Collaborative acquisition and analysis method and device for fire scene melting mark characteristics

Abstract

The invention relates to the technical field of image recognition, in particular to a fire scene melting mark feature collaborative acquisition and analysis method and device. The method comprises the following steps: carrying out image acquisition on a pre-processed melting mark surface to obtain a melting mark surface image containing melting mark appearance characteristics and metallographic structure characteristics; extracting an outer contour line, a transition area boundary line and a grain boundary contour line in the melting mark surface image; assigning the lengths and inclusion areas of the outer contour line, the transition area boundary line and the grain boundary contour line, and calculating to obtain a molten bead characteristic parameter and a grain boundary characteristic parameter; and determining a melting mark type corresponding to the melting mark surface image based on the melting bead characteristic parameters and the grain boundary characteristic parameters. According to the method, the melting mark surface image containing the melting mark appearance characteristics and the metallographic structure characteristics is used as an analysis sample, the melting mark type is judged from different angles based on the extracted multiple characteristic parameters, cooperation and cross validation between different judgment methods are achieved, and the reliability of the qualitative result of the melting mark is further guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of image recognition technology, and in particular to a method and device for collaboratively collecting and analyzing melt mark features at a fire scene. Background Art

[0002] The information disclosed in this background technology section is only intended to enhance the understanding of the overall background of the invention, and should not necessarily be regarded as an admission or any form of suggestion that the information constitutes the prior art already known to a person skilled in the art.

[0003] In actual fire investigations, conductor melt marks are the most numerous and typical melt mark evidence at the fire scene. Melt marks can be divided into three types of melting properties according to the cause of formation: primary short circuit melt marks, secondary short circuit melt marks, and fire melt marks. Primary short circuit melt marks are melting marks formed by electrical faults such as short circuits in conductors under normal environmental conditions, secondary short circuit melt marks are melting marks formed by electrical faults such as short circuits in conductors under fire environmental conditions, and fire melt marks are melting marks formed by conductors being heated and melted in fire. Therefore, accurately characterizing the melting properties of conductor melt marks at the fire scene is the key to scientifically and accurately analyzing and identifying the cause of the fire.

[0004] At present, the qualitative analysis of melt marks is based on GB / T16840.1-2008 "Technical Identification Methods for Physical Evidence of Electrical Fire Traces Part 1: Macroscopic Method" and GB / T16840.4-2021 "Technical Identification Methods for Physical Evidence of Electrical Fire Traces Part 4: Metallographic Analysis Method". The former is qualitatively analyzed by comparing the ratio of the diameter of the molten bead and the diameter of the wire, while the latter is based on the metallographic structure characteristics including the grain morphology. However, the qualitative analysis method of melt marks based on the above standards still has the following problems: (1) There are fuzzy areas in the definition of key parameters. For example, the specific methods of obtaining core parameters such as melt bead diameter, wire diameter, and grain morphology, and the reasonable value range are not clearly explained or there is overlap, which is prone to disputes when understanding and using them. Different investigators may come to different conclusions on the same sample; (2) When using the metallographic method to analyze melt marks, it is necessary to collect metallographic images of the melt mark cross section. Only after the operator accurately prepares the sample and obtains the metallographic image that can be analyzed can the analysis be carried out according to the standard. When the collected metallographic image does not meet the analysis requirements, the subsequent conclusions are inaccurate.

[0005] Although the existing technology uses automatic identification methods to avoid errors caused by subjective judgment of investigators, it still cannot avoid the defects brought by macroscopic methods and metallographic methods: based on the macroscopic method (such as 202011321909.9), only appearance features are used for analysis, and the definition of appearance features is vague and the judgment criteria are based on human definition; based on the metallographic method (such as 202310315581.7), metallographic images of the cross section of the melt mark need to be collected for analysis, and the analysis results are still subject to the strict requirements of metallographic images. In addition, due to the different image standards and positions collected by different analysis methods, it is impossible to comprehensively consider multi-dimensional parameters such as melt bead diameter, wire diameter, grain morphology, etc. in the same image, and it is difficult to achieve cross-validation between different methods, which is not conducive to high-precision identification of melt marks. Summary of the invention

[0006] In order to address the deficiencies in the prior art, the present invention provides a method and device for collaboratively collecting and analyzing melt mark features at a fire scene. By collecting and pre-processing a melt mark surface image that simultaneously shows the appearance features of the melt mark and the metallographic structural features such as the melt mark surface grains, the characteristic parameters of the molten bead and the characteristic parameters of the grain boundary are extracted and calculated, and the melt mark type is analyzed in collaboration with the macroscopic method and the metallographic analysis method, thereby reducing the difficulty of image collection and improving the analysis accuracy.

[0007] In order to achieve the above object, the present invention is implemented through the following technical solutions: In a first aspect, the present invention provides a method for collaboratively collecting and analyzing melt mark characteristics at a fire scene, comprising the following steps: Capturing images of the pre-treated melt mark surface to obtain a melt mark surface image containing melt mark appearance features and metallographic structure features; Extracting outer contour lines, transition zone boundary lines and grain boundary contour lines in the melt mark surface image; Assigning values ​​to the lengths and included areas of the outer contour line, transition zone boundary line and grain boundary contour line, and calculating and obtaining the characteristic parameters of the molten bead and the characteristic parameters of the grain boundary; Based on the melt bead characteristic parameters and the grain boundary characteristic parameters, the melt mark type corresponding to the melt mark surface image is determined.

[0008] Furthermore, the pretreatment includes using a surface corrosive agent to treat the surface of the melt mark to remove combustion residues on the surface, so that the surface of the melt mark shows the appearance characteristics and metallographic structure characteristics of the melt mark.

[0009] Furthermore, a collaborative collection and classification model of melt mark features is established based on the MobileViT neural network to extract the outer contour line, transition zone boundary line and grain boundary contour line in the melt mark surface image.

[0010] Furthermore, before extracting the outer contour line, transition zone boundary line and grain boundary contour line in the melt mark surface image, the method also includes using a melt mark sample image set marked with the melt mark outer contour line, transition zone boundary line and grain boundary contour line to iteratively train the melt mark feature collaborative acquisition and classification model.

[0011] Furthermore, after extracting the outer contour line, transition zone boundary line and grain boundary contour line in the melt mark surface image, the method further includes adding the extracted melt mark surface image to the melt mark sample image set.

[0012] Furthermore, the assignment includes scale marking, pixel counting, chain code or model calculation.

[0013] Furthermore, the characteristic parameters of the molten bead include the projected circumference of the molten bead, the projected area of ​​the molten bead and the diameter of the molten bead, and the characteristic parameters of the grain boundary include the projected area of ​​the grain boundary, the projected circumference of the grain boundary and the grain boundary axis ratio.

[0014] Furthermore, it is characterized in that, based on the characteristic parameters of the molten bead and the characteristic parameters of the grain boundary, determining the type of the molten mark corresponding to the molten mark surface image specifically includes: The diameter of the molten bead is calculated by the projection perimeter and the projection area of ​​the molten bead, and the first molten mark type determination result is obtained according to the corresponding relationship between the ratio of the molten bead diameter to the wire diameter and the molten mark type; The type of grain is determined according to the grain boundary projection area, the grain boundary projection perimeter and the grain boundary axis ratio, and the second melt mark type determination result is obtained according to the corresponding relationship between the grain type and the melt mark type; The proportion of grains with different grain boundary projection areas is calculated, and the third melt mark type determination result is obtained according to the distribution of grain boundary projection areas of different melt mark types; The determination of the melt mark type is a repeated result among the first melt mark type determination result, the second melt mark type determination result and the third melt mark type determination result.

[0015] Furthermore, the correspondence between the ratio of the molten bead diameter to the wire diameter and the melt mark type, the correspondence between the grain type and the melt mark type, and the grain boundary projection area distribution of different melt mark types are obtained by analyzing the melt mark sample set that is labeled with the ratio of the molten bead diameter to the wire diameter, the grain type, the grain boundary projection area distribution, and the melt mark type.

[0016] Further, after determining the type of the melt mark corresponding to the melt mark surface image, the method further includes: Sending the type of the melt mark corresponding to the melt mark surface image to the mobile terminal corresponding to the fire scene staff; Alternatively, the type of melt mark corresponding to the melt mark surface image is displayed in text form.

[0017] In a second aspect, the present invention provides a device for collaboratively collecting and analyzing melt mark characteristics at a fire scene, the device comprising: An acquisition module is used to acquire images of the pre-processed melt mark surface to obtain a melt mark surface image containing melt mark appearance features and metallographic structure features; An extraction module, used to extract the outer contour line, transition zone boundary line and grain boundary contour line in the melt mark surface image; A calculation module, used for assigning values ​​to the lengths and included areas of the outer contour line, the transition zone boundary line and the grain boundary contour line, and calculating and obtaining the characteristic parameters of the molten bead and the characteristic parameters of the grain boundary; A determination module determines the type of melt mark corresponding to the melt mark surface image based on the melt bead characteristic parameters and the grain boundary characteristic parameters.

[0018] The beneficial effects achieved by one or more technical solutions of the present invention are as follows: The collaborative collection and analysis method of melt mark features at fire scenes of the present invention performs multi-dimensional feature extraction and parameter calculation on the melt mark surface image. It no longer relies solely on macroscopic morphological features or microscopic metallographic features. It can obtain multi-dimensional parameters such as molten bead diameter, wire diameter, grain morphology, etc. on the same melt mark surface image and make a comprehensive consideration. The melt mark type is determined from different angles based on the extracted multiple feature parameters, thereby achieving collaboration and cross-validation between different determination methods, and further ensuring the reliability of the qualitative results of the melt mark.

[0019] The fire scene melt mark feature collaborative collection and analysis method of the present invention utilizes the melt mark surface image containing the melt mark appearance features and metallographic structure features as the analysis sample. There is no need to prepare specific melt mark cross-section metallographic images like the traditional metallographic method, which simplifies the sample preparation process, facilitates on-site sample preparation and collection, and improves analysis efficiency.

[0020] The fire scene melt mark feature collaborative collection and analysis device of the present invention can quickly and accurately perform feature collaborative collection and analysis on the copper wire melt marks at the fire scene through a collection module, an extraction module, and a calculation and determination module. The high degree of integration facilitates on-site use, reduces the time for melt mark identification, and reduces the difficulty of melt mark identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0022] Figure 1 A flow chart of a method for collaboratively collecting and analyzing melt mark characteristics at a fire scene provided by an embodiment of the present invention; Figure 2 This is a distribution diagram of the grain boundary projection area in an embodiment of the present invention; Figure 3 A schematic diagram of the internal structure of a device for collaborative collection and analysis of melt mark features at a fire scene provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0024] In actual fire investigations, copper wire melt marks are the most common and typical trace evidence. Accurately identifying copper wire melt marks at the fire scene is the key to scientifically and accurately analyzing and determining the cause of the fire.

[0025] The series of standards on "Technical Identification Methods for Electrical Fire Trace Evidence" provide the commonly used methods for laboratory identification of melt marks, namely macroscopic method, metallographic analysis method and microscopic morphology analysis method. Among them, the macroscopic method uses the appearance characteristics of the melt mark for identification, the metallographic analysis method uses the metallographic structure characteristics of the melt mark cross section for identification, and the microscopic morphology analysis method uses the microscopic morphology of the melt mark cross section for identification.

[0026] In practical applications, the most widely used method of identification of traces in physical evidence identification experiments is to use a combination of macroscopic methods and metallographic analysis methods to identify the melting properties of the melt marks. However, both methods have the defect of being subject to the subjective judgment of fire investigators, and misjudgments or missed judgments often occur. In the prior art, the automatic identification methods of melt marks based on image recognition are mostly single feature recognition methods based on macroscopic methods or metallographic analysis methods. Although this type of method reduces the image of the subjective judgment of fire investigators, it is still subject to the singleness and strict requirements of the image, and it is impossible to achieve collaborative analysis of multiple feature criteria.

[0027] Figure 1 A flow chart of a method for collaboratively collecting and analyzing melt mark characteristics at a fire scene provided by an embodiment of the present invention. Figure 1 As shown, the method for collaborative collection and analysis of melt mark characteristics at a fire scene provided by an embodiment of the present invention includes the following steps: Step 101, collecting images of the pre-treated melt mark surface to obtain a melt mark surface image containing melt mark appearance features and metallographic structure features; Step 102, extracting the outer contour line, transition zone boundary line and grain boundary contour line in the melt mark surface image; Step 103, assigning values ​​to the lengths and included areas of the outer contour line, transition zone boundary line and grain boundary contour line, and calculating and obtaining the characteristic parameters of the molten bead and the characteristic parameters of the grain boundary; Step 104 : determining the type of melt mark corresponding to the melt mark surface image based on the melt bead characteristic parameters and the grain boundary characteristic parameters.

[0028] Through research, it has been confirmed that the grain boundary characteristics of the melt mark surface of copper wires in building fires are consistent with the grain boundary characteristics of the melt mark cross section. Therefore, in the embodiments of the present application, the melt mark surface can be pre-processed to simultaneously show the appearance characteristics and metallographic structure characteristics of the melt mark, and the melt mark surface image can be collected as the analysis object. The outer contour line, transition zone boundary line and grain boundary contour line are further extracted, and the length and included area of ​​the above lines are assigned values, and the melt bead characteristic parameters and grain boundary characteristic parameters are calculated. The macroscopic method is used to determine the melt bead characteristic parameters, and the metallographic method is used to determine the grain boundary characteristic parameters. The determination results of the two methods are combined to determine the type of melt mark corresponding to the melt mark surface image.

[0029] In one or more embodiments of the present invention, the pretreatment includes using a surface corrosive agent to treat the surface of the melt mark to remove combustion residues on the surface so that the surface of the melt mark shows the appearance characteristics and metallographic structure characteristics of the melt mark.

[0030] The treatment method of the prior art is to use alcohol for cleaning, which can only remove the combustion residues on the surface of the melt mark. After the treatment, only the appearance feature can be revealed. When the image is intelligently analyzed, only the macro method result is obtained, and the accuracy is not high. In the implementation of the present invention, the surface corrosive agent is used for treatment, which can not only remove the combustion residues on the surface, but also reveal the appearance features and metallographic structure features of the melt mark. Moreover, the characteristic parameters can be assigned during the image analysis. The judgment results of the macro method and the metallographic method can be obtained simultaneously through image analysis, which can improve the accuracy of the judgment.

[0031] In one or more embodiments of the present invention, a collaborative collection and classification model of melt mark features is established based on the MobileViT neural network to extract the outer contour line, transition zone boundary line and grain boundary contour line in the melt mark surface image.

[0032] The collaborative collection and classification model of melt mark features is mainly composed of convolutional layer (Conv), MobileNetV2 (MV2), mobile vision block (MobileViT Block), global pooling layer (Global Pooling) and fully connected layer (Linear Layer).

[0033] The melt mark image enters the convolution layer (Conv) for initial feature extraction. In this convolution layer (Conv), the melt mark image is transformed into an initial melt mark feature map, and the size changes from 256×256 to 128×128.

[0034] The initial melt mark feature map first enters the MobileNetV2 layer to extract image feature information, and the image size is changed from 128×128 to 64×64 to improve the feature learning ability of the model.

[0035] The initial melt feature map then enters the MobileViT block to extract various qualitative features of the image in a collaborative manner. The initial melt feature map of the MobileViT block is input to a 3×3 convolutional layer (Conv) and a 1×1 convolutional layer (Conv) for local feature modeling; the feature map is then divided into 2×2 pixel windows in each channel, and the pixels at relative positions in each window are expanded and processed by the Transformer; in the Transformer layer, a series of attention mechanisms are used to weight different positions in each channel to emphasize important feature areas and suppress unimportant areas; the feature map processed by the Transformer is adjusted to 256×256 through a 1×1 convolutional block (Conv), and feature fusion is performed through a 3×3 convolutional layer (Conv) to capture local and global information of qualitative features and form the final melt feature map.

[0036] The final melt feature map is transformed from 256×256 to 8×8 through a 1×1 convolution layer (Conv), and then enters the global pooling layer (Global Pooling), and is transformed from 8×8 to 1×C (C is the number of channels). The average value of the feature information contained in each channel of the final melt feature map is calculated, and the high-dimensional final melt feature map is converted into a feature vector of fixed length. The feature vector is then input into the fully connected layer (Linear Layer) and transformed from 1×C to 1×N (N=2, 2 classifications). The transformed 1×N feature vector is converted into classification probability through the activation function, indicating the possibility that the final melt feature map belongs to different categories. For example, the output result of fire melt mark recognition is (0.9, 0.1), indicating that the recognition probability of fire melt marks is 90%.

[0037] In one or more embodiments of the present invention, before extracting the outer contour line, transition zone boundary line and grain boundary contour line in the melt mark surface image, the method also includes using a melt mark sample image set labeled with the melt mark outer contour line, transition zone boundary line and grain boundary contour line to iteratively train the melt mark feature collaborative acquisition and classification model.

[0038] The melt mark sample image data and the corresponding annotated data are sent to the melt mark feature collaborative acquisition and classification model in batches. In each training iteration, the network performs forward propagation based on the input data to obtain the prediction result, calculates the loss value through the loss function, and then uses the optimizer to perform back propagation based on the loss value to update the network parameters. After multiple rounds of training, until the network converges, that is, the loss value no longer decreases significantly or reaches the pre-set training stop condition. By using the melt mark sample image set to iterate the training of the melt mark feature collaborative acquisition and classification model, the segmentation efficiency and accuracy of the melt mark feature collaborative acquisition and classification model are improved. It should be noted that the method of obtaining the melt mark sample image set in the embodiment of the present application can be obtained on the Internet or downloaded from a specific fire protection website, and the embodiment of the present application is not limited to this.

[0039] In one or more embodiments of the present invention, after extracting the outer contour line, transition zone boundary line and grain boundary contour line in the melt mark surface image, the method further includes adding the extracted melt mark surface image to the melt mark sample image set.

[0040] The melt mark surface image with the outer contour line, transition zone boundary line and grain boundary contour line extracted is added to the melt mark sample image set, and the semantic segmentation algorithm is repeatedly trained and iterated to further improve the segmentation efficiency and accuracy of the semantic segmentation algorithm.

[0041] In one or more embodiments of the present invention, the assignment includes scale marking, pixel counting, chain code or model calculation.

[0042] In one or more embodiments of the present invention, the molten bead characteristic parameters include the molten bead projected circumference, the molten bead projected area and the molten bead diameter, and the grain boundary characteristic parameters include the grain boundary projected area, the grain boundary projected circumference and the grain boundary axis ratio.

[0043] In one or more embodiments of the present invention, it is characterized in that, based on the characteristic parameters of the molten bead and the characteristic parameters of the grain boundary, determining the type of the molten mark corresponding to the molten mark surface image specifically includes: The diameter of the molten bead is calculated by the projection perimeter and the projection area of ​​the molten bead, and the first molten mark type determination result is obtained according to the corresponding relationship between the ratio of the molten bead diameter to the wire diameter and the molten mark type; The type of grain is determined according to the grain boundary projection area, the grain boundary projection perimeter and the grain boundary axis ratio, and the second melt mark type determination result is obtained according to the corresponding relationship between the grain type and the melt mark type; The proportion of grains with different grain boundary projection areas is calculated, and the third melt mark type determination result is obtained according to the distribution of grain boundary projection areas of different melt mark types; The determination of the melt mark type is a repeated result among the first melt mark type determination result, the second melt mark type determination result and the third melt mark type determination result.

[0044] In the existing macroscopic method, the type of melt mark is determined by the ratio of the diameter of the melt bead and the diameter of the wire. In the existing metallographic method, the type of melt mark is determined by the area and type of grain. In the former, the ratio ranges of different types overlap, and in the latter, the types of grains are repeated and the description of the grain boundary area is vaguely described as "fine" or "coarse". Therefore, it is difficult to accurately obtain the type of melt mark based on a single judgment result. The macroscopic method is judged by the characteristic parameters of the melt bead, and the metallographic method is judged by the characteristic parameters of the grain boundary. The grain type and the grain boundary area are separated, and the three judgment results of the two methods are combined for judgment, which is conducive to improving the accuracy of the melt mark type.

[0045] In one or more embodiments of the present invention, the correspondence between the ratio of the molten bead diameter to the wire diameter and the melt mark type, the correspondence between the grain type and the melt mark type, and the grain boundary projection area distribution of different melt mark types are obtained by analyzing a melt mark sample set that is labeled with the ratio of the molten bead diameter to the wire diameter, the grain type, the grain boundary projection area distribution, and the melt mark type.

[0046] The above correspondence and grain boundary projection area distribution can be obtained through data grouping, statistics, and correlation analysis, or by constructing a model and training. Accurately obtaining the correspondence between the ratio of the molten bead diameter to the wire diameter and the type of melt mark, the correspondence between the type of grain and the type of melt mark, and the grain boundary projection area distribution of different melt mark types can avoid the ambiguity and overlap of the judgment basis in the existing standards, and improve the feasibility of the melt mark type judgment result. The grain boundary projection area distribution obtained in one embodiment of the present invention is as follows: Figure 2 As shown in the figure, through the distribution of grain boundary projection area, it can be intuitively concluded that the primary short circuit is mainly small cellular crystals, the secondary short circuit is coarse columnar crystals, and the fire melting mark is coarse equiaxed crystals. The specific values ​​of small and coarse are given. Such values ​​are key parameters to improve classification accuracy.

[0047] In one or more embodiments of the present invention, after determining the type of the melt mark corresponding to the melt mark surface image, the method further includes: Sending the type of the melt mark corresponding to the melt mark surface image to the mobile terminal corresponding to the fire scene staff; Alternatively, the type of melt mark corresponding to the melt mark surface image is displayed in text form.

[0048] Compared with the existing melt mark identification method, the fire scene melt mark feature collaborative collection and analysis method provided in the embodiment of the present invention has the following advantages: Through preprocessing, the surface image of the fusion mark containing the appearance characteristics and metallographic structure characteristics of the fusion mark is obtained as the analysis sample, which can be prepared, collected and analyzed on site, effectively improving the efficiency of fusion mark type collection and analysis, shortening the time and steps required for analysis, and reducing the operating burden of on-site personnel; Multi-dimensional feature extraction and parameter calculation are performed on the melt mark surface image, which no longer relies solely on macroscopic morphological features or microscopic metallographic features, and accurately calculates the parameter values ​​required for judgment, achieving coordination and cross-validation between different judgment methods, and further ensuring the reliability of the qualitative results of the melt mark.

[0049] The above is a method embodiment provided in the embodiments of the present application. Based on the same inventive concept, the embodiments of the present application also provide a device for collaborative collection and analysis of melt mark features at a fire scene.

[0050] Figure 3 A schematic diagram of the internal structure of a fire scene melt mark feature collaborative collection and analysis device provided by an embodiment of the present invention. Figure 3 As shown, the fire scene melting mark feature collaborative collection and analysis device provided by the embodiment of the present invention includes: The acquisition module 201 is used to acquire images of the pre-processed melt mark surface to obtain a melt mark surface image containing melt mark appearance features and metallographic structure features; An extraction module 202 is used to extract the outer contour line, transition zone boundary line and grain boundary contour line in the melt mark surface image; A calculation module 203 is used to assign values ​​to the lengths and included areas of the outer contour line, the transition zone boundary line and the grain boundary contour line, and calculate the characteristic parameters of the molten bead and the characteristic parameters of the grain boundary; The determination module 204 determines the type of the melt mark corresponding to the melt mark surface image based on the melt bead characteristic parameters and the grain boundary characteristic parameters.

[0051] The device for collaborative collection and analysis of melt mark features at a fire scene provided by an embodiment of the present invention can quickly and accurately perform collaborative collection and analysis of melt marks of copper wires at a fire scene. It is highly integrated and convenient for on-site use, reducing the time and difficulty of melt mark identification.

[0052] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for collaborative collection and analysis of melt mark characteristics at a fire scene, characterized in that: The following steps are involved: Capturing images of the pre-treated melt mark surface to obtain a melt mark surface image containing melt mark appearance features and metallographic structure features; Extracting outer contour lines, transition zone boundary lines and grain boundary contour lines in the melt mark surface image; Assigning values ​​to the lengths and included areas of the outer contour line, transition zone boundary line and grain boundary contour line, and calculating and obtaining the characteristic parameters of the molten bead and the characteristic parameters of the grain boundary; Based on the melt bead characteristic parameters and the grain boundary characteristic parameters, the melt mark type corresponding to the melt mark surface image is determined.

2. The method for collaborative collection and analysis of melt mark characteristics at a fire scene according to claim 1, characterized in that: The pretreatment includes using a surface corrosive agent to treat the surface of the melt mark, removing combustion residues on the surface, and making the melt mark surface show the melt mark appearance characteristics and metallographic structure characteristics.

3. The method for collaborative collection and analysis of melt mark characteristics at a fire scene according to claim 1, characterized in that: Based on the MobileViT neural network, a collaborative collection and classification model of melt mark features is established to extract the outer contour line, transition zone boundary line and grain boundary contour line in the melt mark surface image; Before extracting the outer contour line, transition zone boundary line and grain boundary contour line in the melt mark surface image, the method also includes using a melt mark sample image set marked with the melt mark outer contour line, transition zone boundary line and grain boundary contour line to iteratively train the melt mark feature collaborative collection and classification model.

4. The method for collaborative collection and analysis of melt mark characteristics at a fire scene according to claim 3, characterized in that: After extracting the outer contour line, transition zone boundary line and grain boundary contour line in the melt mark surface image, the method further includes adding the extracted melt mark surface image to the melt mark sample image set.

5. The method for collaborative collection and analysis of melt mark characteristics at a fire scene according to claim 1, characterized in that: The assignment includes scale marking, pixel counting, chain code or model calculation.

6. The method for collaborative collection and analysis of melt mark characteristics at a fire scene according to claim 1, characterized in that: The characteristic parameters of the molten bead include the projected circumference of the molten bead, the projected area of ​​the molten bead and the diameter of the molten bead, and the characteristic parameters of the grain boundary include the projected area of ​​the grain boundary, the projected circumference of the grain boundary and the axial ratio of the grain boundary.

7. The method for collaborative collection and analysis of melt mark characteristics at a fire scene according to claim 6, characterized in that: Determining the type of fusion mark corresponding to the fusion mark surface image based on the fusion bead characteristic parameters and the grain boundary characteristic parameters, specifically includes: The diameter of the molten bead is calculated by the projection perimeter and the projection area of ​​the molten bead, and the first molten mark type determination result is obtained according to the corresponding relationship between the ratio of the molten bead diameter to the wire diameter and the molten mark type; The type of grain is determined according to the grain boundary projection area, the grain boundary projection perimeter and the grain boundary axis ratio, and the second melt mark type determination result is obtained according to the corresponding relationship between the grain type and the melt mark type; The proportion of grains with different grain boundary projection areas is calculated, and the third melt mark type determination result is obtained according to the distribution of grain boundary projection areas of different melt mark types; The determination of the melt mark type is a repeated result among the first melt mark type determination result, the second melt mark type determination result and the third melt mark type determination result.

8. The method for collaborative collection and analysis of melt mark characteristics at a fire scene according to claim 7, characterized in that: The correspondence between the ratio of the molten bead diameter to the wire diameter and the melt mark type, the correspondence between the grain type and the melt mark type, and the grain boundary projection area distribution of different melt mark types are obtained by analyzing the melt mark sample set that is labeled with the ratio of the molten bead diameter to the wire diameter, the grain type, the grain boundary projection area distribution, and the melt mark type.

9. The method for collaborative collection and analysis of melt mark characteristics at a fire scene according to claim 1, characterized in that: After determining the type of the melt mark corresponding to the melt mark surface image, the method further includes: Sending the type of the melt mark corresponding to the melt mark surface image to the mobile terminal corresponding to the fire scene staff; Alternatively, the type of melt mark corresponding to the melt mark surface image is displayed in text form.

10. A device for collaborative collection and analysis of melt mark characteristics at a fire scene, characterized in that: The device comprises: An acquisition module is used to acquire images of the pre-processed melt mark surface to obtain a melt mark surface image containing melt mark appearance features and metallographic structure features; An extraction module, used to extract the outer contour line, transition zone boundary line and grain boundary contour line in the melt mark surface image; A calculation module, used for assigning values ​​to the lengths and included areas of the outer contour line, the transition zone boundary line and the grain boundary contour line, and calculating and obtaining the characteristic parameters of the molten bead and the characteristic parameters of the grain boundary; A determination module determines the type of melt mark corresponding to the melt mark surface image based on the melt bead characteristic parameters and the grain boundary characteristic parameters.

Images