Ballistic trajectory detection method and system based on breech trace of bullet

Abstract

The invention relates to the technical field of ballistic trajectory detection, and discloses a ballistic trajectory detection method and system based on a bullet bore trace, and the method comprises the steps: obtaining a starting end region image and a tail end region image of a to-be-detected bullet bore trace according to 360-degree image information, and carrying out the image processing, extracting morphological characteristics and local traces of the wirebore traces in the starting end region image and the tail end region image after image processing; establishing a matching model in advance, and determining the gun type of the to-be-detected bullet based on the matching model and the to-be-detected bullet type; and obtaining morphological characteristics and local traces of the sample bullets of the guns in the gun types, determining the morphological characteristics and the local traces as preset morphological characteristics and preset local traces, and determining the shooting gun of the to-be-detected bullet according to the relationship between the morphological characteristics and the local traces of the rifle traces and the preset morphological characteristics and the preset local traces of the sample bullets. According to the invention, efficient and accurate shooting gun identification is realized, and the reliability of ballistic trajectory detection is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of ballistic trajectory detection, and in particular to a ballistic trajectory detection method and system based on bullet rifling marks. Background Art

[0002] Ballistic trajectory analysis is a key means of evidence identification in modern criminal investigation technology. During the investigation of gun-related cases, accurately identifying the source of the bullet is of great significance for reconstructing the course of the crime and locking down the gun involved.

[0003] At present, in the actual judicial identification process, the traditional comparison method for bullets extracted from the crime scene is to rub the rifling marks of different bullets together, and observe through a microscope to see if there is consistency. However, due to the combined influence of factors such as the slope of the barrel, the rifling structure, and the impact of gunpowder gas on the bullet in the barrel, the rifling marks formed often have slight individual differences, making the comparison results unstable. In some cases, even if it is known that two bullets come from the same gun, the rifling marks may still have certain morphological changes, making it difficult for the identification personnel to make a clear identification conclusion. This uncertainty problem not only increases the complexity of case analysis, but also requires multiple experts to consult together for ballistic trace identification, which is time-consuming and labor-intensive, and some trace differences are still difficult to explain, affecting the objectivity and efficiency of the identification.

[0004] Therefore, it is urgent to invent a ballistic trajectory detection technology to solve the problem that the traditional rifling mark comparison method is affected by various factors in the barrel of the bullet, resulting in slight differences in the traces, which in turn causes unstable comparison results and unclear identification conclusions, increasing the complexity and uncertainty of the identification work. Summary of the invention

[0005] In view of this, the present invention proposes a ballistic trajectory detection method and system based on bullet rifling marks, aiming to solve the problem that the traditional rifling mark comparison method is affected by multiple factors in the bullet inside the barrel, resulting in slight differences in traces, which in turn causes unstable comparison results and unclear identification conclusions, increasing the complexity and uncertainty of the identification work.

[0006] In one aspect, the present invention provides a ballistic trajectory detection method based on bullet rifling marks, comprising:

[0007] Acquire 360° image information of the bullet to be detected based on the scanning module, and acquire the starting area image and the ending area image of the rifle mark of the bullet to be detected according to the 360° image information;

[0008] Perform image processing on the starting region image and the ending region image of the rifling marks of the bullet to be detected, and extract the morphological features and local marks of the rifling marks in the starting region image and the ending region image after image processing.

[0009] Pre - establish a matching model, and determine the type of firearm of the bullet to be detected based on the matching model and the type of the bullet to be detected.

[0010] Obtain the morphological features and local marks of the sample bullets of each firearm in the firearm type, and determine them as the preset morphological features and preset local marks. Determine the firearm that fired the bullet to be detected according to the relationship between the morphological features and local marks of the rifling marks and the preset morphological features and preset local marks of each sample bullet.

[0011] Further, when obtaining the starting region image and the ending region image of the rifling marks of the bullet to be detected according to the 360° image information, it includes:

[0012] Perform unfolding transformation on the bullet surface image based on an image processing algorithm to establish a polar coordinate projection map.

[0013] Extract the rifling marks on the bullet surface based on an edge detection algorithm, and determine the starting region and the ending region of the rifling marks based on the morphological features and gradient changes of the marks.

[0014] Automatically segment the starting region and the ending region based on a region growing algorithm, and extract the starting region image and the ending region image of the rifling marks.

[0015] Further, when performing image processing on the starting region image and the ending region image of the rifling marks of the bullet to be detected, it includes:

[0016] Perform denoising processing on the obtained starting region image and ending region image based on Gaussian filtering.

[0017] For the denoised starting region image and ending region image, enhance the edge features of the rifling marks based on adaptive histogram equalization.

[0018] Extract the local texture features in the edge features of the enhanced rifling marks based on the gray - level co - occurrence matrix.

[0019] Further, when extracting the morphological features and local marks of the rifling marks in the starting region image and the ending region image after image processing, it includes:

[0020] Extract the overall contour and edge features of the rifling marks based on the edge detection algorithm and the starting region image and the ending region image.

[0021] Extract the key local feature points in the starting region image and the ending region image based on the local feature point detection algorithm.

[0022] Based on the overall contour, edge features, and key local feature points of the rifling marks, obtain the morphological features and local marks of the rifling marks, where:

[0023] Convert the spatial features of the extracted rifling marks into frequency domain features based on Fourier transform;

[0024] Perform dimensionality reduction on the extracted morphological features, local marks, and key local feature points based on principal component analysis, and extract representative features;

[0025] Convert the frequency domain features and representative features into feature vectors, and perform data normalization and mean centering on the feature vectors.

[0026] Furthermore, when pre - establishing a matching model, it includes:

[0027] Obtain the bullet types fired by each firearm and establish a bullet type correlation formula;

[0028] Obtain the distance metric between the bullet type correlation formulas, and perform iterative clustering on the bullet type correlation formulas according to the distance metric. According to the clustering results, establish a matching model.

[0029] Furthermore, when determining the firearm type of the bullet to be detected based on the matching model and the bullet type to be detected, it includes:

[0030] Obtain the matching between each bullet type in the matching model and the bullet type of the bullet to be detected;

[0031] Obtain the bullet type consistent with the bullet to be detected, obtain the corresponding firearm types of this bullet type, and determine the firearm types of the bullet to be detected as the firearm types.

[0032] Furthermore, when determining the shooting firearm of the bullet to be detected according to the relationship between the morphological features and local marks of the rifling marks and the preset morphological features and preset local marks of each sample bullet, it includes:

[0033] Obtain the similarity scores between the rifling marks and each sample bullet according to the overlap degree between the morphological features and the preset morphological features and between the local marks and the preset local marks;

[0034] Arrange the similarity scores in descending order, and determine the shooting firearm corresponding to the similarity score ranked first as the shooting firearm of the bullet to be detected.

[0035] Furthermore, when obtaining the similarity scores between the rifling marks and each sample bullet according to the overlap degree between the morphological features and the preset morphological features and between the local marks and the preset local marks, it includes:

[0036] Obtain the morphological feature overlap degree between the morphological features and the preset morphological features, and determine the initial similarity score between the rifling marks and the sample bullet according to the relationship between the morphological feature overlap degree and the first preset morphological feature overlap degree and the second preset morphological feature overlap degree;

[0037] When the morphological feature overlap degree is lower than the first preset morphological feature overlap degree, determine that the initial similarity score is L1;

[0038] When the morphological feature overlap degree is higher than or equal to the first preset morphological feature overlap degree and lower than the second preset morphological feature overlap degree, determine that the initial similarity score is L2;

[0039] When the morphological feature overlap degree is higher than or equal to the second preset morphological feature overlap degree, determine that the initial similarity score is L3;

[0040] Among them, the first preset morphological feature overlap degree is lower than the second preset morphological feature overlap degree, and L1 < L2 < L3.

[0041] Further, when determining that the initial similarity score is Li, i = 1, 2, 3, it includes:

[0042] Obtain the local trace overlap degree between the local traces and the preset local traces, and determine the adjustment coefficient according to the relationship between the local trace overlap degree and the first preset local trace overlap degree and the second preset local trace overlap degree:

[0043] When the local trace overlap degree is lower than the first preset local trace overlap degree, determine that the adjustment coefficient is k1;

[0044] When the local trace overlap degree is higher than or equal to the first preset local trace overlap degree and lower than the second preset local trace overlap degree, determine that the adjustment coefficient is k2;

[0045] When the local trace overlap degree is higher than or equal to the second preset local trace overlap degree, determine that the adjustment coefficient is k3;

[0046] Among them, the first preset local trace overlap degree is lower than the second preset local trace overlap degree, and 0.8 < k1 < k2 < k3 < 1.2;

[0047] Adjust the initial similarity score Li according to the adjustment coefficient, and determine the adjusted initial similarity score as the similarity score between the rifling marks and the sample bullet.

[0048] Compared with the prior art, the beneficial effects of the present invention are as follows: By obtaining 360° image information of the warhead to be detected based on the scanning module and extracting the images of the starting area and the ending area of the rifling marks, compared with the traditional single-view or local sampling method, this method can comprehensively cover all the mark information on the surface of the warhead, ensuring that the obtained mark data is more complete. This all-round image acquisition method can effectively reduce the comparison errors caused by different observation angles or incomplete image acquisition, and improve the stability and repeatability of ballistic mark analysis. Secondly, in terms of image processing, this method processes the starting area and the ending area of the rifling marks respectively, and extracts their morphological features and local marks. Due to the influence of factors such as the transition of the chamfered bore, the engagement of the rifling, and the impact of gunpowder combustion gases when the warhead is in the barrel, the rifling marks may be deformed, blurred or worn. Through algorithms such as image enhancement, edge detection, and feature point extraction, the recognizability of the marks can be effectively improved, ensuring that key morphological information can be accurately extracted during comparison. This feature extraction method based on image processing avoids the recognition deviation caused by human subjective judgment during the traditional microscope observation process, making the detection results more objective and stable. In addition, this method constructs a matching model, using the morphological features and local marks of the rifling marks of the warhead to be detected as input, and combining the known sample data of gun types to achieve efficient gun matching. Compared with the traditional method that relies on manual experience for mark comparison, this matching model can utilize big data training to improve the generalization ability and matching accuracy of the model. Especially for the situation where the barrel has slight wear or manufacturing errors, this method can still accurately identify based on the overall morphological features of the marks, avoiding matching failures caused by local differences and improving the accuracy of gun identification. Finally, through the known sample library of gun types, the morphological features and local marks of the bullets of each gun sample are obtained, and a standardized preset feature database is established. During the gun matching process, the system can automatically compare the rifling marks of the warhead to be detected with the preset features in the sample database, and determine the shooting gun of the warhead to be detected according to the matching similarity. This data matching-based method can reduce the time of expert consultation and improve the efficiency of case analysis. Especially in the case of involving a large number of warhead comparisons, it can achieve fast and batch analysis, improving the processing ability of forensic identification work.

[0049] On the other hand, the present application also provides a ballistic trajectory detection system based on the rifling marks of the warhead, including:

[0050] A scanning module configured to obtain 360° image information of the warhead to be detected;

[0051] The acquisition module is electrically connected to the scanning module. The acquisition module is configured to obtain the starting area image and the ending area image of the rifling marks of the bullet to be detected according to the 360° image information. The acquisition module is further configured to perform image processing on the starting area image and the ending area image of the rifling marks of the bullet to be detected, and extract the morphological features and local marks of the rifling marks in the processed starting area image and ending area image.

[0052] The classification module is electrically connected to the acquisition module. The analysis module is configured to pre-establish a matching model, and determine the gun type of the bullet to be detected based on the matching model and the type of the bullet to be detected. The analysis module is further configured to obtain the morphological features and local marks of the sample bullets of each gun in the gun type, and determine them as the preset morphological features and preset local marks. According to the relationship between the morphological features and local marks of the rifling marks and the preset morphological features and preset local marks of each sample bullet, the shooting gun of the bullet to be detected is determined.

[0053] It can be understood that the above-mentioned embodiments of the present invention for a ballistic trajectory detection method and system based on bullet rifling marks have the same beneficial effects and will not be elaborated here. Description of the Drawings

[0054] By reading the following detailed description of the preferred embodiments, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present invention. And throughout the drawings, the same reference numerals are used to represent the same components. In the drawings:

[0055] Figure 1 It is a flowchart of a ballistic trajectory detection method based on bullet rifling marks provided by an embodiment of the present invention;

[0056] Figure 2 It is a functional block diagram of a ballistic trajectory detection system based on bullet rifling marks provided by an embodiment of the present invention. Detailed Embodiments

[0057] Hereinafter, exemplary embodiments of the present disclosure will be described in more detail with reference to the drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully conveyed to those skilled in the art. It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other. Hereinafter, the present invention will be described in detail with reference to the drawings and in combination with the embodiments.

[0058] As Figure 1As shown, in some embodiments of the present application, this embodiment provides a ballistic trajectory detection method based on the rifling marks of a warhead, including:

[0059] Step S100: Obtain 360° image information of the warhead to be detected based on a scanning module, and obtain the starting region image and the ending region image of the rifling marks of the warhead to be detected according to the 360° image information.

[0060] Specifically, when obtaining the starting region image and the ending region image of the rifling marks of the warhead to be detected according to the 360° image information, it includes: performing an unfolding transformation on the warhead surface image based on an image processing algorithm to establish a polar coordinate projection map; extracting the rifling marks on the warhead surface based on an edge detection algorithm, and determining the starting region and the ending region of the rifling marks based on the morphological characteristics and gradient changes of the marks; automatically segmenting the starting region and the ending region based on a region growing algorithm, and extracting the starting region image and the ending region image of the rifling marks.

[0061] It is understandable that by using an image processing algorithm to perform unfolding transformation on the surface image of the bullet head, a polar coordinate projection map is established. Since the bullet head is a cylindrical or approximately cylindrical structure, directly analyzing its three-dimensional curved surface image may cause distortion and information loss. Therefore, the polar coordinate projection method is adopted to unfold the 360° image information into a two-dimensional plane image, enabling the rifling marks to be presented in a continuous and intuitive manner. This conversion method can not only retain all the trace information on the bullet head surface but also facilitate subsequent image processing and feature extraction. Secondly, after obtaining the unfolded image, this solution further uses an edge detection algorithm to extract the rifling marks on the bullet head surface. The edge detection algorithm can identify areas with relatively drastic changes in grayscale or color in the image, and the rifling marks usually exhibit significant texture changes and gradient changes. Therefore, they can be effectively extracted by methods such as the Sobel operator, Canny edge detection, or Laplace operator. After extracting the complete rifling marks, by combining the morphological features of the marks (such as curvature, width change, etc.) and the trend of gradient change, the starting area and ending area of the rifling marks can be accurately determined. This feature analysis-based method can adapt to the changes in different gun types and bullet head surface characteristics, improving the accuracy of area positioning. Finally, after determining the starting area and ending area of the rifling marks, this solution further introduces a region growing algorithm to automatically segment the starting area and ending area. The region growing algorithm is a method for region division based on pixel similarity. It can start from a predefined seed point and gradually expand to adjacent pixels that meet specific characteristics, thus forming a complete region. In this solution, based on criteria such as the grayscale feature, gradient distribution, or texture consistency of the marks, the starting area image and ending area image of the rifling marks can be automatically segmented. This automated region segmentation method reduces manual intervention, improves processing efficiency, and ensures that the extracted trace regions are complete and accurate, laying a data foundation for subsequent ballistic trajectory comparison.

[0062] Step S200: Perform image processing on the starting area image and ending area image of the rifling marks of the bullet to be detected, and extract the morphological features and local marks of the rifling marks in the processed starting area image and ending area image.

[0063] Specifically, when performing image processing on the starting area image and ending area image of the rifling marks of the bullet to be detected, it includes: performing denoising processing on the obtained starting area image and ending area image based on Gaussian filtering; for the denoised starting area image and ending area image, enhancing the edge features of the rifling marks based on adaptive histogram equalization; extracting the local texture features in the edge features of the enhanced rifling marks based on the gray-level co-occurrence matrix.

[0064] Specifically, when extracting the morphological features and local traces of rifling marks from the starting area image and the ending area image after image processing, it includes: based on the edge detection algorithm and the starting area image and the ending area image, extracting the overall contour and edge features of the rifling marks; based on the local feature point detection algorithm, extracting the key local feature points in the starting area image and the ending area image; based on the overall contour and edge features of the rifling marks and the key local feature points, obtaining the morphological features and local traces of the rifling marks, where: converting the spatial features of the extracted rifling marks into frequency domain features based on Fourier transform; performing dimensionality reduction processing on the extracted morphological features, local traces and key local feature points based on principal component analysis, and extracting representative features; converting the frequency domain features and representative features into feature vectors, and performing data normalization and mean centering processing on the feature vectors.

[0065] It is understandable that Gaussian filtering is used to denoise the image. Gaussian filtering is a smoothing filtering method that can effectively remove random noise in the image while preserving the main edge information of the rifling marks. In addition, to enhance the visibility of the rifling marks, this solution uses adaptive histogram equalization (CLAHE) to enhance the image, making the details of the dark and bright parts more uniform and improving the edge contrast of the rifling marks. Through these image processing techniques, a clearer rifling mark area can be obtained, providing high-quality input data for subsequent feature extraction. Secondly, after denoising and enhancement processing, this solution uses the gray-level co-occurrence matrix (GLCM) to extract the local texture features of the rifling marks. The gray-level co-occurrence matrix is a statistical method for describing image texture that can analyze the spatial relationship between pixel gray values in the image, thereby extracting the detailed information of the rifling marks. By calculating texture parameters such as energy, contrast, and correlation, the local features of the rifling marks can be further quantified and characterized, improving their distinguishability. In addition, these texture features can be used as important parameters for subsequent pattern matching and comparison, making the identification of gun types more accurate. In terms of rifling mark feature extraction, this solution is based on edge detection algorithms and local feature point detection algorithms to extract the overall contour, edge features, and key local feature points of the rifling marks. Edge detection algorithms (such as the Canny operator or Sobel operator) can accurately identify the overall structure of the rifling marks, while local feature point detection (such as SIFT, SURF, etc.) can identify the local morphological features of key positions on the bullet surface. By comprehensively analyzing the overall contour and key feature points, a complete morphological feature model of the rifling marks can be constructed, making the comparison of marks on different bullets more accurate. In addition, this solution uses Fourier transform and principal component analysis (PCA) to optimize the extracted morphological features and local marks. Fourier transform can convert spatial domain features into frequency domain features, making the periodic structure of the rifling marks more obvious and helping to remove redundant information. In addition, PCA is used for dimensionality reduction processing, converting complex feature data into representative features in a low-dimensional space to reduce computational complexity while retaining the main information of the features. Finally, through the normalization and mean centering processing of the feature vectors, it is ensured that the features of different bullets are compared within the same scale range, improving the comparison accuracy and robustness.

[0066] Step S300: Pre-establish a matching model, and determine the gun type of the bullet to be detected based on the matching model and the type of the bullet to be detected.

[0067] Specifically, when pre-establishing the matching model, it includes: obtaining the bullet types fired by each gun and establishing a bullet type correlation formula; obtaining the distance metrics between the bullet type correlation formulas, and performing iterative clustering on the bullet type correlation formulas according to the distance metrics. According to the clustering results, a matching model is established.

[0068] Specifically, when determining the firearm type of the warhead to be detected based on the matching model and the warhead type to be detected, it includes: obtaining the matching between each bullet type in the matching model and the bullet type of the warhead to be detected; obtaining the bullet type consistent with the warhead to be detected, obtaining each firearm type corresponding to this bullet type, and determining each firearm type as the firearm type of the warhead to be detected.

[0069] It can be understood that by collecting the bullet types fired by different firearms, the association relationship between the bullet type and the firearm type is established. Specifically, after each firearm model is fired, its warhead will carry unique rifling trace characteristics, and these trace characteristics can be classified and modeled. For this reason, this solution first establishes the association formula of each bullet type, that is, records the relationship between the bullet type and the corresponding firearm, providing basic data support for subsequent matching. Secondly, when constructing the matching model, this solution further calculates the distance metric between different bullet type association formulas. The distance metric is used to measure the similarity between different bullet types, and the Euclidean distance, Mahalanobis distance or cosine similarity can be calculated based on the feature vector. Subsequently, through the method of iterative clustering, the bullet types with similar features are classified into the same category. Clustering algorithms (such as K-means, DBSCAN or hierarchical clustering) can automatically discover the patterns in the data, making the relationship between different bullet types clearer, and then forming a complete matching model. In addition, during the firearm type recognition process, this solution uses the established matching model to match the bullet type of the warhead to be detected. First, extract the morphological characteristics and local traces of the warhead to be detected, and compare them with each bullet type in the matching model. By calculating the similarity score, the system can determine the bullet type closest to the warhead to be detected, that is, find the known bullet type that best matches its characteristics. This process can use machine learning methods (such as support vector machine SVM, deep learning CNN, etc.) to improve the matching accuracy, thereby enhancing the recognition ability of the system. Finally, this solution further determines the firearm type of the warhead to be detected through the matched bullet type. Since the matching model has established the mapping relationship between the bullet type and the firearm model, therefore, only need to find the firearm corresponding to the bullet type that matches the warhead to be detected to determine the possible shooting firearm. This method effectively reduces the matching error and improves the recognition accuracy through multi-level comparison (first matching the bullet type, and then matching the firearm type), so as to more accurately determine from which firearm the warhead to be detected originated.

[0070] Step S400: Obtain the morphological characteristics and local traces of the sample bullets of each firearm in the firearm type, and determine them as the preset morphological characteristics and preset local traces. Determine the shooting firearm of the warhead to be detected according to the relationship between the morphological characteristics and local traces of the rifling trace and the preset morphological characteristics and preset local traces of each sample bullet.

[0071] Specifically, when determining the firearm used to fire the bullet to be detected based on the morphological characteristics of the rifling marks and the relationship between the local marks and the preset morphological characteristics and preset local marks of each sample bullet, it includes: obtaining the similarity scores between the rifling marks and each sample bullet according to the overlap degrees between the morphological characteristics and the preset morphological characteristics and between the local marks and the preset local marks; arranging the similarity scores in descending order, and determining the firearm corresponding to the similarity score ranked first as the firearm used to fire the bullet to be detected.

[0072] Specifically, when obtaining the similarity scores between the rifling marks and each sample bullet according to the overlap degrees between the morphological characteristics and the preset morphological characteristics and between the local marks and the preset local marks, it includes: obtaining the overlap degree of the morphological characteristics between the morphological characteristics and the preset morphological characteristics, and determining the initial similarity score between the rifling marks and the sample bullet according to the relationship between the overlap degree of the morphological characteristics and the first preset overlap degree of the morphological characteristics and the second preset overlap degree of the morphological characteristics; when the overlap degree of the morphological characteristics is lower than the first preset overlap degree of the morphological characteristics, then determining the initial similarity score as L1; when the overlap degree of the morphological characteristics is higher than or equal to the first preset overlap degree of the morphological characteristics and lower than the second preset overlap degree of the morphological characteristics, then determining the initial similarity score as L2; when the overlap degree of the morphological characteristics is higher than or equal to the second preset overlap degree of the morphological characteristics, then determining the initial similarity score as L3; where the first preset overlap degree of the morphological characteristics is lower than the second preset overlap degree of the morphological characteristics, and L1 < L2 < L3.

[0073] Specifically, when determining the initial similarity score as Li, i = 1, 2, 3, it includes: obtaining the overlap degree of the local marks between the local marks and the preset local marks, and determining the adjustment coefficient according to the relationship between the overlap degree of the local marks and the first preset overlap degree of the local marks and the second preset overlap degree of the local marks; when the overlap degree of the local marks is lower than the first preset overlap degree of the local marks, then determining the adjustment coefficient as k1; when the overlap degree of the local marks is higher than or equal to the first preset overlap degree of the local marks and lower than the second preset overlap degree of the local marks, then determining the adjustment coefficient as k2; when the overlap degree of the local marks is higher than or equal to the second preset overlap degree of the local marks, then determining the adjustment coefficient as k3; where the first preset overlap degree of the local marks is lower than the second preset overlap degree of the local marks, and 0.8 < k1 < k2 < k3 < 1.2; adjusting the initial similarity score Li according to the adjustment coefficient, and determining the adjusted initial similarity score as the similarity score between the rifling marks and the sample bullet.

[0074] It is understandable that by obtaining sample bullets of different gun types, extracting their morphological features and local traces, and storing these data as preset features in the database. The morphological features of the sample bullets usually include the overall contour, the depth and width of the rifling marks, etc., while the local traces cover fine friction marks, defects, and personalized markings. These preset features are used for subsequent bullet head comparisons, thus providing a reference benchmark for identifying the shooting gun of the bullet to be detected. During the comparison process, first, the overlap degree between the morphological features of the bullet to be detected and the preset morphological features of the sample bullets is calculated to evaluate their overall similarity. For this purpose, this solution sets two thresholds: the first preset morphological feature overlap degree and the second preset morphological feature overlap degree. If the overlap degree of the morphological features of the bullet to be detected is low (below the first preset threshold), the initial similarity score is low (L1); if the overlap degree is between the first and second thresholds, a medium similarity score (L2) is assigned; when the overlap degree is high (exceeding the second threshold), the similarity score is the highest (L3). This hierarchical scoring method helps to screen out candidate guns with higher morphological feature matching degrees and improve the recognition accuracy. After obtaining the initial similarity score, this solution further adjusts the score by combining the local trace overlap degree. Specifically, two preset thresholds (the first and second preset local trace overlap degrees) are also set for the matching situation of the local traces, and the corresponding adjustment coefficients (k1, k2, k3) are determined according to the matching situation. When the local trace overlap degree is low, the adjustment coefficient is small (k1), causing the final similarity score to decrease; when the local trace overlap degree is high, the adjustment coefficient is large (k3), increasing the similarity score. This multi-level adjustment mechanism ensures that the comparison result can comprehensively consider both the overall morphological features and the tiny details of the local traces, improving the accuracy of gun identification. Finally, the shooting gun of the bullet to be detected is identified based on the calculated similarity score. The similarity scores of all sample bullets and the bullet to be detected are arranged in reverse order, and the gun corresponding to the sample bullet with the highest similarity is determined as the possible shooting gun. This method improves the reliability of the comparison process, reduces the misjudgment rate, and makes the ballistic trajectory detection more scientific and accurate through the dual comparison of morphological features and local traces, combined with hierarchical scoring and dynamic adjustment.

[0075] In the above embodiments, by obtaining 360° image information of the warhead to be detected based on the scanning module and extracting the images of the starting area and the ending area of the rifling marks, compared with the traditional single-view or local sampling methods, this method can comprehensively cover all the mark information on the warhead surface, ensuring that the obtained mark data is more complete. This all-round image acquisition method can effectively reduce the comparison errors caused by different observation angles or incomplete image acquisition, and improve the stability and repeatability of ballistic mark analysis. Secondly, in terms of image processing, this method processes the starting area and the ending area of the rifling marks respectively to extract their morphological features and local marks. Due to the influence of factors such as the transition of the bevel rifling, the engagement of the rifling, and the impact of gunpowder combustion gas in the barrel, the rifling marks may be deformed, blurred or worn. Through algorithms such as image enhancement, edge detection, and feature point extraction, the recognizability of the marks can be effectively improved, ensuring that the key morphological information can be accurately extracted during comparison. This feature extraction method based on image processing avoids the recognition deviation caused by human subjective judgment during the traditional microscope observation, making the detection results more objective and stable. In addition, this method constructs a matching model, taking the morphological features and local marks of the rifling marks of the warhead to be detected as inputs, and combining the known sample data of gun types to achieve efficient gun matching. Compared with the traditional method that relies on manual experience for mark comparison, this matching model can utilize big data training to improve the generalization ability and matching accuracy of the model. Especially for the situation where there are slight wear or manufacturing errors in the barrel, this method can still accurately identify based on the overall morphological features of the marks, avoiding matching failures caused by local differences and improving the accuracy of gun identification. Finally, through the known sample library of gun types, the morphological features and local marks of the bullets of each gun sample are obtained, and a standardized preset feature database is established. During the gun matching process, the system can automatically compare the rifling marks of the warhead to be detected with the preset features in the sample database, and determine the shooting gun of the warhead to be detected according to the matching similarity. This data-based matching method can reduce the time of expert consultation and improve the efficiency of case analysis. Especially in the case of involving a large number of warhead comparisons, it can achieve fast and batch analysis, improving the processing ability of forensic identification work.

[0076] In another preferred manner based on the above embodiments, as Figure 2 shown, this embodiment provides a ballistic trajectory detection system based on the rifling marks of the warhead, including: a scanning module, a collection module, and a classification module.

[0077] Specifically, the scanning module is configured to obtain 360° image information of the warhead to be detected; the acquisition module is electrically connected to the scanning module, and the acquisition module is configured to obtain the start region image and the end region image of the rifling marks of the warhead to be detected according to the 360° image information. The acquisition module is further configured to perform image processing on the start region image and the end region image of the rifling marks of the warhead to be detected, and extract the morphological features and local marks of the rifling marks in the processed start region image and end region image; the classification module is electrically connected to the acquisition module, and the analysis module is configured to pre-establish a matching model, and determine the type of firearm of the warhead to be detected based on the matching model and the type of the warhead to be detected; the analysis module is further configured to obtain the morphological features and local marks of the sample bullets of each firearm in the firearm type, and determine them as preset morphological features and preset local marks, and determine the shooting firearm of the warhead to be detected according to the relationship between the morphological features and local marks of the rifling marks and the preset morphological features and preset local marks of each sample bullet.

[0078] It can be understood that the above-mentioned embodiments of the present invention, a ballistic trajectory detection method and system based on warhead rifling marks, have the same beneficial effects and will not be repeated.

[0079] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0080] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of the flows and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for realizing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 a block or multiple blocks.

[0081] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured product including an instruction device, and the instruction device realizes the functions in the processFigure 1 one process or multiple processes and / or blocks Figure 1 the functions specified in one block or multiple blocks.

[0082] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0083] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: the specific implementation manners of the present invention can still be modified or equivalently replaced, and any modification or equivalent replacement that does not depart from the spirit and scope of the present invention shall be covered by the protection scope of the claims of the present invention.

Claims

1. A ballistic trajectory detection method based on bullet head rifling marks, characterized in that Including: Obtaining 360° image information of the warhead to be detected based on a scanning module, and obtaining the start region image and the end region image of the rifling marks of the warhead to be detected according to the 360° image information; Performing image processing on the start region image and the end region image of the rifling marks of the warhead to be detected, and extracting the morphological features and local marks of the rifling marks in the processed start region image and end region image; Pre-establishing a matching model, and determining the firearm type of the warhead to be detected based on the matching model and the type of the warhead to be detected; Obtaining the morphological features and local marks of the sample bullets of each firearm in the firearm type, and determining them as preset morphological features and preset local marks, and determining the firearm that fired the warhead to be detected according to the relationship between the morphological features and local marks of the rifling marks and the preset morphological features and preset local marks of each sample bullet.

2. The ballistic trajectory detection method based on the rifling marks of the warhead according to claim 1, characterized in that, When obtaining the start region image and the end region image of the rifling marks of the warhead to be detected according to the 360° image information, it includes: Performing unfolding transformation on the warhead surface image based on an image processing algorithm to establish a polar coordinate projection map; Extracting the rifling marks on the warhead surface based on an edge detection algorithm, and determining the start region and the end region of the rifling marks based on the morphological features and gradient changes of the marks; Automatically segmenting the start region and the end region based on a region growing algorithm, and extracting the start region image and the end region image of the rifling marks.

3. The ballistic trajectory detection method based on the rifling marks of the warhead as described in claim 2, wherein When performing image processing on the start region image and the end region image of the rifling marks of the warhead to be detected, it includes: Performing denoising processing on the obtained start region image and end region image based on Gaussian filtering; For the denoised start region image and end region image, enhancing the edge features of the rifling marks based on adaptive histogram equalization; Extracting local texture features in the edge features of the enhanced rifling marks based on a gray-level co-occurrence matrix.

4. The ballistic trajectory detection method based on the rifling marks of the warhead according to claim 3, characterized in that When extracting the morphological features and local marks of the rifling marks in the processed start region image and end region image, it includes: Extracting the overall contour and edge features of the rifling marks based on an edge detection algorithm and the start region image and the end region image; Extracting key local feature points in the start region image and the end region image based on a local feature point detection algorithm; Obtaining the morphological features and local marks of the rifling marks based on the overall contour and edge features of the rifling marks and the key local feature points, where: Converting the spatial features of the extracted rifling marks into frequency domain features based on Fourier transform; Performing dimensionality reduction processing on the extracted morphological features, local marks and key local feature points based on principal component analysis, and extracting representative features; Converting the frequency domain features and representative features into feature vectors, and performing data normalization and mean centering processing on the feature vectors.

5. The ballistic trajectory detection method based on the rifling marks of the warhead as claimed in claim 1, wherein, When pre-establishing a matching model, it includes: Obtaining the bullet types fired by each firearm, and establishing a bullet type correlation formula; Obtaining the distance metrics between the bullet type correlation formulas, and performing iterative clustering on the bullet type correlation formulas according to the distance metrics, and establishing a matching model according to the clustering results.

6. The ballistic trajectory detection method based on the rifling marks of the warhead as claimed in claim 5, wherein When determining the firearm type of the warhead to be detected based on the matching model and the type of the warhead to be detected, it includes: Match each bullet type in the matching model with the bullet type of the bullet to be detected; Obtain the bullet type consistent with the bullet to be detected, obtain each gun type corresponding to the bullet type, and determine each gun type as the gun type of the bullet to be detected.

7. The ballistic trajectory detection method based on the rifling marks of the warhead according to claim 6, wherein, When determining the shooting gun of the bullet to be detected according to the morphological characteristics of the rifling marks and the relationship between the local marks and the preset morphological characteristics and preset local marks of each sample bullet, it includes: Obtain the similarity scores between the rifling marks and each sample bullet according to the overlap degree between the morphological characteristics and the preset morphological characteristics and the overlap degree between the local marks and the preset local marks; Arrange each similarity score in descending order, and determine the shooting gun corresponding to the similarity score ranked first as the shooting gun of the bullet to be detected.

8. The ballistic trajectory detection method based on the rifling marks of the warhead as claimed in claim 7, wherein When obtaining the similarity scores between the rifling marks and each sample bullet according to the overlap degree between the morphological characteristics and the preset morphological characteristics and the overlap degree between the local marks and the preset local marks, it includes: Obtain the overlap degree of the morphological characteristics between the morphological characteristics and the preset morphological characteristics, and determine the initial similarity score between the rifling marks and the sample bullet according to the relationship between the overlap degree of the morphological characteristics and the first preset overlap degree of the morphological characteristics and the second preset overlap degree of the morphological characteristics; When the overlap degree of the morphological characteristics is lower than the first preset overlap degree of the morphological characteristics, then determine the initial similarity score as L1; When the overlap degree of the morphological characteristics is higher than or equal to the first preset overlap degree of the morphological characteristics and lower than the second preset overlap degree of the morphological characteristics, then determine the initial similarity score as L2; When the overlap degree of the morphological characteristics is higher than or equal to the second preset overlap degree of the morphological characteristics, then determine the initial similarity score as L3; Among them, the first preset overlap degree of the morphological characteristics is lower than the second preset overlap degree of the morphological characteristics, and L1 < L2 < L3.

9. The ballistic trajectory detection method based on the rifling marks of the warhead as claimed in claim 8, wherein, When determining the initial similarity score as Li, i = 1, 2, 3, it includes: Obtain the overlap degree of the local marks between the local marks and the preset local marks, and determine the adjustment coefficient according to the relationship between the overlap degree of the local marks and the first preset overlap degree of the local marks and the second preset overlap degree of the local marks: When the overlap degree of the local marks is lower than the first preset overlap degree of the local marks, then determine the adjustment coefficient as k1; When the overlap degree of the local marks is higher than or equal to the first preset overlap degree of the local marks and lower than the second preset overlap degree of the local marks, then determine the adjustment coefficient as k2; When the overlap degree of the local marks is higher than or equal to the second preset overlap degree of the local marks, then determine the adjustment coefficient as k3; Among them, the first preset overlap degree of the local marks is lower than the second preset overlap degree of the local marks, and 0.8 < k1 < k2 < k3 < 1.2; Adjust the initial similarity score Li according to the adjustment coefficient, and determine the adjusted initial similarity score as the similarity score between the rifling marks and the sample bullet.

10. A ballistic trajectory detection system based on bullet rifling marks, which adopts a ballistic trajectory detection method based on bullet rifling marks as described in any one of claims 1-9, is characterized in that, It includes: A scanning module configured to obtain 360° image information of the bullet to be detected; The acquisition module, electrically connected to the scanning module, is configured to obtain the starting region image and the ending region image of the rifling marks of the bullet to be detected according to the 360° image information. The acquisition module is further configured to perform image processing on the starting region image and the ending region image of the rifling marks of the bullet to be detected, and extract the morphological features and local marks of the rifling marks in the processed starting region image and ending region image. The classification module, electrically connected to the acquisition module, is configured to pre-establish a matching model and determine the gun type of the bullet to be detected based on the matching model and the type of the bullet to be detected. The analysis module is further configured to obtain the morphological features and local marks of the sample bullets of each gun in the gun type, and determine them as the preset morphological features and preset local marks. According to the relationship between the morphological features and local marks of the rifling marks and the preset morphological features and preset local marks of each sample bullet, the shooting gun of the bullet to be detected is determined.

Images