An event camera-based method for measuring spatiotemporal characteristics of a target plate penetrated by a fragment

Abstract

The application provides a method for measuring space-time characteristics of a fragment penetrating a target plate based on an event camera, and solves the problems that the existing method cannot detect the spatial distribution of the fragment in real time, cannot obtain time information of the fragment movement, or has high cost, small dynamic range of a high-speed camera, large bandwidth, waste of storage and computing resources. The method comprises the following steps: placing at least one event camera behind the target plate, calibrating the event camera to obtain calibration parameters of the event camera; collecting an event stream of the fragment penetrating the target plate in an explosion process of a warhead by using the event camera, and performing noise reduction processing on the collected event stream; detecting the event stream after the noise reduction processing to obtain image coordinates and time information when different fragments pass through the target plate; and obtaining physical space coordinates and time information when the fragments pass through the target plate by using the obtained calibration parameters of the event camera, the image coordinates and the time information, so as to complete the measurement of the space-time characteristics of the fragment penetrating the target plate.

Description

Technical Field

[0001] This invention relates to the measurement of the spatiotemporal characteristics of fragment perforation, specifically to a method for measuring the spatiotemporal characteristics of fragment penetration of a target plate based on an event camera. Background Technology

[0002] On the modern battlefield, fragments generated by the explosion of a warhead are one of the most common destructive elements. Fragments are characterized by high flight speed, wide kill range, and strong penetration ability. Therefore, measuring the spatiotemporal characteristics of fragments is one of the effective methods to assess the destructive effectiveness of a warhead on a target.

[0003] When measuring the spatiotemporal characteristics of fragments, the target plate method is a commonly used contact measurement method. This method involves placing a target plate within the blast radius, allowing fragments generated after the warhead detonation to impact the target plate. The spatial distribution parameters of the fragments are obtained by statistically analyzing the location and number of perforations in the target plate. However, this method cannot detect the spatial distribution of fragments in real time, nor can it obtain the temporal information of fragment motion. Alternatively, imaging methods can be used to measure the spatiotemporal characteristics of fragments. Imaging methods analyze the spatiotemporal information of fragments by capturing images of fragment flight or after impact with a target plate using a high-speed camera. However, this method has drawbacks: high cost, small dynamic range of high-speed cameras leading to large bandwidth limitations, and much of the information in the captured images is redundant, resulting in a significant waste of storage and computing resources. Summary of the Invention

[0004] The purpose of this invention is to solve the technical problems of existing fragment measurement methods, such as the inability to detect the spatial distribution of fragments in real time, the inability to obtain the temporal information of fragment motion, or the high cost, small dynamic range, large bandwidth, and waste of storage and computing resources of high-speed camera methods for photographing fragment flight. The invention provides a method for measuring the spatiotemporal characteristics of fragment penetration into a target plate based on an event camera.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A method for measuring the spatiotemporal characteristics of fragment penetration into a target plate based on an event camera, characterized by the following steps:

[0007] Step 1: Place at least one event camera behind the target plate, calibrate the event camera, and obtain the calibration parameters of the event camera;

[0008] Step 2: Use an event camera to collect the event stream of fragments penetrating the target plate during the explosion of the warhead, and perform noise reduction processing on the collected event stream;

[0009] Step 3: Use a target detection algorithm to detect the event stream after noise reduction to obtain the image coordinates and time information when different fragments pass through the target plate;

[0010] Step 4: Using the calibration parameters of the event camera obtained in Step 1, the image coordinates and time information obtained in Step 3, the physical space coordinates and time information of the fragment when it passes through the target plate are obtained, and the spatiotemporal characteristics of the fragment penetrating the target plate are measured.

[0011] Furthermore, step 1 specifically includes:

[0012] Two event cameras were placed at different observation positions behind the target plate. The two event cameras were calibrated using the multi-camera calibration method to obtain the intrinsic and extrinsic parameters of the two event cameras respectively.

[0013] Step 2 specifically involves using two event cameras to simultaneously acquire event streams of fragments penetrating the target plate during the warhead explosion, and then performing noise reduction processing on the event streams acquired by each event camera.

[0014] Furthermore, step 3 specifically includes the following steps:

[0015] Step 3.1: Use an object detection algorithm to detect one of the denoised event streams within time ΔT, and calculate the correlation between different events within time ΔT according to the following formula. :

[0016]

[0017] Where 1≤i≤N, 1≤j≤N and N represents the total number of events in the denoised event stream within time ΔT. , For the i-th event and the j-th event Spatial distance between them , The i-th event x-axis image coordinates, y-axis image coordinates , The j-th event x-axis image coordinates, y-axis image coordinates , For the i-th event and the j-th event The time distance between them , The i-th event The j-th event Time, for standard deviation for Standard deviation;

[0018] Step 3.2, if Preset probability threshold Then the j-th event Constitutes the i-th event The supporting events, and the j-th event and the i-th event The target event clusters are labeled with the same ID; the centroid of each event cluster with the same target event cluster ID is calculated.

[0019] Step 3.3: Use the tracker of the corresponding event camera to record the centroid of each event cluster obtained in Step 3.2, and mark each event cluster as a different fragment ID;

[0020] Step 3.4: Repeat steps 3.1 to 3.2 to detect the event stream after noise reduction within the new time ΔT. Calculate the centroid of each event cluster with the same target event cluster ID within the new time ΔT, and calculate the distance D between the centroid of each event cluster within the new time ΔT and the centroids of different fragments in the tracker obtained in step 3.3.

[0021] Step 3.5: If the distance D between the centroid of an event cluster and the centroid of a fragment in the tracker within the new time ΔT is less than the distance threshold d, then update the centroid of the fragment in the tracker obtained in Step 3.3 to the centroid of the event cluster; if the distance D between the centroid of an event cluster and the centroids of different fragments in the tracker within the new time ΔT is greater than d, then update the tracker in Step 3.3, mark the event cluster as a new fragment ID and add it to the tracker; return to Step 3.4 until the detection of the event stream after noise reduction processing within all time periods is completed, and obtain the image coordinates and time information of different fragments passing through the target plate in the tracker of the corresponding event camera;

[0022] Step 3.6: Use the methods from Step 3.1 to Step 3.5 to detect the remaining denoised event stream and obtain the image coordinates and time information of different fragments passing through the target plate in the tracker of the remaining event camera;

[0023] Step 4 specifically involves using the intrinsic and extrinsic parameters of the two event cameras obtained in Step 1, and the image coordinates and time information of different fragments passing through the target plate from the trackers of the two event cameras obtained in Steps 3.5 and 3.6, to obtain the physical space coordinates and time information of the fragments passing through the target plate in the world coordinate system and the target plate coordinate system.

[0024] Furthermore, in step 4, the following equations are used to calculate the physical space coordinates of the fragment in the world coordinate system when it passes through the target plate:

[0025] ;

[0026] ;

[0027] in, , These are the x-axis and y-axis image coordinates of the fragment centroid in the tracker of one of the event cameras, respectively. , These are the internal and external parameters of the camera for this event, respectively; , These are the x-axis and y-axis image coordinates of the fragment centroid in the tracker of another event camera at the same time. , These are the intrinsic and extrinsic parameters of the camera for this event. , , These are the x-axis, y-axis, and z-axis coordinates of the fragment in the world coordinate system, respectively.

[0028] The physical space coordinates of the fragment in the target plate coordinate system when it passes through the target plate are calculated using the following formula:

[0029] ;

[0030] in, , These are the x-axis and y-axis coordinates of the fragment in the target plate coordinate system, respectively. , These are the rotation matrix and translation vector from the world coordinate system to the target coordinate system, respectively.

[0031] Furthermore, step 1 specifically includes:

[0032] An event camera is placed behind the target plate. The event camera is calibrated using a calibration method based on straight line features to obtain the conversion coefficients between the image pixels and spatial distance of the event camera.

[0033] Step 3 specifically includes the following steps:

[0034] Step 3.1: Use an object detection algorithm to detect the denoised event stream within time ΔT, and calculate the correlation between different events within time ΔT according to the following formula. :

[0035] ;

[0036] Where 1≤i≤N, 1≤j≤N and N represents the total number of events in the denoised event stream within time ΔT. , For the i-th event and the j-th event Spatial distance between them , The i-th event x-axis image coordinates, y-axis image coordinates , The j-th event x-axis image coordinates, y-axis image coordinates , For the i-th event and the j-th event The time distance between them , The i-th event The j-th event Time, for standard deviation for Standard deviation;

[0037] Step 3.2, if Preset probability threshold Then the j-th event Constitutes the i-th event The supporting events, and the j-th event and the i-th event The target event clusters are labeled with the same ID; the centroid of each event cluster with the same target event cluster ID is calculated.

[0038] Step 3.3: Use the tracker of the event camera to record the centroid of each event cluster obtained in Step 3.2, and mark each event cluster as a different fragment ID;

[0039] Step 3.4: Repeat steps 3.1 to 3.2 to detect the event stream after noise reduction within the new time ΔT. Calculate the centroid of each event cluster with the same target event cluster ID within the new time ΔT, and calculate the distance D between the centroid of each event cluster within the new time ΔT and the centroids of different fragments in the tracker obtained in step 3.3.

[0040] Step 3.5: If the distance D between the centroid of an event cluster and the centroid of a fragment in the tracker within the new time ΔT is less than the distance threshold d, then update the centroid of the fragment in the tracker obtained in Step 3.3 to the centroid of the event cluster; if the distance D between the centroid of an event cluster and the centroids of different fragments in the tracker within the new time ΔT is greater than d, then update the tracker in Step 3.3, mark the event cluster as a new fragment ID and add it to the tracker; return to Step 3.4 until the detection of the event stream after noise reduction processing within all time periods is completed, and obtain the image coordinates and time information of different fragments passing through the target plate in the tracker of the event camera;

[0041] Step 4 specifically involves using the conversion coefficients between the image pixels and spatial distance of the event camera obtained in Step 1, and the image coordinates and time information obtained in Step 3.5, to obtain the physical spatial coordinates and time information of the fragment in the target plate coordinate system when it passes through the target plate.

[0042] Further, in step 4, the physical space coordinates of the fragment in the target plate coordinate system when it passes through the target plate are calculated according to the following formula:

[0043] ;

[0044] in, , These are the x-axis and y-axis image coordinates of the centroid of the fragment in the tracker, respectively. , These represent the x-axis and y-axis image coordinates of the origin of the target coordinate system in the image coordinate system, respectively. , , , respectively, are the x-axis and y-axis coordinates of the fragment in the target plate coordinate system, and m is the conversion coefficient between image pixels and spatial distance.

[0045] Further, in step 3.2, the centroid of each event cluster with the same target event cluster ID is calculated according to the following formula:

[0046] ;

[0047] Where n is the total number of events in an event cluster with the same target event cluster ID. , , These represent the x-axis image coordinates, y-axis image coordinates, and time of the p-th event in an event cluster with the same target event cluster ID. , , These are the x-axis image coordinates, y-axis image coordinates, and time of the centroid of an event cluster with the same target event cluster ID;

[0048] In step 3.4, the distance D between the centroid of each event cluster within the new time interval ΔT and the centroid of different fragments in the tracker obtained in step 3.3 is calculated according to the following formula:

[0049] ;

[0050] in, , These are the x-axis and y-axis image coordinates of the centroid of the event cluster within the new time △T, respectively. , These are the x-axis and y-axis image coordinates of the centroid of the fragment in the tracker, respectively.

[0051] Furthermore, in step 2, a denoising algorithm is used to denoise the event stream, specifically as follows:

[0052] First, calculate the neighborhood event count matrix Q; the size of Q is the same as the size of the spatial neighborhood, which is a 5×5 matrix, and the elements of Q are the number of events generated by each pixel in the spatial neighborhood of the newly arrived event;

[0053] The random noise judgment value S is then calculated using the following formula:

[0054] ;

[0055] in, This represents the Hadamard product of the random noise filter F and the matrix Q. , Describes the norm of 1;

[0056] If S is greater than the noise threshold, the event is retained; otherwise, the event is filtered out.

[0057] The beneficial effects of this invention are:

[0058] 1. This invention provides a method for measuring the spatiotemporal characteristics of fragment penetration into a target plate based on an event camera. This method can acquire the temporal and spatial distribution of the fragments. Since event cameras are significantly cheaper than high-speed cameras, this method is inexpensive. Furthermore, event cameras collect less data, occupy less bandwidth, and save considerable storage and computing resources.

[0059] 2. Unlike traditional cameras, the event camera in this invention does not output images, but outputs an asynchronous event stream based on the brightness changes of each pixel. Therefore, the event camera has advantages such as high dynamic range and high temporal resolution. Due to the large dynamic range of the event camera, there is no problem of image brightness saturation.

[0060] 3. When using one event camera, the present invention can acquire the spatiotemporal characteristics of the fragment in the target coordinate system; when using two event cameras, it can acquire the spatiotemporal characteristics of the fragment in both the target coordinate system and the world coordinate system; when there are multiple target plates, it can conveniently acquire the spatiotemporal characteristics of all fragments that penetrate the target plate in the world coordinate system. Attached Figure Description

[0061] Figure 1 This is a flowchart of a method for measuring the spatiotemporal characteristics of fragments penetrating a target plate based on an event camera, according to the present invention.

[0062] Figure 2 This is a layout diagram of an event camera observation target plate in Embodiment 1 of the present invention;

[0063] Figure 3 This is a layout diagram of the two event camera observation target plates in Embodiment 2 of the present invention.

[0064] Explanation of reference numerals in the attached figures:

[0065] 1-Target plate, 2-Event camera. Detailed Implementation

[0066] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0067] Example 1

[0068] like Figure 1 As shown, a method for measuring the spatiotemporal characteristics of fragment penetration into a target plate based on an event camera includes the following steps:

[0069] Step 1, as follows Figure 2 As shown, an event camera 2 is placed behind the target plate 1. The event camera 2 is calibrated using a calibration method based on straight-line features, obtaining the conversion coefficients between image pixels and spatial distance of the event camera 2; where, Therefore The world coordinate system with the origin as its coordinate system. Therefore The target plate coordinate system is the origin. Therefore The event camera coordinate system is centered at the origin.

[0070] Step 2: Use event camera 2 to collect the event stream of fragments penetrating target plate 1 during the warhead explosion, and use a noise reduction algorithm to denoise the event stream; only focus on the events when fragments penetrate target plate 1. Based on the existing events, determine whether new events are noise events. The specific noise reduction steps are as follows:

[0071] First, calculate the neighborhood event count matrix Q; the size of Q is the same as the size of the spatial neighborhood, which is a 5×5 matrix, and the elements of Q are the number of events generated by each pixel in the spatial neighborhood of the newly arrived event;

[0072] The random noise judgment value S is then calculated using the following formula:

[0073] ;

[0074] in, This represents the Hadamard product of the random noise filter F and the matrix Q. , Describes the norm of 1;

[0075] If S is greater than the noise threshold, the event is retained; otherwise, the event is filtered out.

[0076] Step 3: Use a target detection algorithm to detect the denoised event stream and obtain the image coordinates and time information of different fragments passing through target plate 1; specifically, this includes the following steps:

[0077] Step 3.1: Use an object detection algorithm to detect the denoised event stream within time ΔT, and calculate the correlation between different events within time ΔT according to the following formula. :

[0078] ;

[0079] Where 1≤i≤N, 1≤j≤N and N represents the total number of events in the denoised event stream within time ΔT. , For the i-th event and the j-th event Spatial distance between them , The i-th event x-axis image coordinates, y-axis image coordinates , The j-th event x-axis image coordinates, y-axis image coordinates , For the i-th event and the j-th event The time distance between them , The i-th event The j-th event Time, for standard deviation for Standard deviation;

[0080] Step 3.2, if Preset probability threshold Then the j-th event Constitutes the i-th event The supporting events, and the j-th event and the i-th event The target event cluster ID is marked as the same; otherwise, the j-th event... It does not constitute the i-th event. Supporting events are not tagged with target event cluster IDs; the centroid of each event cluster with the same target event cluster ID is calculated according to the following formula:

[0081] ;

[0082] Where n is the total number of events in an event cluster with the same target event cluster ID. , , These represent the x-axis image coordinates, y-axis image coordinates, and time of the p-th event in an event cluster with the same target event cluster ID. , , These are the x-axis image coordinates, y-axis image coordinates, and time of the centroid of an event cluster with the same target event cluster ID.

[0083] Step 3.3: Each different event cluster represents an event in which different fragments penetrate the target plate 1. The centroid of each event cluster obtained in step 3.2 is recorded using the tracker of the event camera 2, and each event cluster is marked with a different fragment ID and stored in the tracker.

[0084] Step 3.4: Repeat steps 3.1 to 3.2 to detect the noise-reduced event stream within the new time interval ΔT. Calculate the centroid of each event cluster with the same target event cluster ID within the new time interval ΔT, and calculate the distance D between the centroid of each event cluster within the new time interval ΔT and the centroids of different fragments in the tracker obtained in step 3.3 according to the following formula:

[0085] ;

[0086] in, , These are the x-axis and y-axis image coordinates of the centroid of the event cluster within the new time △T, respectively. , These are the x-axis and y-axis image coordinates of the centroid of the fragment in the tracker, respectively.

[0087] Step 3.5: If the distance D between the centroid of an event cluster and the centroid of a fragment in the tracker within the new time ΔT is less than d, then update the centroid of the fragment in the tracker obtained in Step 3.3 to the centroid of the event cluster; if the distance D between the centroid of an event cluster and the centroids of different fragments in the tracker within the new time ΔT is greater than d, then update the tracker in Step 3.3, mark the event cluster as a new fragment ID and add it to the tracker, and the centroid of the new fragment is the centroid of the event cluster; return to Step 3.4 until the detection of the event stream after noise reduction within all time periods is completed, and obtain the image coordinates and time information of different fragments passing through the target plate 1 in the tracker of event camera 2.

[0088] Step 4: Using the conversion coefficient between image pixels and spatial distance of event camera 2 obtained in Step 1, and the image coordinates and time information obtained in Step 3.5, obtain the physical spatial coordinates and time information of the fragment in the target plate coordinate system when it passes through the target plate 1, thus completing the measurement of the spatiotemporal characteristics of the fragment penetrating the target plate 1. Calculate the physical spatial coordinates of the fragment in the target plate coordinate system when it passes through the target plate 1 according to the following formula:

[0089] ;

[0090] in, , The origin of the target plate coordinate system is respectively (In this embodiment, the origin) The x-axis and y-axis image coordinates of the lower left corner of target plate 1 in the image coordinate system are shown below. , , , respectively, are the x-axis and y-axis coordinates of the fragment in the target plate coordinate system, and m is the conversion coefficient between image pixels and spatial distance.

[0091] Example 2

[0092] This embodiment of the method for measuring the spatiotemporal characteristics of fragment penetration into a target plate based on an event camera includes the following steps:

[0093] Step 1, as follows Figure 3 As shown, two event cameras 2 are placed at different observation positions behind the target plate 1. A multi-camera calibration method is used to calibrate the two event cameras 2, obtaining their intrinsic and extrinsic parameters respectively; among which, Therefore One of the event camera coordinate systems with the origin as its coordinate system. Therefore Let p1 and p2 be the image points of spatial point p in the two event camera images, with p1 and p2 being the image points of spatial point p in the two event camera images, respectively.

[0094] Step 2: Simultaneously acquire event streams of fragments penetrating target plate 1 during the warhead explosion using two event cameras 2, and apply noise reduction algorithms to the event streams acquired by each event camera 2; only focus on events where fragments penetrate target plate 1. Based on existing events, determine whether new events are noise events; the specific noise reduction steps are as follows:

[0095] First, calculate the neighborhood event count matrix Q; the size of Q is the same as the size of the spatial neighborhood, which is a 5×5 matrix, and the elements of Q are the number of events generated by each pixel in the spatial neighborhood of the newly arrived event;

[0096] The random noise judgment value S is then calculated using the following formula:

[0097] ;

[0098] in, This represents the Hadamard product of the random noise filter F and the matrix Q. , Describes the norm of 1;

[0099] If S is greater than the noise threshold, the event is retained; otherwise, the event is filtered out.

[0100] Step 3: Use a target detection algorithm to detect the denoised event stream, and obtain the image coordinates and time information of different fragments passing through the target plate 1 in each event camera 2; specifically including the following steps:

[0101] In step 3.1, an object detection algorithm is used to detect one of the denoised event streams within time ΔT, and the correlation between different events within time ΔT is calculated according to the following formula. :

[0102] ;

[0103] Where 1≤i≤N, 1≤j≤N and N represents the total number of events in the denoised event stream within time ΔT. , For the i-th event and the j-th event Spatial distance between them , The i-th event x-axis image coordinates, y-axis image coordinates , The j-th event x-axis image coordinates, y-axis image coordinates , For the i-th event and the j-th event The time distance between them , The i-th event The j-th event Time, for standard deviation for The standard deviation.

[0104] Step 3.2, if Preset probability threshold Then the j-th event Constitutes the i-th event The supporting events, and the j-th event and the i-th event The target event cluster ID is marked as the same; otherwise, the j-th event... It does not constitute the i-th event. Supporting events are not tagged with target event cluster IDs; the centroid of each event cluster with the same target event cluster ID is calculated according to the following formula:

[0105] ;

[0106] Where n is the total number of events in an event cluster with the same target event cluster ID. , , These represent the x-axis image coordinates, y-axis image coordinates, and time of the p-th event in an event cluster with the same target event cluster ID. , , These are the x-axis image coordinates, y-axis image coordinates, and time of the centroid of an event cluster with the same target event cluster ID.

[0107] Step 3.3: Use the tracker of the corresponding event camera 2 to record the centroid of each event cluster obtained in step 3.2, and mark each event cluster with a different fragment ID and store it in the tracker.

[0108] Step 3.4: Repeat steps 3.1 to 3.2 to detect the noise-reduced event stream within the new time interval ΔT. Calculate the centroid of each event cluster with the same target event cluster ID within the new time interval ΔT, and calculate the distance D between the centroid of each event cluster within the new time interval ΔT and the centroids of different fragments in the tracker obtained in step 3.3 according to the following formula:

[0109] ;

[0110] in, , These are the x-axis and y-axis image coordinates of the centroid of the event cluster within the new time △T, respectively. , These are the x-axis and y-axis image coordinates of the centroid of the fragment in the tracker, respectively.

[0111] Step 3.5: If the distance D between the centroid of an event cluster and the centroid of a fragment in the tracker within the new time ΔT is less than the distance threshold d, then update the centroid of the fragment in the tracker obtained in Step 3.3 to the centroid of the event cluster; if the distance D between the centroid of an event cluster and the centroids of different fragments in the tracker within the new time ΔT is greater than d, then update the tracker in Step 3.3, mark the event cluster as a new fragment ID and add it to the tracker; return to Step 3.4 until the detection of the event stream after noise reduction processing within all time periods is completed, and obtain the image coordinates and time information of different fragments passing through the target plate 1 in the tracker of the corresponding event camera 2.

[0112] Step 3.6: Use the methods in Steps 3.1 to 3.5 to detect the other noise-reduced event stream and obtain the image coordinates and time information of different fragments passing through the target plate 1 in the tracker of the other event camera 2.

[0113] Step 4: Using the intrinsic and extrinsic parameters of the two event cameras 2 obtained in Step 1, and the image coordinates and time information of different fragments passing through the target plate 1 from the trackers of the two event cameras 2 obtained in Steps 3.5 and 3.6, obtain the physical space coordinates and time information of the fragments passing through the target plate 1 in the world coordinate system and the target plate coordinate system, and complete the measurement of the spatiotemporal characteristics of the fragments penetrating the target plate 1. Simultaneously solve the following equations to calculate the physical space coordinates of the fragments passing through the target plate 1 in the world coordinate system:

[0114] ;

[0115] ;

[0116] in, , These are the x-axis and y-axis image coordinates of the fragment centroid in the tracker of one of the events, camera 2. , These are the intrinsic and extrinsic parameters of camera 2 for this event; , These are the x-axis and y-axis image coordinates of the fragment centroid in the tracker of camera 2, which represents another event at the same time. , These are the intrinsic and extrinsic parameters of camera 2 for this event. , , These are the x-axis, y-axis, and z-axis coordinates of the fragment in the world coordinate system, respectively. Then, the physical space coordinates of the fragment in the target plate coordinate system when it passes through target plate 1 are calculated using the following formula:

[0117] ;

[0118] in, , These are the rotation matrix and translation vector from the world coordinate system to the target coordinate system, respectively.

[0119] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions within the technical scope disclosed in the present invention should be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for measuring the spatiotemporal characteristics of fragment penetration into a target plate based on an event camera, characterized in that, Includes the following steps: Step 1: Place at least one event camera (2) behind the target plate (1), calibrate the event camera (2), and obtain the calibration parameters of the event camera (2); Step 1 is as follows: Two event cameras (2) were placed at different observation positions behind the target plate (1). The two event cameras (2) were calibrated using the multi-camera calibration method to obtain the intrinsic and extrinsic parameters of the two event cameras (2). Step 2 specifically involves: using two event cameras (2) to simultaneously collect event streams of fragments penetrating the target plate (1) during the explosion of the warhead, and performing noise reduction processing on the event streams collected by each event camera (2); Step 2: Use the event camera (2) to collect the event stream of fragments penetrating the target plate (1) during the explosion of the warhead, and perform noise reduction processing on the collected event stream; Step 3: Use the target detection algorithm to detect the event stream after noise reduction, and obtain the image coordinates and time information when different fragments pass through the target plate (1); Step 3 specifically includes the following steps: Step 3.1: Use an object detection algorithm to detect one of the denoised event streams within time ΔT, and calculate the correlation between different events within time ΔT according to the following formula. : ； Where 1≤i≤N, 1≤j≤N and N represents the total number of events in the denoised event stream within time ΔT. , For the i-th event and the j-th event Spatial distance between them , The i-th event x-axis image coordinates, y-axis image coordinates , The j-th event x-axis image coordinates, y-axis image coordinates , For the i-th event and the j-th event The time distance between them , The i-th event The j-th event Time, for standard deviation for Standard deviation; Step 3.2, if Preset probability threshold Then the j-th event Constitutes the i-th event The supporting events, and the j-th event and the i-th event The target event clusters are labeled with the same ID; the centroid of each event cluster with the same target event cluster ID is calculated. Step 3.3: Use the tracker of the corresponding event camera (2) to record the centroid of each event cluster obtained in step 3.2, and mark each event cluster as a different fragment ID; Step 3.4: Repeat steps 3.1 to 3.2 to detect the event stream after noise reduction within the new time ΔT. Calculate the centroid of each event cluster with the same target event cluster ID within the new time ΔT, and calculate the distance D between the centroid of each event cluster within the new time ΔT and the centroids of different fragments in the tracker obtained in step 3.

3. Step 3.5: If the distance D between the centroid of an event cluster and the centroid of a fragment in the tracker within the new time △T is less than the distance threshold d, then update the centroid of the fragment in the tracker obtained in Step 3.3 to the centroid of the event cluster; if the distance D between the centroid of an event cluster and the centroids of different fragments in the tracker within the new time △T is greater than d, then update the tracker in Step 3.3, mark the event cluster as a new fragment ID and add it to the tracker; return to Step 3.4 until the detection of the event stream after noise reduction processing in all time periods is completed, and obtain the image coordinates and time information of different fragments passing through the target plate (1) in the tracker of the corresponding event camera (2); Step 3.6: Use the methods in Steps 3.1 to 3.5 to detect the other noise-reduced event stream and obtain the image coordinates and time information of different fragments passing through the target plate (1) in the tracker of the other event camera (2); Step 4: Using the calibration parameters of the event camera (2) obtained in Step 1, the image coordinates and time information obtained in Step 3, the physical space coordinates and time information when the fragment passes through the target plate (1) are obtained, and the spatiotemporal characteristics of the fragment penetrating the target plate (1) are measured. Step 4 specifically involves using the intrinsic and extrinsic parameters of the two event cameras (2) obtained in Step 1, and the image coordinates and time information of different fragments passing through the target plate (1) in the trackers of the two event cameras (2) obtained in Steps 3.5 and 3.6, to obtain the physical space coordinates and time information of the fragments passing through the target plate (1) in the world coordinate system and the target plate coordinate system.

2. The method for measuring the spatiotemporal characteristics of fragment penetration into a target plate based on an event camera according to claim 1, characterized in that, In step 4, the following equations are used to calculate the physical space coordinates of the fragment in the world coordinate system when it passes through the target plate (1): ； ； in, , The x-axis and y-axis image coordinates of the centroid of the fragment in the tracker of one of the event cameras (2) are respectively. , These are the intrinsic and extrinsic parameters of the camera (2) for this event, respectively; , The x-axis and y-axis image coordinates of the centroid of the fragment in the tracker of another event camera (2) at the same time are respectively. , These are the intrinsic and extrinsic parameters of the camera (2) for this event, respectively. , , These are the x-axis, y-axis, and z-axis coordinates of the fragment in the world coordinate system, respectively. The physical space coordinates of the fragment in the target plate coordinate system when it passes through the target plate (1) are calculated according to the following formula: ； in, , These are the x-axis and y-axis coordinates of the fragment in the target plate coordinate system, respectively. , These are the rotation matrix and translation vector from the world coordinate system to the target coordinate system, respectively.

3. The method for measuring the spatiotemporal characteristics of fragment penetration into a target plate based on an event camera according to claim 1, characterized in that: In step 3.2, the centroid of each event cluster with the same target event cluster ID is calculated according to the following formula: ； Where n is the total number of events in an event cluster with the same target event cluster ID. , , These represent the x-axis image coordinates, y-axis image coordinates, and time of the p-th event in an event cluster with the same target event cluster ID. , , These are the x-axis image coordinates, y-axis image coordinates, and time of the centroid of an event cluster with the same target event cluster ID; In step 3.4, the distance D between the centroid of each event cluster within the new time interval ΔT and the centroid of different fragments in the tracker obtained in step 3.3 is calculated according to the following formula: ； in, , These are the x-axis and y-axis image coordinates of the centroid of the event cluster within the new time △T, respectively. , These are the x-axis and y-axis image coordinates of the centroid of the fragment in the tracker, respectively.

4. The method for measuring the spatiotemporal characteristics of fragment penetration into a target plate based on an event camera according to claim 1, characterized in that, In step 2, a noise reduction algorithm is used to denoise the event stream, specifically as follows: First, calculate the neighborhood event count matrix Q; the size of Q is the same as the size of the spatial neighborhood, which is a 5×5 matrix, and the elements of Q are the number of events generated by each pixel in the spatial neighborhood of the newly arrived event; The random noise judgment value S is then calculated using the following formula: ； in, This represents the Hadamard product of the random noise filter F and the matrix Q. , Describes the norm of 1; If S is greater than the noise threshold, the event is retained; otherwise, the event is filtered out.

Images