This is a schematic diagram showing the distribution of maximum and minimum key points on the edge contour corresponding to each suspected flame sub-region in a suspected flame image provided by an embodiment of the present invention.

[0044] In this embodiment of the invention, the flame edge features of a suspected flame image are determined based on the maximum and minimum keypoints corresponding to at least two suspected flame sub-regions. For example, all maximum and minimum keypoints in the suspected flame image sequence are transformed to the same Cartesian coordinate system, and the coordinates of each maximum and minimum keypoint in the same Cartesian coordinate system are determined. For example, if the suspected flame image sequence contains M suspected flame images, and the suspected flame region in each suspected flame image is divided into N suspected flame sub-regions, then each suspected flame image contains N maximum and N minimum keypoints. The maximum and minimum keypoints corresponding to the j-th suspected flame sub-region in the i-th suspected flame image can be represented as H. ij (x h_ij y h_ij L ij (x l_ij y l_ij Based on the maximum and minimum key points corresponding to N suspected flame sub-regions in the suspected flame image in the same Cartesian coordinate system, the flame edge features of the suspected flame image are determined.

[0045] Optionally, the flame edge features include an average vibration amplitude difference; determining the flame edge features of the suspected flame image based on the maximum and minimum key points corresponding to the at least two suspected flame sub-regions includes: determining the maximum vibration amplitude of the maximum key point and the minimum vibration amplitude of the corresponding minimum key point for each suspected flame sub-region, and calculating the vibration amplitude difference between the maximum and minimum vibration amplitudes; and calculating the average vibration amplitude difference of the suspected flame image based on the vibration amplitude differences corresponding to the at least two suspected flame sub-regions.

[0046] For example, for each suspected flame sub-region in a suspected flame image, the maximum vibration amplitude of the maximum key point and the minimum vibration amplitude of the minimum key point in the suspected flame sub-region are determined. For example, the ordinate y of the maximum key point is... h_ij The maximum vibration amplitude at the key point of maximum value is used as the ordinate of the key point of minimum value. l_ij The minimum vibration amplitude is used as the key point for minima. The difference between the maximum and minimum vibration amplitudes of each suspected flame sub-region is calculated and used as the vibration amplitude difference. For example, the vibration amplitude difference corresponding to the j-th suspected flame sub-region in the i-th suspected flame image in a suspected flame image sequence can be expressed as y. h_ij -y l_ijUnderstandably, the vibration amplitude difference corresponding to the N suspected flame sub-regions in each suspected flame image can be calculated using the above method. Based on the vibration amplitude differences corresponding to the N suspected flame sub-regions in the suspected flame image, the average vibration amplitude difference of the suspected flame image is calculated. This average vibration amplitude difference reflects the difference in the undulation of the edge contour between the real flame and the falsely detected flame target in a single frame. For example, the average vibration amplitude difference α of the i-th suspected flame image in the suspected flame image sequence... i It can be represented as:

[0047] Optionally, the flame edge features include edge roundness; determining the flame edge features of the suspected flame image based on the maximum and minimum keypoints corresponding to the at least two suspected flame sub-regions includes: determining the mean of the first maximum vibration amplitude of the maximum keypoints and the mean of the first minimum vibration amplitude of the corresponding minimum keypoints for the at least two suspected flame sub-regions; calculating the standard deviation of the first maximum vibration amplitude based on the maximum vibration amplitude of the maximum keypoints and the mean of the first maximum vibration amplitude for the at least two suspected flame sub-regions, and calculating the standard deviation of the first minimum vibration amplitude based on the minimum vibration amplitude of the minimum keypoints and the mean of the first minimum vibration amplitude for the at least two suspected flame sub-regions; and calculating the edge roundness of the suspected flame image based on the standard deviation of the first maximum vibration amplitude and the standard deviation of the first minimum vibration amplitude.

[0048] In the embodiments of the invention, the mean of the maximum vibration amplitude of the N maximum key points corresponding to the N suspected flame sub-regions in the suspected flame image, and the mean of the minimum vibration amplitude of the N minimum key points are calculated. For ease of description, the mean of the maximum vibration amplitude of the N maximum key points in the suspected flame image is called the first maximum vibration amplitude mean, and the mean of the minimum vibration amplitude of the N minimum key points in the suspected flame image is called the first minimum vibration amplitude mean. For example, the first maximum vibration amplitude mean and the first minimum vibration amplitude mean of the i-th suspected flame image in the suspected flame image sequence can be represented as σ. hi σ liThe standard deviation of the first maximum vibration amplitude is calculated based on the maximum vibration amplitude of N maximum keypoints corresponding to N suspected flame sub-regions in the suspected flame image and the mean of the first maximum vibration amplitude; the standard deviation of the first minimum vibration amplitude is calculated based on the minimum vibration amplitude of N minimum keypoints corresponding to N suspected flame sub-regions in the suspected flame image and the mean of the first minimum vibration amplitude. The mean of the standard deviations of the first maximum vibration amplitude and the first minimum vibration amplitude is used as the edge roundness of the suspected flame image. The edge roundness reflects the difference in overall roundness between the edge contour of a real flame and a falsely detected flame target in a single frame. For example, the edge roundness β of the i-th suspected flame image in the suspected flame image sequence... i It can be represented as:

[0049] Optionally, determining the local brightness uniformity of each suspected flame image in the suspected flame image sequence includes: for each suspected flame image in the suspected flame image sequence, randomly selecting m brightness key points on the first line connecting the maximum value key point to the centroid and the second line connecting the corresponding minimum value key point to the centroid for each suspected flame sub-region of the suspected flame image; for each suspected flame sub-region in the suspected flame image, determining the first brightness mean of all the brightness key points on the first and / or second lines connecting the suspected flame sub-region, and calculating a first brightness standard deviation based on the brightness value of each brightness key point on the first and / or second lines connecting the suspected flame sub-region and the first brightness mean; and determining the local brightness uniformity of the suspected flame image based on the first brightness standard deviation corresponding to at least two suspected flame sub-regions.

[0050] For example, for each suspected flame sub-region of a suspected flame image, m luminance key points are randomly selected on the first line connecting the maximum key point and the centroid of the suspected flame sub-region, and m luminance key points are randomly selected on the second line connecting the minimum key point and the centroid. Therefore, 2m luminance key points can be identified in the suspected flame sub-region. Optionally, m uniformly distributed luminance key points can be selected on the first and second lines respectively. The m-th luminance key point in the j-th suspected flame sub-region of the i-th suspected flame image in the suspected flame image sequence can be represented as H. ij_m (x ij_m y ij_m For example, Figure 5This is a schematic diagram illustrating the distribution of brightness key points according to an embodiment of the present invention. It is understood that, following the above method, 2m*N brightness key points can be determined from a suspected flame image, resulting in a total of 2m*N*M brightness key points determined in the suspected flame image sequence. Each suspected flame image is grayscaled, and the mean brightness value of all brightness key points along the first and / or second connecting lines of each suspected flame sub-region in the suspected flame image is calculated. For ease of description, the mean brightness value of all brightness key points along the first and / or second connecting lines is referred to as the first brightness mean. It is understood that the mean brightness value of the m brightness key points along the first connecting line of each suspected flame sub-region can be used as the first brightness mean, or the mean brightness value of the m brightness key points along the second connecting line of each suspected flame sub-region can be used as the first brightness mean, or the mean brightness value of the 2m brightness key points along the first and second connecting lines of each suspected flame sub-region can be used as the first brightness mean. Based on the brightness value of each brightness key point along the first and / or second connecting lines of each suspected flame sub-region and the first brightness mean, the first brightness standard deviation is calculated. Understandably, the above method can be used to calculate the N first brightness standard deviations corresponding to the N suspected flame sub-regions in the suspected flame image. The local brightness uniformity of the suspected flame image is then calculated based on these N first brightness standard deviations. For example, if the average brightness value of 2m brightness key points on the first and second lines connecting each suspected flame sub-region is taken as the first brightness mean, then the local brightness uniformity ω of the i-th suspected flame image in the suspected flame image sequence can be expressed as: Among them, L ij_k L represents the brightness value of the k-th brightness key point in the j-th suspected flame sub-region of the i-th suspected flame image in a suspected flame image sequence. ij_a This represents the first brightness mean of the j-th suspected flame sub-region in the i-th suspected flame image in the suspected flame image sequence.

[0051] S230. Determine the uniformity of edge vibration amplitude change and brightness amplitude change of the suspected flame image sequence.

[0052] Analysis of multiple consecutive frames of real flame images revealed that the flame's shape changes in almost every frame, which can be interpreted as the vibration of the flame tongue. Furthermore, the brightness and saturation of the flame in the same area across different frames also exhibit a certain regularity. Therefore, this study analyzes the changes in the edge contours of suspected flame regions between suspected flame images in a suspected flame image sequence to determine the uniformity of edge vibration amplitude changes. This uniformity reflects the uniformity of the undulations in the edge contours of suspected flame regions between suspected flame images in the suspected flame image sequence. Similarly, analysis of the brightness changes in suspected flame regions between suspected flame images in a suspected flame image sequence determines the uniformity of brightness amplitude changes. This uniformity reflects the uniformity of brightness changes in suspected flame regions between suspected flame images in the suspected flame image sequence.

[0053] Optionally, determining the uniformity of edge vibration amplitude variation in the suspected flame image sequence includes: determining the mean of the second maximum vibration amplitude of the maximum key points and the mean of the second minimum vibration amplitude of the minimum key points in the suspected flame sub-regions at the same corresponding position of all suspected flame images in the suspected flame image sequence; calculating the standard deviation of the second maximum vibration amplitude based on the maximum vibration amplitude of the maximum key points and the mean of the second maximum vibration amplitude in the suspected flame sub-regions at the same corresponding position of each suspected flame image in the suspected flame image sequence, and calculating the standard deviation of the second minimum vibration amplitude based on the minimum vibration amplitude of the minimum key points and the mean of the second minimum vibration amplitude in the suspected flame sub-regions at the same corresponding position of each suspected flame image in the suspected flame image sequence; and calculating the uniformity of edge vibration amplitude variation in the suspected flame image sequence based on the standard deviation of the second maximum vibration amplitude and the standard deviation of the second minimum vibration amplitude.

[0054] For example, the mean of the second maximum vibration amplitude of the maximum keypoints and the mean of the second minimum vibration amplitude of the minimum keypoints in the suspected flame sub-regions at the same corresponding position in all suspected flame images in the suspected flame image sequence are calculated. It can be understood that the mean of the second maximum vibration amplitude is the average of the maximum vibration amplitudes of the M maximum keypoints in the suspected flame sub-regions at the same corresponding position in the suspected flame image sequence, and the mean of the second minimum vibration amplitude is the average of the maximum vibration amplitudes of the M minimum keypoints in the suspected flame sub-regions at the same corresponding position in the suspected flame image sequence. For example, the mean of the second maximum vibration amplitude of all j-th suspected flame sub-regions in the suspected flame image sequence can be expressed as σ. hj The mean of the second minimum vibration amplitude of all j-th suspected flame sub-regions in the suspected flame image sequence can be expressed as σ. ljThe suspected flame sub-regions at the same corresponding position in the suspected flame image sequence are grouped into a set of suspected flame sub-regions. For each group of suspected flame sub-regions, the standard deviation of the second maximum vibration amplitude is calculated based on the mean of the maximum and second maximum vibration amplitudes of each maximum keypoint in the group. Similarly, the standard deviation of the second minimum vibration amplitude is calculated based on the mean of the minimum and second minimum vibration amplitudes of each minimum keypoint in the group. It can be understood that, following the above method, N standard deviations of the second maximum vibration amplitude and N standard deviations of the second minimum vibration amplitude in the suspected flame image sequence can be determined. The average of the N standard deviations of the second maximum vibration amplitude and the N standard deviations of the second minimum vibration amplitude is taken as the uniformity of the edge vibration amplitude variation of the suspected flame image sequence. For example, the uniformity of the edge vibration amplitude variation of the suspected flame image sequence can be expressed as...

[0055] Optionally, determining the uniformity of brightness amplitude variation in the suspected flame image sequence includes: determining the second average brightness of all brightness key points on the first and / or second connecting lines of suspected flame sub-regions at the same corresponding position in all suspected flame images in the suspected flame image sequence; calculating a second brightness standard deviation based on the first and second average brightness of the suspected flame sub-regions at the same corresponding position in each suspected flame image in the suspected flame image sequence; and calculating the uniformity of brightness amplitude variation in the suspected flame image sequence based on the second brightness standard deviation.

[0056] In this embodiment of the invention, the second average brightness value of all brightness key points on the first and / or second connecting lines of suspected flame sub-regions at the same corresponding position in all suspected flame image sequences is calculated. For example, the average brightness value of all brightness key points (m*M brightness key points in total) on the first connecting lines of M suspected flame sub-regions at the same corresponding position in the suspected flame image sequence can be used as the second average brightness value; alternatively, the average brightness value of all brightness key points (m*M brightness key points in total) on the second connecting lines of M suspected flame sub-regions at the same corresponding position in the suspected flame image sequence can be used as the second average brightness value; furthermore, the average brightness value of all brightness key points (2m*M brightness key points in total) on the first and second connecting lines of M suspected flame sub-regions at the same corresponding position in the suspected flame image sequence can be used as the second average brightness value. For example, the second average brightness value of all j-th suspected flame sub-regions in the suspected flame image sequence can be represented as L. ij_a_a Understandably, the second average brightness can be calculated using the following formula: Among them, T ij_aLet represent the first mean brightness value of the j-th suspected flame sub-region in the i-th suspected flame image in the suspected flame image sequence. All suspected flame sub-regions at the same corresponding position in the suspected flame image sequence are grouped into a set of suspected flame sub-regions. For each group of suspected flame sub-regions, a second brightness standard deviation is calculated based on the first and second mean brightness values ​​of each suspected flame sub-region within that group. It can be understood that N second brightness standard deviations in the suspected flame image sequence can be determined using the above method. The average of the N second brightness standard deviations is taken as the uniformity of brightness amplitude variation in the suspected flame image sequence. For example, the uniformity of brightness amplitude variation in the suspected flame image sequence can be expressed as:

[0057] S240. Based on the flame edge features, the local area brightness uniformity, the edge vibration amplitude variation uniformity, and the brightness amplitude variation uniformity, detect whether the suspected flame area in the suspected flame image sequence is a real flame area.

[0058] For example, if the flame edge features include the average vibration amplitude difference and edge roundness, then the sum of the average vibration amplitude difference, edge roundness, local area brightness uniformity, edge vibration amplitude variation uniformity, and brightness amplitude variation uniformity can be calculated, and this sum is used as the flame confidence level. It is then determined whether the flame confidence level is greater than a preset threshold. If so, the suspected flame region in the suspected flame image sequence can be determined to be a real flame region; otherwise, the suspected flame region in the suspected flame image sequence is determined not to be a real flame region.

[0059] Optionally, detecting whether a suspected flame region in the suspected flame image sequence is a real flame region based on the flame edge features, the local area brightness uniformity, the edge vibration amplitude variation uniformity, and the brightness amplitude variation uniformity includes: normalizing the flame edge features, the local area brightness uniformity, the edge vibration amplitude variation uniformity, and the brightness amplitude variation uniformity, respectively; calculating the flame confidence level based on the normalized flame edge features, the local area brightness uniformity, the edge vibration amplitude variation uniformity, the brightness amplitude variation uniformity, and the corresponding feature weights; and determining whether the suspected flame region in the suspected flame image sequence is a real flame region based on the flame confidence level.

[0060] For example, the average value α of the average vibration amplitude difference, the average value β of the edge roundness, and the average value ω of the local brightness uniformity are calculated for all suspected flame images in the suspected flame image sequence. The average value α of the average vibration amplitude difference, the average value β of the edge roundness, the average value ω of the local brightness uniformity, and the average value γ of the edge vibration amplitude variation uniformity of the suspected flame image sequence are also calculated. ij and the uniformity of brightness amplitude variation ω i_j Normalization is performed to adjust the value range of each quantity to the range of (0,1). The normalized quantities can be represented as α′, β′, ω′, γ′, and ω0′. In this embodiment, considering the importance of each quantity to flame detection, corresponding feature weights are assigned to each quantity. The flame confidence level is calculated based on the normalized average vibration amplitude difference, edge roundness, local area brightness uniformity, edge vibration amplitude variation uniformity, brightness amplitude variation uniformity, and their corresponding feature weights. The flame confidence level can be expressed as: W = aα′ + bβ′ + cγ′ + dω′ + eω0′, where a, b, c, d, and e represent the feature weights of each quantity, and the sum of a, b, c, d, and e is 1. For example, to reduce the influence of the specificity of certain suspected flame images, the feature weights a, b, and d can be less than 0.3. The system determines whether a suspected flame region in a suspected flame image sequence is a real flame region based on flame confidence level, where the flame confidence level ranges from (0,1). For example, it checks if the flame confidence level is greater than a preset confidence threshold (e.g., 0.8). If so, the suspected flame region in the suspected flame image sequence is determined to be a real flame region; otherwise, it is determined not to be a real flame region. The preset confidence threshold can be set according to the specific application scenario. For example, to improve flame recognition accuracy, the preset confidence threshold can be set to a larger value, such as 0.9; to improve the flame detection rate, the preset confidence threshold can be set to a smaller value, such as 0.6. In this embodiment, when a suspected flame region in a suspected flame image sequence is detected to be a real flame region, a fire is determined to have occurred at the monitoring location corresponding to the suspected flame image sequence, and a fire alarm is triggered.

[0061] The flame detection method of this invention, in response to a flame detection event being triggered, acquires a sequence of suspected flame images; wherein the suspected flame image sequence consists of at least two suspected flame images; the flame edge features and local area brightness uniformity of each suspected flame image in the suspected flame image sequence are determined respectively; the edge vibration amplitude variation uniformity and brightness amplitude variation uniformity of the suspected flame image sequence are determined; based on the flame edge features, the local area brightness uniformity, the edge vibration amplitude variation uniformity, and the brightness amplitude variation uniformity, it is detected whether the suspected flame area in the suspected flame image sequence is a real flame area. Through the technical solution provided by this invention, the dynamic features of flame combustion are mined, which not only solves the problem of false flame detection caused by various light sources such as lamps, effectively improving the detection accuracy of real flames, but also simultaneously supports flame recognition in visible light mode, nighttime scenes, and infrared scenes.

[0062] Figure 6 This is a schematic diagram of a flame detection device provided in an embodiment of the present invention. Figure 6 As shown, the device includes:

[0063] The suspected flame image sequence acquisition module 610 is used to acquire a suspected flame image sequence in response to a flame detection event being triggered; wherein the suspected flame image sequence consists of at least two suspected flame images;

[0064] The flame edge feature determination module 620 is used to determine the flame edge features and local brightness uniformity of each suspected flame image in the suspected flame image sequence.

[0065] The uniformity of change determination module 630 is used to determine the uniformity of edge vibration amplitude change and brightness amplitude change of the suspected flame image sequence.

[0066] The flame detection module 640 is used to detect whether a suspected flame region in the suspected flame image sequence is a real flame region based on the flame edge features, the brightness uniformity of the local area, the edge vibration amplitude variation uniformity, and the brightness amplitude variation uniformity.

[0067] Optionally, the flame edge feature determination module includes:

[0068] The suspected flame sub-region generation unit is used to determine the centroid of the suspected flame region in each suspected flame image in the suspected flame image sequence, and to divide the suspected flame region in the suspected flame image into equal-angle regions with the centroid as the center, thereby generating at least two suspected flame sub-regions.

[0069] The extreme key point determination unit is used to determine the maximum and minimum key points on the edge contour corresponding to each of the suspected flame sub-regions;

[0070] The flame edge feature determination unit is used to determine the flame edge features of the suspected flame image based on the maximum and minimum key points corresponding to the at least two suspected flame sub-regions.

[0071] Optionally, the flame edge feature includes the average vibration amplitude difference;

[0072] The flame edge feature determination unit is used for:

[0073] The maximum vibration amplitude of the maximum key point and the minimum vibration amplitude of the corresponding minimum key point are determined for each of the suspected flame sub-regions, and the vibration amplitude difference between the maximum vibration amplitude and the minimum vibration amplitude is calculated.

[0074] The average vibration amplitude difference of the suspected flame image is calculated based on the vibration amplitude difference corresponding to the at least two suspected flame sub-regions.

[0075] Optionally, the flame edge feature includes edge roundness;

[0076] The flame edge feature determination unit is used for:

[0077] The mean value of the first maximum vibration amplitude of the maximum key point corresponding to at least two suspected flame sub-regions and the mean value of the first minimum vibration amplitude of the corresponding minimum key point are determined respectively.

[0078] The standard deviation of the first maximum vibration amplitude is calculated based on the maximum vibration amplitude of the maximum key point corresponding to the at least two suspected flame sub-regions and the mean of the first maximum vibration amplitude; and the standard deviation of the first minimum vibration amplitude is calculated based on the minimum vibration amplitude of the minimum key point corresponding to the at least two suspected flame sub-regions and the mean of the first minimum vibration amplitude.

[0079] The edge roundness of the suspected flame image is calculated based on the standard deviation of the first maximum vibration amplitude and the standard deviation of the first minimum vibration amplitude.

[0080] Optionally, the uniformity of change determination module is used for:

[0081] The mean of the second maximum vibration amplitude of the maximum key point and the mean of the second minimum vibration amplitude of the minimum key point are determined respectively for the same corresponding position of the suspected flame sub-region in all suspected flame images in the suspected flame image sequence.

[0082] The standard deviation of the second maximum vibration amplitude is calculated based on the maximum vibration amplitude of the maximum key point of the suspected flame sub-region at the same corresponding position in each suspected flame image in the suspected flame image sequence and the mean of the second maximum vibration amplitude; and the standard deviation of the second minimum vibration amplitude is calculated based on the minimum vibration amplitude of the minimum key point of the suspected flame sub-region at the same corresponding position in each suspected flame image in the suspected flame image sequence and the mean of the second minimum vibration amplitude.

[0083] The uniformity of edge vibration amplitude variation in the suspected flame image sequence is calculated based on the standard deviation of the second maximum vibration amplitude and the standard deviation of the second minimum vibration amplitude.

[0084] Optionally, the flame edge feature determination module is used for:

[0085] For each suspected flame image in the suspected flame image sequence, m brightness key points are randomly selected on the first line connecting the maximum key point and the centroid and the second line connecting the corresponding minimum key point and the centroid in each suspected flame sub-region of the suspected flame image.

[0086] For each suspected flame sub-region in the suspected flame image, determine the first brightness mean of all brightness key points on the first and / or second connecting lines of the suspected flame sub-region, and calculate the first brightness standard deviation based on the brightness value of each brightness key point on the first and / or second connecting lines of the suspected flame sub-region and the first brightness mean.

[0087] The local brightness uniformity of the suspected flame image is determined based on the first brightness standard deviation corresponding to the at least two suspected flame sub-regions.

[0088] Optionally, the uniformity of change determination module is used for:

[0089] Determine the second brightness average of all brightness key points on the first and / or second lines connecting the suspected flame sub-regions at the same corresponding position in all suspected flame images in the suspected flame image sequence;

[0090] The second brightness standard deviation is calculated based on the first brightness mean and the second brightness mean of the suspected flame sub-region at the same corresponding position in each suspected flame image in the suspected flame image sequence;

[0091] The uniformity of brightness amplitude variation in the suspected flame image sequence is calculated based on the second brightness standard deviation.

[0092] Optionally, the flame detection module is used for:

[0093] The flame edge features, the local area brightness uniformity, the edge vibration amplitude variation uniformity, and the brightness amplitude variation uniformity are all normalized.

[0094] The flame confidence level is calculated based on the normalized flame edge features, the local area brightness uniformity, the edge vibration amplitude variation uniformity, the brightness amplitude variation uniformity, and the corresponding feature weights.

[0095] Based on the flame credibility, determine whether the suspected flame area in the suspected flame image sequence is a real flame area.

[0096] The flame detection device provided in the embodiments of the present invention can execute the flame detection method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of executing the method.

[0097] Figure 7 A schematic diagram of a server 10, which can be used to implement embodiments of the present invention, is shown. The server is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The server can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0098] like Figure 7 As shown, server 10 includes at least one processor 11 and memory, such as read-only memory (ROM) 12 and random access memory (RAM) 13, communicatively connected to at least one processor 11. The memory stores computer programs executable by at least one processor. Processor 11 can perform various appropriate actions and processes based on the computer program stored in ROM 12 or loaded into RAM 13 from storage unit 18. RAM 13 can also store various programs and data required for the operation of server 10. Processor 11, ROM 12, and RAM 13 are interconnected via bus 14. Input / output (I / O) interface 15 is also connected to bus 14.

[0099] Multiple components in server 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of monitors, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows server 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0100] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as flame detection methods.

[0101] In some embodiments, the flame detection method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on server 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the flame detection method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the flame detection method by any other suitable means (e.g., by means of firmware).

[0102] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0103] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0104] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0105] To provide interaction with the user, the systems and techniques described herein can be implemented on a server having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the server. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0106] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0107] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0108] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0109] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A flame detection method, characterized in that, include: In response to a flame detection event being triggered, a sequence of suspected flame images is acquired; wherein the suspected flame image sequence consists of at least two suspected flame images; The flame edge features and local brightness uniformity of each suspected flame image in the suspected flame image sequence are determined respectively; Determine the uniformity of edge vibration amplitude variation and brightness amplitude variation in the suspected flame image sequence; Based on the flame edge features, the local area brightness uniformity, the edge vibration amplitude variation uniformity, and the brightness amplitude variation uniformity, it is determined whether the suspected flame area in the suspected flame image sequence is a real flame area.

2. The method according to claim 1, characterized in that, Determining the flame edge features of each suspected flame image in the suspected flame image sequence includes: For each suspected flame image in the suspected flame image sequence, the centroid of the suspected flame region in the suspected flame image is determined, and the suspected flame region in the suspected flame image is divided into equal-angle regions with the centroid as the center, generating at least two suspected flame sub-regions; Determine the maximum and minimum key points on the edge contour corresponding to each of the suspected flame sub-regions; The flame edge features of the suspected flame image are determined based on the maximum and minimum key points corresponding to the at least two suspected flame sub-regions.

3. The method according to claim 2, characterized in that, The flame edge characteristics include the average vibration amplitude difference; Based on the maximum and minimum keypoints corresponding to the at least two suspected flame sub-regions, the flame edge features of the suspected flame image are determined, including: The maximum vibration amplitude of the maximum key point and the minimum vibration amplitude of the corresponding minimum key point are determined for each of the suspected flame sub-regions, and the vibration amplitude difference between the maximum vibration amplitude and the minimum vibration amplitude is calculated. The average vibration amplitude difference of the suspected flame image is calculated based on the vibration amplitude difference corresponding to the at least two suspected flame sub-regions.

4. The method according to claim 2, characterized in that, The flame edge features include edge roundness; Based on the maximum and minimum keypoints corresponding to the at least two suspected flame sub-regions, the flame edge features of the suspected flame image are determined, including: The mean value of the first maximum vibration amplitude of the maximum key point corresponding to at least two suspected flame sub-regions and the mean value of the first minimum vibration amplitude of the corresponding minimum key point are determined respectively. The standard deviation of the first maximum vibration amplitude is calculated based on the maximum vibration amplitude of the maximum key point corresponding to the at least two suspected flame sub-regions and the mean of the first maximum vibration amplitude; and the standard deviation of the first minimum vibration amplitude is calculated based on the minimum vibration amplitude of the minimum key point corresponding to the at least two suspected flame sub-regions and the mean of the first minimum vibration amplitude. The edge roundness of the suspected flame image is calculated based on the standard deviation of the first maximum vibration amplitude and the standard deviation of the first minimum vibration amplitude.

5. The method according to claim 2, characterized in that, Determining the uniformity of edge vibration amplitude variation in the suspected flame image sequence includes: The mean of the second maximum vibration amplitude of the maximum key point and the mean of the second minimum vibration amplitude of the minimum key point are determined respectively for the same corresponding position of the suspected flame sub-region in all suspected flame images in the suspected flame image sequence. The standard deviation of the second maximum vibration amplitude is calculated based on the maximum vibration amplitude of the maximum key point of the suspected flame sub-region at the same corresponding position in each suspected flame image in the suspected flame image sequence and the mean of the second maximum vibration amplitude; and the standard deviation of the second minimum vibration amplitude is calculated based on the minimum vibration amplitude of the minimum key point of the suspected flame sub-region at the same corresponding position in each suspected flame image in the suspected flame image sequence and the mean of the second minimum vibration amplitude. The uniformity of edge vibration amplitude variation in the suspected flame image sequence is calculated based on the standard deviation of the second maximum vibration amplitude and the standard deviation of the second minimum vibration amplitude.

6. The method according to claim 2, characterized in that, Determining the local brightness uniformity of each suspected flame image in the suspected flame image sequence includes: For each suspected flame image in the suspected flame image sequence, m brightness key points are randomly selected on the first line connecting the maximum key point and the centroid and the second line connecting the corresponding minimum key point and the centroid in each suspected flame sub-region of the suspected flame image. For each suspected flame sub-region in the suspected flame image, determine the first brightness mean of all brightness key points on the first and / or second connecting lines of the suspected flame sub-region, and calculate the first brightness standard deviation based on the brightness value of each brightness key point on the first and / or second connecting lines of the suspected flame sub-region and the first brightness mean. The local brightness uniformity of the suspected flame image is determined based on the first brightness standard deviation corresponding to the at least two suspected flame sub-regions.

7. The method according to claim 6, characterized in that, Determining the uniformity of brightness amplitude variation in the suspected flame image sequence includes: Determine the second brightness average of all brightness key points on the first and / or second lines connecting the suspected flame sub-regions at the same corresponding position in all suspected flame images in the suspected flame image sequence; The second brightness standard deviation is calculated based on the first brightness mean and the second brightness mean of the suspected flame sub-region at the same corresponding position in each suspected flame image in the suspected flame image sequence; The uniformity of brightness amplitude variation in the suspected flame image sequence is calculated based on the second brightness standard deviation.

8. A flame detection device, characterized in that, include: A suspected flame image sequence acquisition module is used to acquire a suspected flame image sequence in response to a flame detection event being triggered; wherein, the suspected flame image sequence consists of at least two suspected flame images; The flame edge feature determination module is used to determine the flame edge features and local brightness uniformity of each suspected flame image in the suspected flame image sequence. The uniformity of change module is used to determine the uniformity of edge vibration amplitude change and brightness amplitude change of the suspected flame image sequence. The flame detection module is used to detect whether a suspected flame region in the suspected flame image sequence is a real flame region based on the flame edge features, the brightness uniformity of the local area, the edge vibration amplitude variation uniformity, and the brightness amplitude variation uniformity.

9. A server, characterized in that, The server includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the flame detection method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the flame detection method according to any one of claims 1-7.

Images