Collision test measuring device and collision test measuring method

Abstract

To easily measure movement of a vehicle and a collision target object during a collision test.SOLUTION: A collision test measuring device 20 measures movement of a vehicle 11 and a barrier 12 during a collision test in which the vehicle 11 collides with the barrier 12. A storage section 22 stores a video of the vehicle 11 and the barrier 12 captured with a camera 13 during the collision test. A first data acquisition section 24 detects a feature point from a specific still image through image recognition based on shapes of the vehicle 11 and the barrier 12 shown in the specific still image and acquires first data in which coordinates of the feature point are associated with a capture time of the specific still image. A second data acquisition section 25 detects a characteristic point from each of a plurality of subsequent still images by tracking the feature point detected by the first data acquisition section 24 through the image recognition and acquires second data in which the coordinates of the feature point are associated with the captured time of the subsequent still images.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a crash test measuring device and a crash test measuring method. [Background technology]

[0002] In recent years, CAE (Computer Aided Engineering) analysis has been used to evaluate the collision safety performance of vehicles at the design stage of the vehicle (for example, Patent Document 1). Furthermore, vehicle crashworthiness performance is also evaluated through crash tests using actual vehicles. In these crash tests, cameras capture images of the vehicle and the impact object, such as an impactor or a dummy. Numerous detection target locations are preset on the vehicle and the impact object. Coordinates of the same detection target location are extracted from each of a plurality of still images constituting a video captured by the camera. Then, based on the movement of the same detection target location, the movement of the vehicle and the impact object during the crash test, specifically, the deformation of the vehicle, the displacement of the impactor, the deformation of the dummy, etc., are measured and analyzed. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] International Publication No. 2011 / 016499 Summary of the Invention [Problem to be solved by the invention]

[0004] Typically, in a collision test using an actual vehicle, target marks are manually attached or lines or other marks are drawn on the surface of the vehicle or the surface of the object to be hit at positions corresponding to the detection target areas so that the detection target areas can be easily identified.

[0005] Furthermore, the task of extracting the coordinates of the detection target parts from the multiple still images that make up the captured video is also performed manually by specifying the coordinates of the detection target parts while viewing the still images.

[0006] In this manner, in a crash test using an actual vehicle, manual work must be performed on each of the many target parts to be detected, making the crash test extremely time-consuming. [Means for solving the problem]

[0007] Various aspects of the device for solving the above problems will be described. [Aspect 1] A collision test measuring device that measures the movement of a vehicle and a colliding object during a collision test in which the vehicle and the colliding object are collided, the collision test measuring device comprising: a video storage unit that stores video captured by a camera of the vehicle and the colliding object during the collision test; a first data acquisition unit that uses image recognition to detect feature points from a specific still image, which is one of a plurality of still images that constitute the video, based on the shape of the vehicle and the shape of the colliding object that appear in the specific still image, and acquires first data linking the coordinates of the detected feature points in the specific still image to the capture time of the specific still image; and a second data acquisition unit that uses image recognition to track the feature points detected by the first data acquisition unit from each of a plurality of rear still images, which are still images captured after the specific still image, of the plurality of still images, to detect the feature points in each of the rear still images, and acquires second data linking the coordinates of the detected feature points in the rear still image to the capture time of the rear still image.

[0008] According to the above configuration, by using a video of the vehicle and the object being hit during a collision test, it is possible to automatically detect feature points from multiple still images that make up the video through image recognition, track the feature points, and obtain data that links the coordinates of the feature points in the still images with the capture times of the still images. Therefore, it is possible to easily measure the movements of the vehicle and the object being hit during the collision test.

[0009] [Aspect 2] A collision test measuring device as described in [Aspect 1], comprising a display device, and a line data display unit that creates line data indicating the movement trajectory of the characteristic point based on the first data and the second data, and displays the line data on the display device in a superimposed state with the video or one of the multiple still images.

[0010] According to the above configuration, by viewing the moving or still image and the line data displayed on the display device, the movement of the characteristic points during the collision test can be intuitively grasped, thereby enabling the movement of each part of the vehicle and each part of the object to be hit during the collision test to be intuitively grasped.

[0011] [Aspect 3] A collision test measuring device according to [Aspect 1] or [Aspect 2], wherein the first data acquisition unit performs feature point detection to detect the feature points by corner detection and edge detection.

[0012] According to the above configuration, since feature point detection is performed by corner detection, it is possible to detect as feature points portions in the specific still image where angular shapes are displayed. Moreover, since feature point detection is also performed by edge detection, it is possible to detect as feature points portions in the specific still image where shapes indicating edges are displayed. Therefore, it is possible to detect various parts of a vehicle or a collision object as feature points.

[0013] [Embodiment 4] A collision test measurement method for measuring the movement of a vehicle and a collision object during a collision test in which the vehicle and the collision object are collided, the collision test measurement method comprising: a first step of storing a video of the vehicle and the collision object captured by a camera during the collision test in a storage device; a second step of, after the first step, detecting feature points from a specific still image, which is one of a plurality of still images constituting the video stored in the storage device, by image recognition based on the shape of the vehicle and the shape of the collision object captured in the specific still image, and acquiring first data linking the coordinates of the detected feature points in the specific still image to the capture time of the specific still image; and a third step of, after the second step, tracking the feature points detected in the first step from each of a plurality of rear still images, which are still images captured after the specific still image, of the plurality of still images, by image recognition, to detect the feature points in each of the rear still images, and acquiring second data linking the coordinates of the detected feature points in the rear still image to the capture time of the rear still image.

[0014] According to the above method, by using a video of the vehicle and the object being hit during a collision test, it is possible to automatically detect feature points from multiple still images that make up the video through image recognition, track the feature points, and obtain data that links the coordinates of the feature points in the still images with the capture times of the still images. Therefore, it is possible to easily measure the movements of the vehicle and the object being hit during the collision test. [Effects of the Invention]

[0015] According to the present invention, the movement of the vehicle and the object being hit during the crash test can be easily measured. [Brief explanation of the drawings]

[0016] [Figure 1] 1 is a schematic diagram showing the general configuration of a testing site where a crash test of a measurement object is performed. [Figure 2] FIG. 10 is an explanatory diagram for explaining corner detection. [Figure 3] FIG. 10 is an explanatory diagram for explaining edge detection. [Figure 4] 10 is a flowchart showing the execution procedure of a measurement process. [Figure 5] FIG. 10 is a diagram showing a coordinate system for identifying the positions of feature points. [Figure 6] FIG. 10 is a diagram conceptually showing the relationship between specific still image data and the coordinates of feature points. [Figure 7] 10 is a table conceptually showing the data structure of a data list. [Figure 8] FIG. 10 is a diagram showing an example of the display content of a second display area on the display device. DETAILED DESCRIPTION OF THE INVENTION

[0017] An embodiment of a crash test measuring device and a crash test measuring method will be described below. <Test equipment configuration> As shown in FIG. 1, a vehicle 11 to be tested is brought into a testing site 10. Also provided at the testing site 10 are a barrier 12, a camera 13, and a crash test measuring device 20. The barrier 12 is an object that the vehicle 11 collides against during the crash test. The barrier 12 is made of, for example, aluminum honeycomb, and is fixedly installed at the testing site 10. The crash test is performed, for example, by causing the vehicle 11 to crash head-on into the barrier 12 at a predetermined speed.

[0018] The camera 13 photographs the vehicle 11 and the barrier 12 during the crash test. The camera 13 is fixedly installed in the test site 10 so as to photograph an area including the entire vehicle 11 and the entire barrier 12 from the side of the vehicle 11. It is preferable to use a camera 13 with a high frame rate (for example, several hundred to several thousand fps).

[0019] <Collision test measuring device 20> The crash test measuring device 20 includes a control unit 21, a memory unit 22, and a display device 23. The control unit 21 is configured with a processor. The memory unit 22 is a storage device configured with a non-volatile memory that can be written to and read from at any time, such as a hard disk drive or solid state drive. The memory unit 22 stores control programs and data. The crash test measuring device 20 is configured with, for example, a general-purpose PC (personal computer). The crash test measuring device 20 is configured to execute various arithmetic processes based on various programs and data. The crash test measuring device 20 measures the movements of the vehicle 11 and the barrier 12 during the crash test through the arithmetic processes. The crash test measuring device 20 includes, as functional units, a first data acquisition unit 24, a second data acquisition unit 25, and a line data display unit 26. Note that the functional units of the crash test measuring device 20 are not physically separated, but will be described separately for convenience of explanation.

[0020] <Measurement of the movement of vehicle 11 and barrier 12> <Measurement Overview> When a crash test is performed in which the vehicle 11 is caused to crash into the barrier 12, the camera 13 captures images of the vehicle 11 and the barrier 12 during the crash test. The video (more specifically, video data Da) captured by the camera 13 is read into the crash test measuring device 20 and stored in a memory unit 22 of the crash test measuring device 20. In this embodiment, the memory unit 22 corresponds to a video storage unit.

[0021] Here, the video data Da is composed of a plurality of still images (more specifically, still image data Db). The crash test measuring device 20 acquires specific still image data Db1, which is one of the plurality of still image data Db (in this embodiment, the still image data Db captured first). Then, from this specific still image data Db1, image recognition is performed to detect a feature point P based on the shape of the vehicle 11 and the shape of the barrier 12 captured in the specific still image data Db1. The crash test measuring device 20 also acquires first data DT1, which is data in which the coordinates of the detected feature point P in the specific still image data Db1 are linked to the capture time of the specific still image data Db1. In this embodiment, a plurality of (for example, several hundred) feature points P are detected in the specific still image data Db1. Then, the first data DT1 is acquired for each feature point P.

[0022] As an example, the OpenCV (registered trademark) library of the Python (registered trademark) language is used in the image recognition processing in the crash test measuring device 20. OpenCV is an open source library that compiles functional programs such as image recognition processing.

[0023] Furthermore, the feature point detection for detecting the feature points P uses, for example, an algorithm called AKAZE implemented in OpenCV. Specifically, the feature point detection is performed using "corner detection" and "edge detection." As shown in an example in FIG. 2, by performing feature point detection using "corner detection," portions of the specific still image data Db1 where angular shapes are displayed are detected as feature points P. As shown in an example in FIG. 3, by performing feature point detection using "edge detection," portions of the specific still image data Db1 where shapes indicating edges are displayed are detected as feature points P. Examples of portions where shapes indicating edges are the outer edge of the vehicle 11, the outer edge of the barrier 12, and the boundary between adjacent components. In this embodiment, "corner detection" and "edge detection" are used as the feature point detection methods, making it possible to detect various parts of the vehicle 11 and the barrier 12 as feature points P.

[0024] Then, based on the characteristic point P of the specific still image data Db1, the collision test measuring device 20 detects the characteristic point P from each of the multiple still image data Db (hereinafter referred to as subsequent still image data Db2) that was taken after the specific still image data Db1 among the multiple still image data Db.

[0025] The detection of the feature point P from each of the plurality of subsequent still image data Db2 is performed by an optical flow method. More specifically, the feature point P in each of the subsequent still image data Db2 is detected by tracking the feature point P of the specific still image data Db1 in the plurality of subsequent still image data Db2. OpenCV implements an algorithm for realizing detection using the optical flow method. In this embodiment, the detection of the feature point P from each of the plurality of subsequent still image data Db2 is performed using this algorithm.

[0026] The crash test measuring device 20 also acquires second data DT2, which is data in which the coordinates of the feature points P in the rear still image data Db2 are linked to the capture times of the rear still image data Db2. In this embodiment, a plurality of feature points P (e.g., several hundred) are detected for each rear still image data Db2. Then, the second data DT2 is acquired for each feature point P.

[0027] In this embodiment, the first data DT1 and the second data DT2 are a data list (t, X, Y) that defines, for each feature point P, the relationship between the coordinates (X, Y) of the feature point P and the shooting time (t) of the still image data Db that includes the feature point P. By referring to this data list (t, X, Y), a user of the crash test measuring device 20 can understand the movements of the vehicle 11 and the barrier 12 during the crash test.

[0028] According to this embodiment, the following [First Process], [Second Process], and [Third Process] are automatically performed by image recognition using video data Da of the vehicle 11 and the barrier 12 captured during a collision test. Therefore, the movements of the vehicle 11 and the barrier 12 during the collision test can be easily measured.

[0029] [First process] A process for detecting feature points P from the specific still image data Db1. [Second process] A process of tracking and detecting the feature points P in a plurality of subsequent still image data Db2.

[0030] [Third Process] A process of acquiring the first data DT1 or the second data DT2 for each feature point P. <Measurement details> The measurement process for measuring the movements of the vehicle 11 and the barrier 12 during the crash test will now be described in detail.

[0031] Fig. 4 shows the execution procedure of the measurement process. Note that the series of processes shown in the flowchart of Fig. 4 conceptually show the execution procedure of the measurement process, and the actual process is executed by the collision test measurement device 20 as a process at predetermined intervals.

[0032] 4, in this process, first, video data Da is read and stored in the memory unit 22 of the crash test measuring device 20 (step S11). In this embodiment, when a crash test is performed, the vehicle 11 and the barrier 12 are photographed by the camera 13. That is, the video data Da is acquired by the camera 13. In the process of step S11, this video data Da is read and stored in the memory unit 22. In this embodiment, the process of step S11 corresponds to the first step.

[0033] Thereafter, the frame rate of the video data Da is manually input or automatically acquired, and the reciprocal of the acquired frame rate (1 / frame rate) is defined as the time interval between the individual still image data Db that make up the video data Da (step S12). For example, if the frame rate is "1000 fps," then its reciprocal, "1 / 1000 fps," or "1 / 1000 seconds," is defined as the time interval between the individual still image data Db that make up the video data Da.

[0034] Then, the measurement start time is defined as [t=0], and the measurement end time is defined as [t=N] (step S13). Then, the first still image data Db (the specific still image data Db1) among the multiple still image data Db constituting the video data Da is acquired as the still image data Db corresponding to time [t=0] (step S14). As shown in Fig. 5, each still image data Db is expressed in pixel coordinates. If the horizontal coordinate is "X" and the vertical coordinate is "Y", for example, a 1920 x 1080 pixel image is expressed as coordinates (X = 0 to 1919, Y = 0 to 1079).

[0035] Thereafter, as shown in Fig. 6, a feature point P is detected by image recognition from specific still image data Db1 corresponding to time [t=0] (step S15 in Fig. 4). In the processing of step S15, multiple feature points P (P1, P2, P3, ...) are detected. Fig. 6 shows only one (feature point P1) of the multiple feature points P detected in the processing of step S15 together with the specific still image data Db1.

[0036] Furthermore, in the processing of step S15, for each feature point P, the coordinates (X, Y) of the feature point P in the specific still image data Db1 are linked to the time [t=0] to obtain a data list (0, X, Y). In this embodiment, this data list (0, X, Y) corresponds to the first data DT1. In this embodiment, the processing of step S15 corresponds to the second step, and is performed by the first data obtaining unit 24, which is one of the functional units of the crash test measuring device 20.

[0037] Thereafter, the time [t] is advanced to time [t=1] (step S16). Then, the second still image data Db (the subsequent still image data Db2) among the plurality of still image data Db constituting the moving image data Da is acquired as the still image data Db corresponding to the time [t=1] (step S17).

[0038] Thereafter, based on the feature point P of the specific still image data Db1, the feature point P is detected from the subsequent still image data Db2 corresponding to the time [t=1] by the optical flow method (step S18). More specifically, the feature point P is tracked between adjacent frames (in this case, between the specific still image data Db1 and the subsequent still image data Db2 corresponding to the time [t=1]), thereby detecting the feature point P in the subsequent still image data Db2.

[0039] Furthermore, in the processing of step S18, for each detected feature point P, the coordinates (X, Y) of the feature point P in the subsequent still image data Db2 and the time [t=1] are linked to each other to obtain a data list (1, X, Y).

[0040] Thereafter, the time [t] is advanced to "t=t+1" (step S19), and it is determined whether the advanced time [t] has exceeded the measurement end time [t=N] (step S20).

[0041] If the time [t] has not exceeded the measurement end time [t=N] (step S20: NO), the processes of steps S17 to S20 are repeatedly executed until the time [t] exceeds the measurement end time [t=N]. As a result, in the next processing loop, a data list (2, X, Y) corresponding to time [t=2] is acquired, and in the next processing loop, a data list (3, X, Y) corresponding to time [t=3] is acquired, and so on, and data lists (t, X, Y) are sequentially added. In this embodiment, the data list (t, X, Y) acquired through the processes of steps S17 to S20 corresponds to the second data DT2. In this embodiment, the processes of steps S17 to S20 correspond to the third step and are performed by the second data acquisition unit 25, which is one of the functional units of the crash test measurement device 20.

[0042] Then, the processes of steps S17 to S20 are repeatedly executed, and when the time [t] exceeds the measurement end time [t=N] (step S20: YES), this process is ended. <Actions and effects of measurement processing> In this embodiment, through the measurement process, a data list (t, X, Y) is created for each feature point P, which defines the relationship between the coordinates (X, Y) of the feature point P and the capture time (t) of the still image data Db including the feature point P. FIG. 7 is a table conceptually showing the data structure of the data list (t, X, Y). According to this embodiment, by referring to the data list (t, X, Y) shown in FIG. 7, it is possible to understand the movements of the vehicle 11 and the barrier 12 during the collision test.

[0043] The crash test measuring device 20 of this embodiment executes a process (measurement result display process) for displaying the measurement results on the display device 23. The measurement result display process is executed as follows. As shown in FIG. 1, the display device 23 has a first display area 231 that displays a data list (t, X, Y) in the form of a matrix table (see FIG. 7).

[0044] When the measurement process (see Figure 4) is completed, the crash test measurement device 20 creates the matrix table based on the data list (t, X, Y) and displays it in the first display area 231 of the display device 23, either automatically or by operating an input operation unit such as a keyboard.

[0045] The display device 23 also has a second display area 232. The crash test measuring device 20 is configured to be able to display, in the second display area 232, line data DL indicating the movement trajectory of the characteristic point P during the crash test, superimposed on the video data Da or the still image data Db.

[0046] When the measurement process is completed, the crash test measuring device 20 generates line data DL indicating the movement trajectory of each feature point P (P1, P2, P3, ...) based on the data list (t, X, Y), either automatically or by operating the input operation unit. The line data DL is basically data indicating a line connecting the same feature points P in adjacent frames (still image data Db). Then, this line data DL is displayed on the display device 23 in a state where it is superimposed on the video data Da or one of the plurality of still image data Db.

[0047] FIG. 8 shows an example of the display content of the second display area 232. In the example shown in FIG. 8, rear still image data Db2 corresponding to the time [t=M] immediately before the vehicle 11 collides with the barrier 12 and line data DL indicating the trajectory of the feature point P1 during the period from time "t=0" to time "t=M" are displayed in an overlapping state. Although FIG. 8 shows an example of only one piece of line data DL, in reality, the second display area 232 of the display device 23 displays the line data DL corresponding to all of the feature points P separately. In this embodiment, the process of forming the line data DL and displaying it in the second display area 232 is performed by the line data display unit 26, which is one of the functional units of the collision test measurement device 20.

[0048] <Actions and effects of measurement result display processing> According to this embodiment, the movement of the characteristic point P during the crash test can be intuitively grasped by looking at the matrix table displayed in the first display area 231 of the display device 23 and the video data Da, still image data Db, and line data DL displayed in the second display area 232. This allows the movement of each part of the vehicle 11 and each part of the barrier 12 during the crash test to be intuitively grasped.

[0049] <Effects of this embodiment> The effects of this embodiment will be described. (1) The crash test measuring device 20 measures the movements of the vehicle 11 and the barrier 12 during a crash test in which the vehicle 11 and the barrier 12 collide with each other. The crash test measuring device 20 includes a memory unit 22 that stores video data Da of the vehicle 11 and the barrier 12 captured by a camera 13 during the crash test, a first data acquisition unit 24 that acquires first data DT1, and a second data acquisition unit 25 that acquires second data DT2.

[0050] According to the above configuration, the movements of the vehicle 11 and the barrier 12 during a crash test can be easily measured. (2) The crash test measuring device 20 includes a display device 23 and a line data display unit 26. The line data display unit 26 creates line data DL indicating the movement trajectory of the characteristic point P based on a data list (t, X, Y) consisting of the first data DT1 and the second data DT2. Then, the line data DL is displayed on the display device 23 in a state where it is superimposed on the video data Da or one of the multiple still image data Db.

[0051] According to the above configuration, the movement of each characteristic point P during the crash test can be intuitively grasped by looking at the video data Da, still image data Db, and line data DL displayed in the second display area 232 of the display device 23. This allows the movement of each part of the vehicle 11 and the movement of each part of the barrier 12 during the crash test to be intuitively grasped.

[0052] (3) The first data acquisition unit 24 performs feature point detection to detect the feature points P by corner detection and edge detection. According to the above configuration, feature point detection is performed by "corner detection," so that parts in the specified still image data Db1 where angular shapes are displayed can be detected as feature points P. Moreover, feature point detection is also performed by "edge detection," so that parts in the specified still image data Db1 where shapes indicating edges are displayed can be detected as feature points P. Therefore, various parts of the vehicle 11 and the barrier 12 can be detected as feature points P.

[0053] <Example of change> The above embodiment can be modified as follows: The above embodiment and the following modifications can be combined with each other within the scope of technical compatibility.

[0054] The second display area 232 of the display device 23 is not limited to displaying the line data DL corresponding to all of the feature points P separately, but may display only one piece of line data DL or only some of all of the line data DL. In this case, the display content of the second display area 232 may be switched by operating an input operation unit such as a keyboard.

[0055] Feature point detection is not limited to being performed by corner detection and edge detection, but can be performed by combining various detection methods such as corner detection, edge detection, and blob detection. For example, feature point detection can be performed by corner detection, edge detection, and blob detection, or by only one of corner detection, edge detection, and blob detection.

[0056] - For feature point detection, in addition to the "AKAZE" algorithm, any algorithm can be used, such as the "KAZE" algorithm, the "SIFT" algorithm, the "SURF" algorithm, or the "Shi-Tomasi" algorithm.

[0057] The method for tracking the feature point P is not limited to optical flow methods such as KLT tracking, but any method such as motion tracking can be used as long as it is capable of tracking an object in the video captured by the camera 13.

[0058] In the image recognition process in the crash test measuring device 20, source libraries other than the "OpenCV" library can be used as long as they are source libraries that compile functional programs for image recognition processing, etc. An example of such a source library is the "scikit-image" library in the Python language.

[0059] Instead of defining the measurement start time [t=0] as the time when the first still image data Db among the multiple still image data Db constituting the video data Da is captured, it may be defined as the time when the second or subsequent still image data Db among the multiple still image data Db is captured. For example, it is possible to define the measurement start time [t=0] as the time when the still image data Db immediately before the vehicle 11 starts moving during the collision test is captured. In this case, the operation for specifying the still image data Db corresponding to the measurement start time [t=0] may be performed by operating an input operation unit such as a keyboard.

[0060] The measurement end time [t=N] may be defined as the time when the last still image data Db among the multiple still image data Db constituting the video data Da was captured, or as the time when the still image data Db before the last among the multiple still image data Db was captured. For example, the measurement end time [t=N] may be defined as the time when the still image data Db immediately after each part of the vehicle 11 stops moving during a crash test is captured. In this case, the operation for specifying the still image data Db corresponding to the measurement end time [t=N] may be performed by operating an input operation unit such as a keyboard.

[0061] The process of displaying the matrix table in the first display area 231 of the display device 23 may be omitted. Also, the process of displaying the line data DL and the video data Da or one of the plurality of still image data Db in a superimposed state in the second display area 232 of the display device 23 may be omitted. When both of these processes are omitted, it is also possible to omit the display device 23.

[0062] The crash test measuring device 20 and crash test measuring method according to the above embodiment can be applied to a crash test in which the front of the vehicle 11 collides with a barrier 12 as a crash object, as well as other crash tests. Examples of such crash tests include a crash test in which a dolly as a crash object collides with the side of the vehicle 11, and a crash test in which the rear of the vehicle 11 collides with a crash object. Other examples include a crash test in which a head impactor as a crash object collides with the outer or inner surface of the vehicle 11, and a crash test in which a leg impactor as a crash object collides with the bumper of the vehicle 11.

[0063] To measure the movement of the vehicle 11 and the object being hit during a crash test, video footage of the entire vehicle 11 and the entire object being hit can be used, as well as video footage of only a portion of the vehicle 11 or only a portion of the object being hit. This configuration also makes it possible to measure the movement of the vehicle 11 and the object being hit during a crash test.

[0064] In order to measure the movement of the vehicle 11 and the object being hit during the crash test, it is possible to use video taken by a camera 13 installed outside the vehicle 11, as well as video taken by a camera 13 installed inside the vehicle 11. With this configuration, it is also possible to measure the movement of the vehicle 11 and the object being hit during the crash test. [Explanation of symbols]

[0065] P, P1, P2, P3...feature points Da...Video data Db: Still image data Db1: Specific still image data Db2...still image data DT1...first data DT2...Second data DL...Line data 10...Testing site 11...Vehicle 12...Barrier 13...Camera 20...Crash test measuring device 21...Control unit 22...Storage section 23…Display device 231...1st display area 232…Second display area 24...First data acquisition unit 25...Second data acquisition unit 26...Line data display section

Claims

1. A crash test measurement device that measures movements of a vehicle and a collision object during a crash test in which the vehicle and the collision object collide, a video storage unit that stores video images of the vehicle and the collision object taken by a camera during the collision test; a first data acquisition unit that detects feature points from a specific still image that is one of the plurality of still images that constitute the video, based on the shape of the vehicle and the shape of the collision object captured in the specific still image through image recognition, and acquires first data in which the coordinates of the detected feature points in the specific still image are linked to the capture time of the specific still image; a second data acquisition unit that detects feature points in each of a plurality of subsequent still images, which are still images captured after the specific still image among the plurality of still images, by tracking the feature points detected by the first data acquisition unit through image recognition, and acquires second data that associates the coordinates of the detected feature points in the subsequent still image with the capture time of the subsequent still image; A crash test measuring device comprising:

2. a display device; a line data display unit that creates line data indicating a movement trajectory of the feature point based on the first data and the second data, and displays the line data and one of the moving image or the plurality of still images superimposed on the display device.

2. A crash test measuring device according to claim 1.

3. the first data acquisition unit detects the feature points by corner detection and edge detection; 3. A crash test measuring device according to claim 1 or 2.

4. A crash test measurement method for measuring movements of a vehicle and a collision object during a crash test in which the vehicle and the collision object collide, comprising: a first step of storing a video of the vehicle and the collision object captured by a camera during the collision test in a storage device; a second step of detecting, after the first step, feature points from a specific still image that is one of the plurality of still images that constitute the moving image stored in the storage device, by image recognition based on the shape of the vehicle and the shape of the collision object that appear in the specific still image, and acquiring first data in which the coordinates of the detected feature points in the specific still image are linked to the shooting time of the specific still image; a third step, after the second step, of detecting the feature points in each of a plurality of subsequent still images, which are still images captured after the specific still image among the plurality of still images, by tracking the feature points detected in the first step through image recognition, and acquiring second data linking the coordinates of the detected feature points in the subsequent still image with the capture times of the subsequent still images; A crash test measurement method comprising:

Images