Method for judging whether combustion improver is on fire or not based on fire flame video
By processing and analyzing the fire scene videos and extracting the flame characteristics, the subjective problem of ignition aids in fire investigations is solved, and scientific and accurate judgment of the nature of the fire is achieved.
Patent Information
- Application Number
- CN202510527583.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-25
- Publication Date
- 2025-08-08
Smart Images
Figure CN120451861A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of fire identification, and in particular to a method for judging whether a fire is caused by an accelerant based on a fire flame video. Background Art
[0002] After a fire accident occurs, scientifically and rationally identifying the cause of the fire and timely and accurately determining the nature of the fire are important tasks of fire investigation. They are of great significance for investigating the responsibility for the fire accident, summarizing the experience and lessons, and safeguarding the safety of people’s lives and property.
[0003] In arson cases, suspects often use accelerants to ignite the fire quickly. Therefore, whether there is an accelerant involved in the fire is an important basis for determining the nature of the fire.
[0004] Currently, fire investigations begin by identifying unusual combustion patterns through on-site inspections, such as irregular "pool fire" patterns, localized variations in carbonization depth, or fire spread rates significantly higher than normal. These patterns may suggest the involvement of liquid accelerants. Furthermore, suspicious containers, residual volatile odors, or signs of separation between the fire source and the fuel further support the hypothesis of artificial use of accelerants. However, such analysis and judgment methods rely on the investigator's experience, are highly subjective, and do not conform to the principles of scientific evidence.
[0005] Considering the shortcomings of existing inspection and identification methods and the problems they cause, it is necessary to propose new methods to supplement the existing methods. Summary of the Invention
[0006] The purpose of the present invention is to provide a method for determining whether the fire is caused by an accelerant based on a video of the fire flame. As a prerequisite for identifying the species of the accelerant, the method can quickly and effectively determine the nature of the case.
[0007] To achieve the above object, the present invention provides the following solutions:
[0008] A method for determining whether a fire is caused by an accelerant based on a fire flame video, comprising:
[0009] Obtain fire scene video, perform contrast enhancement or super-resolution reconstruction based on the shooting conditions, and read the conversion ratio between pixel units and centimeter units;
[0010] Use the frame rate scale to mark the time point of each frame of the converted fire scene video;
[0011] The flame area of each frame of the fire scene video is extracted and refined using the HSV color space and morphological corrosion and expansion methods, and then binarized.
[0012] The flame features are extracted from the binary flame area, and the time series data of the flame features changing with time is determined at each frame time point using the frame rate scale in Matlab. The flame features include: flame height, flame width, flame area, and number of flame corners.
[0013] Determine whether there is an oxidant involved in the combustion at the scene based on the flame characteristic time series data.
[0014] Optionally, the fire scene video is acquired, contrast enhancement or super-resolution reconstruction is performed according to the shooting conditions, and the conversion ratio between pixel units and centimeter units is read, specifically including:
[0015] Get the height and width of known objects in the fire scene video;
[0016] The pixel values in the fire scene video are converted between centimeters and pixels using the known height and width of the object.
[0017] Optionally, the method of extracting the flame area of each frame of the fire scene video using the HSV color space and the erosion and dilation method in morphology and performing binarization specifically includes:
[0018] Use HSV color space to identify and extract flame areas;
[0019] The flame region is refined using erosion and dilation in morphological methods.
[0020] According to the method of the present invention for determining whether the fire is caused by an accelerant based on a fire flame video, the present invention can achieve the following technical effects:
[0021] The method acquires a fire scene video, performs contrast enhancement or super-resolution reconstruction based on the shooting conditions, and reads the conversion ratio between pixel units and centimeter units; extracts the flame area; extracts flame features based on the binarized flame area combined with the time point of each frame; and determines time series data that changes over time based on the flame features; the flame features include: flame height, flame width, flame area, and number of flame corners; determines a situation graph based on the time series data that changes over time; and determines whether the fire was caused by an accelerant based on the situation graph of the flame features changing over time. The present invention determines whether the fire was caused by an accelerant based on the flame features presented in the fire scene video, thereby improving the accuracy and efficiency of case nature determination. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, 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.
[0023] Figure 1 A schematic diagram of a method for analyzing whether a fire is caused by an oxidant provided by the present invention;
[0024] Figure 2 This is a schematic diagram of converting between pixel units and actual centimeter units using objects in a fire scene;
[0025] Figure 3 This is a sample image of the flame area identified using the HSV method;
[0026] Figure 4 This is a time series diagram of the flame height changes of common combustible haystacks, solid combustible nitrocellulose, and liquid combustion aid gasoline.
[0027] Figure 5 This is a time series diagram of the flame width changes of solid combustible nitrocellulose and liquid combustion aid gasoline.
[0028] Figure 6 This is a time series graph of the flame tip angle number of common combustible haystack, solid combustible nitrocellulose, and liquid combustion aid gasoline.
[0029] Figure 7 A diagram showing the characteristics of combustion residues. DETAILED DESCRIPTION
[0030] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.
[0031] The purpose of the present invention is to provide a method for analyzing whether a fire is caused by an accelerant, which can improve the efficiency and accuracy of determining the nature of a case during a fire investigation.
[0032] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.
[0033] like Figure 1 As shown, the present invention provides a method for identifying the species of an oxidant based on flame characteristics, comprising:
[0034] S101, obtain fire scene video. If the clarity is not enough to accurately identify the flame area, pre-processing such as contrast enhancement and super-resolution reconstruction can be considered;
[0035] S102 reads the conversion ratio between pixel unit and centimeter unit;
[0036] The pixel values in the fire scene video are converted between centimeters and pixels using the known height and width of the object.
[0037] S103 uses the frame rate to scale the time point of each frame of video;
[0038] S104 uses the HSV color space and morphological corrosion and expansion methods to extract and refine the flame area of each frame of the fire scene video and perform binarization;
[0039] S104 specifically includes:
[0040] Use HSV color space to identify and extract flame areas;
[0041] After extraction, the flame area is refined using morphological methods such as erosion and dilation.
[0042] Determine whether to use deconvolution to eliminate the influence of light interference based on the video quality, clarity, and focus conditions.
[0043] S105, extracting flame features based on the binary flame area;
[0044] S106 , using the extracted flame features and the time point of each frame of the video scaled by the frame rate, determining how the flame features change over time.
[0045] The following combination Figure 2-Figure 6 To explain:
[0046] After S101 is completed, the height and width values of a known object in the video are used to perform a conversion between pixel values and actual length and width (unit: cm). Figure 2 As shown in the figure, a beam with a height of 3m is used to calibrate the horizontal and vertical lengths. In this example, the horizontal and vertical conversion ratios obtained are both 1 pixel = 0.15cm.
[0047] Figure 3 It can be observed that the method defined by HSV threshold can effectively identify the flame area and flame edge, which is convenient for extracting the time-varying data of the flame specific characteristics to obtain a time series diagram.
[0048] Figure 4-6 Distinguish between ordinary combustibles and violently burning materials (haystacks vs. nitrocellulose / gasoline).
[0049] The video used in the embodiment is a real fire scene video. After obtaining the video, an adaptive player is selected according to the video format.
[0050] Use the videoFile function provided by Matlab2021b to read the video file. Other video playback software can adjust the function accordingly. Then use the framerate function to read the video frame rate. After obtaining the number of frames per second, mark the time point of each frame of the video.
[0051] Measure the size of any landmark in the video (in centimeters). For example, a common floor height is 3 meters. Use the imread function to read the number of pixels occupied by the landmark in the video in both horizontal and vertical directions as 60 pixels. The conversion ratio between pixels and actual size in centimeters is 1 cm = 0.2 pixels.
[0052] The HSV method is used to determine the flame area and extract three typical characteristics: flame height, width, and number of sharp corners. If there is an explosive growth in height and width within 0-5 seconds of ignition, at a rate of 15-20 cm / s, and the width expansion rate is significantly higher than the height expansion rate, and the flame height / width growth rate ratio reaches 3, it can be determined to be a gasoline combustion accelerator.
[0053] The following combination Figure 4-6 Describe in detail how to distinguish between ordinary combustibles and flammable materials.
[0054] The flame height evolution of common combustibles exhibits a typical three-stage curve: slow rise, plateau, and slow fall. Within five seconds of ignition, the flame height increases by only 4 cm / s, with the maximum height remaining stable below 1.5 meters. In contrast, nitrocellulose and gasoline, two highly combustible materials, exhibit explosive growth of 15-20 cm / s in the initial stages of ignition (0-5 seconds), with the maximum growth rate concentrated within the first 10 seconds, and the flame height can exceed 3 meters within 20 seconds.
[0055] The number of flame apexes in haystack combustion remained stable at ≤3, decreasing monotonically as the combustion progressed. However, the number of apexes in highly flammable materials peaked at 5-8 within 10 seconds of ignition and exhibited oscillations with a period of 2-3 seconds. The amplitude of the apex number in gasoline combustion reached 1.6 times that of nitrocellulose, a finding closely related to the differences in turbulent combustion intensity caused by the fuel phase.
[0056] The calculation of the number of flame sharp corners is as follows: First, obtain the set of contour coordinate points of the binary flame area. Each contour consists of a series of ordered boundary points, which are arranged clockwise or counterclockwise to form a closed area. The contour points are discretized and sampled at intervals of 30 points (the 30 pixels in this example image correspond to 5 centimeters. When the shooting equipment and angle change, the number of interval points needs to change with the conversion ratio between centimeter units and pixel units). For example, if a contour has 1000 points, only the points at positions 1, 31, 61, 91... are detected for sharp corners, where the first point is set to the connection between the hypotenuse and the ground in the edge of the flame.
[0057] For each sampling point (denoted as b), take its preceding point a and its succeeding point c, forming three consecutive points abc. Calculate the angle using the following steps: vector ab = ab (the reverse vector from b to a), vector bc = bc (the reverse vector from b to c). Calculate the angular direction (in radians) of each vector ab and bc, and then calculate the difference between the two. If the difference is greater than 1 / 3π (i.e., 60 degrees), point b is considered to have a sharp corner, and a counter is incremented by 1. Repeat this process for all sampling points, and the final accumulated counter value is the number of sharp corners in the flame frame.
[0058] The following combination Figure 4-6 Explain how to distinguish between liquid combustion aids and non-combustible materials (gasoline vs. nitrocellulose).
[0059] Analysis of flame width evolution revealed that gasoline reaches its maximum width (approximately 1.8 times that of nitrocellulose) at the moment of ignition (<1 second), followed by a unique biphasic curve of a steep decline followed by a gradual rise. While nitrocellulose also exhibits width expansion during the initial ignition phase, its curve exhibits a single, gradual growth phase, lacking the rapid contraction characteristic of liquid fuels.
[0060] Comprehensive analysis of the flame height / width growth rate ratio (R_hw): At the moment of ignition (0-1 second), the R_hw of nitrocellulose is ≈ 1, indicating that its height and width grow synchronously; while the R_hw of gasoline is ≈ 3, indicating that its width expansion rate is significantly higher than the height growth. Figure 7 The combustion residue characteristics are as follows: after the liquid combustion aid is burned, there will be no solid flocs floating down, while solid flammable materials such as nitrocellulose will continue to float down along with carbonized flocs with a diameter of 2-5 mm.
[0061] In summary, a three-tiered discrimination system can be established through dynamic flame parameter monitoring: first, screening for intense combustion based on the initial (0-5 seconds) flame height growth threshold (>15 cm / s); second, distinguishing between liquid and solid fuels using the transient R_hw ratio (within 1 second); and finally, confirming the discrimination results by combining the characteristics of the combustion residue. This system organically combines flame morphological parameters (height, width, and number of sharp corners) with material phase characteristics (liquid volatilization, solid decomposition) to improve discrimination accuracy, especially in complex combustion scenarios, where it is important to monitor the mutation characteristics of key parameters within 10 seconds after ignition.
[0062] Specific real-world cases are used as examples for illustration. The above examples are intended only to help understand the method and core concept of the present invention. At the same time, those skilled in the art will appreciate that variations in the specific implementation and scope of application are possible based on the principles of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.
Claims
1. A method for determining whether a fire is caused by an accelerant based on fire flame video analysis, characterized in that: include: Obtain fire scene video and perform pre-processing such as contrast enhancement and super-resolution reconstruction according to the recording conditions; Read the conversion ratio between pixel unit and centimeter unit; Use the frame rate scale to mark the time point of each frame of the converted fire scene video; The flame area of each frame of the fire scene video is extracted and refined using the HSV color space and morphological corrosion and expansion methods, and then binarized. The flame features are extracted from the binary flame area, and the time series data of the flame features changing over time is determined by combining the time point of each frame and the conversion between pixel units and centimeter units. The flame features include: flame height, flame width, flame area, and number of flame corners. The presence of combustion aids is determined based on the time series data of flame characteristics.
2. The method for determining whether a fire is caused by an oxidant based on flame characteristics analysis according to claim 1, characterized in that: The fire scene video is obtained, and pre-processed according to the recording conditions, such as contrast enhancement and super-resolution reconstruction, and conversion between centimeter units and pixel units is performed, specifically including: Get the height and width of known objects in the fire scene video; The pixel values in the fire scene video are converted between centimeters and pixels using the known height and width of the object.
3. The method for determining whether the fire is caused by an accelerant based on fire flame video analysis according to claim 1, characterized in that: The method of using the HSV color space and the erosion and dilation method in morphology to extract the flame area of each frame of the fire scene video and perform binarization specifically includes: Use HSV color space to identify and extract flame areas; The flame region is refined using erosion and dilation in morphological methods.
4. The method for determining whether the fire is caused by an accelerant based on fire flame video analysis according to claim 1, characterized in that: The extracted flame features and the time point of each frame of the video scaled by the frame rate are used to determine the change of flame features over time, that is, the time series data of the flame height, flame width, flame area, and flame tip number over time, including: Using the binary flame area, a matrix is established to store the height, width, and area of the flame area.
5. The method for determining whether the fire is caused by an accelerant based on fire flame video analysis according to claim 4, characterized in that: The number of sharp corners is calculated as follows: first, a set of contour coordinate points of the binary flame area is obtained; each contour is composed of a series of ordered boundary points, which are arranged clockwise or counterclockwise to form a closed area; the contour points are discretized and sampled at intervals of 30 points; For each sampling point (denoted as b), take its previous point a and next point c to form three consecutive points abc; The angle is calculated by the following steps: vector ab = ab (the reverse vector from b to a), vector bc = bc (the reverse vector from b to c); the direction angles (in radians) of vectors ab and bc are calculated respectively, and the difference between the direction angles of the two vectors is calculated. If the angle difference is greater than 1 / 3π (i.e., 60 degrees), it is determined that there is a sharp corner at point b, and the counter is increased by 1. This process is repeated for all sampling points, and the final accumulated counter value is the number of sharp corners in the flame frame; A matrix is established to store the data of the change of flame tip number over time.