Abnormal travel detecting device and abnormal travel detecting method

Abstract

To quickly and surely determine an abnormally traveling vehicle existing in a distant area or a near area.SOLUTION: The abnormal travel detecting device 100 includes the divided speed recording device V1 that divides the amount of movement of the vehicles 32d detected on the screen IM into the plurality of zones ZO and stores the measurement information MI including the divided zones ZO and the speeds ZV in the storage unit 32a, the collection device V1 that collects the measurement information MI obtained for the plurality of vehicles 32b, and the determination device 32b that compares the measurement information MI obtained for the determination target vehicles VO with the collection result SD obtained by the collection device 32c and determines whether the speeds ZV of the determination target vehicles VO are within the allowable range.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an abnormal driving detection device and an abnormal driving detection method for detecting an abnormally driving vehicle that may cause an accident, for example. [Background technology]

[0002] A dangerous vehicle detection device is known that grasps the behavior of each vehicle traveling in a monitored area, learns the normal behavior of vehicles in the monitored area, and detects vehicles that exhibit behavior that deviates from the learned normal behavior as dangerous vehicles.A method of calculating the vehicle's traveling speed is disclosed in Patent Document 1, which uses the time it takes for the vehicle to pass from a speed measurement start line to a speed measurement end line that is set in advance in the monitored area.

[0003] When using a pair of lines as a reference for passing, as in the device of Patent Document 1, it is difficult to set a wide distance between the lines, which tends to shorten the measurement time, resulting in a smaller number of frames that can be acquired and a larger error. Also, when the angle of view changes, the lines must be reset, which reduces work efficiency. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Laid-Open No. 2011-170568 Summary of the Invention [Problem to be solved by the invention]

[0005] The present invention has been made in view of the above-mentioned background art, and has as its object to quickly and reliably determine whether an abnormally moving vehicle is present in a distant area or a nearby area.

[0006] In order to achieve the above object, the abnormal driving detection device of the present invention comprises a zone speed recording device that classifies the movement amount of a vehicle detected on a screen into one of a plurality of zones and stores measurement information including the zone and the speed in a memory unit, a compilation device that compiles the measurement information obtained for a plurality of vehicles, and a determination device that compares the measurement information obtained for the vehicle to be determined with the compilation result by the compilation device and determines whether the speed of the vehicle to be determined is within an allowable range. Note that in this specification, speed means the speed on the screen.

[0007] In the abnormal driving detection device, the speed zone recording device classifies the movement amount of a vehicle detected on the screen into one of several zones and stores measurement information including the zone and speed in a memory unit. Therefore, while treating the vehicle's movement amount as discrete, data collection makes it possible to grasp the movement amount over the entire vehicle route and obtain the speed on the screen for each zone. The measurement information obtained for multiple vehicles to be learned is compiled by a compilation device and becomes statistically processed reference data. The determination device compares the measurement information obtained for the vehicle to be determined with the compilation results from the compilation device to determine whether the speed of the vehicle to be determined is within an acceptable range. Such determination results can be consistent regardless of the distance or proximity of the vehicle's detected position on the screen.

[0008] In order to achieve the above-mentioned object, the abnormal driving detection method of the present invention classifies the movement amount of a vehicle detected on a screen into one of a plurality of zones, uses a zone speed recording device that stores measurement information including the zone and speed in a memory unit, acquires and compiles measurement information for a plurality of vehicles, and compares the measurement information obtained for the vehicle to be judged with the compiled result of the measurement information obtained for a plurality of vehicles to determine whether the speed of the vehicle to be judged is within an acceptable range. [Brief explanation of the drawings]

[0009] [Figure 1] 1A and 1B are block diagrams illustrating an abnormal driving detection device according to a first embodiment. [Figure 2]FIG. 2 is a diagram showing an installation state of an abnormal driving detection device. [Figure 3] FIG. 1A is a diagram for explaining the acquisition of tracking information by an image processing device, and FIG. 1B is a conceptual diagram for explaining the acquisition of measurement information by a section speed recording device or the like. [Figure 4] FIG. 10 is a conceptual diagram illustrating sorting and counting of measurement information. [Figure 5] FIG. 10 is a conceptual diagram illustrating the results of collecting measurement information. [Figure 6] 10(A) to 10(C) are conceptual diagrams illustrating the allowable speed ranges in each zone. [Figure 7] 10A to 10C are conceptual diagrams illustrating the determination when speeds are obtained in multiple zones. [Figure 8] FIG. 10 is a diagram illustrating correction and complementation of data related to the tolerance range. [Figure 9] 10 is a flowchart illustrating an example of the operation of the abnormal driving detection device. [Figure 10] FIG. 6 is a diagram illustrating an abnormal driving detection device according to a second embodiment. [Figure 11] FIG. 10 is a diagram illustrating an abnormal driving detection device according to a third embodiment. [Figure 12] FIG. 10 is a diagram illustrating a modified example of an abnormal driving detection device according to the third embodiment. [Figure 13] FIG. 10 is a diagram illustrating a modified abnormal driving detection device. DETAILED DESCRIPTION OF THE INVENTION

[0010] [First embodiment] Hereinafter, an abnormal driving detection device and method according to a first embodiment will be described with reference to the drawings.

[0011] Fig. 1(A) is a block diagram showing an example of the configuration of an abnormal driving detection device of an embodiment, and Fig. 1(B) is a functional block diagram explaining the abnormal driving detection device. Fig. 2 is a conceptual side view showing the installation state of the abnormal driving detection device 100 shown in Fig. 1. Fig. 3(A) is a diagram explaining the acquisition of tracking information, and Fig. 3(B) is a conceptual diagram explaining the acquisition of measurement information.

[0012] Referring to Figure 1(A) etc., the abnormal driving detection device 100 includes a camera 10 that photographs the road RO (see Figure 2) on which the vehicle is traveling, and an information processing device 20 that makes a determination regarding the driving state of the vehicle using the photographed image PI (see Figure 3(A)) acquired by the camera 10.

[0013] The camera 10 has a shooting range PA (see FIG. 3(A)) that is the target space including the target road RO, the area above it, and its surroundings, and continuously captures images of the target at a predetermined time interval (for example, about 30 fps). The captured images PI, including road images captured by the camera 10, are output to the information processing device 20. The frame rate of the continuous shooting by the camera 10 is not limited to 30 fps, but may be, for example, 5 fps, 10 fps, or even about 100 fps or more.

[0014] The information processing device 20 is a computer, and includes a main control device 21, a storage device 22, a user interface 23, and a communication device 24. The information processing device 20 detects abnormally driving vehicles that may cause an accident by using the images PI captured by the camera 10. To this end, the information processing device 20 performs a setting process to collect data in advance and set criteria, and a determination process to detect information about individual vehicles and determine whether or not they are driving abnormally based on the set criteria, as will be described in detail later.

[0015] The main control device 21 operates based on programs stored in the storage device 22. The storage device 22 stores basic programs for operating the information processing device 20 and application software that runs on the basic programs, and also stores data and processing results required for executing the programs and software. The application software includes software for operating the camera 10 to acquire captured images PI, software for acquiring captured images PI in advance and executing a setting process, and software for executing a determination process that detects vehicle information from captured images PI acquired after the start of actual operation and makes a determination. The storage device 22 temporarily stores captured images PI captured by the camera 10 and permanently stores continuous video files at times specified by the operator, specific image files related to specific events such as the start of operation or the occurrence of an abnormality, and image files processed as needed. For example, if an abnormally moving vehicle is detected, under the control of the main control device 21, a series of captured images PI for a predetermined number of seconds before and after the detection are stored as a video in the storage device 22, allowing the operator to repeatedly review the video later. The user interface 23 includes a display, a touch panel, a speaker, a microphone, etc., and receives instructions from an operator and presents the operator with the results of processing by the information processing device 20. The information processing device 20 can communicate with an external management server (not shown) via a communication network using the communication device 24. Note that for images PI captured by the camera 10, the information processing device 20 can start or stop recording upon request from an operator who directly operates the information processing device 20 or a remote operator, and the original image file or a processed file can also be downloaded to a remote terminal.

[0016] 1(B), the information processing device 20 includes an image processing device 31 that extracts an image of a vehicle V1 from a captured image PI, a learning processing device 33 that performs setting processing, a detection processing device 34 that performs determination processing, and a user interface unit 35. The learning processing device 33 includes a section speed recording device 32a, a counting device 32b, and a memory unit 32d. The detection processing device 34 includes a section speed recording device 32a, a determination device 32c, and a memory unit 32d.

[0017] Referring to FIG. 2, the abnormal driving detection device 100 is mounted on a sign vehicle WV located adjacent to a maintenance work site MW where construction work or fallen objects are being carried out on a road RO, such as a highway. The abnormal driving detection device 100 includes a camera 10 fixed to a sign 3 fixed to a loading platform 2a, an information processing device main body 20a fixed to the loading platform 2a, and a user interface 23 installed in the driver's seat 2b. The maintenance work site MW is located in front of the sign vehicle WV. A human-shaped warning sign 4 is located behind the sign vehicle WV, upstream in the vehicle's direction of travel, and supports a rotating warning light 4a. Upstream of the human-shaped warning sign 4 in the vehicle's direction of travel, a vehicle V1 is traveling on the road RO and approaching the maintenance work site MW. The sign 3 urges the vehicle V1 to slow down and change lanes. Note that the warning sign 4 does not have to be a human-shaped sign; it may be the same as the sign on the sign vehicle WV, or it may be replaced with something else.

[0018] 2 and 3(A), a target area including the road RO behind the sign vehicle WV and the upstream direction of vehicle travel is captured by the camera 10. The capturing range PA, which corresponds to the angle of view of the camera 10, includes the area outside the road RO. The captured image PI, including the road image output from the camera 10, is processed by the information processing device main body 20a, and the processing result is output to the user interface 23. Note that a vanishing point EX exists at the back of the road RO, but it is not essential that the vanishing point EX be included within the angle of view.

[0019] The captured image PI shown in FIG. 3(A) depicts a road RO having a first lane RL1 and a second lane RL2. In the captured image PI, four vehicles V1 traveling on the second lane RL2 are depicted above the second lane RL2, but this is for convenience of explanation. The vehicle V11 at the farthest back is depicted in the frame in which a specific vehicle V1 is first recognized, vehicle V12 is depicted in the next frame, vehicle V13 is depicted in the frame after that, and vehicle V14 is depicted in the frame after that. In other words, vehicle V1 moves sequentially from the back to the front to the four positions indicated by vehicles V11 to V14.

[0020] Returning to FIG. 1B , in the information processing device 20, the image processing device 31 operates the camera 10 during the setting process to acquire a captured image PI including the road RO ahead as a preliminary image SI, and extracts one or more images of a vehicle V1 from the preliminary image P0. During the determination process, the image processing device 31 also operates the camera 10 to acquire the captured image PI as a target image OI and extracts one or more images of a vehicle V1 from the target image OI. Known pattern recognition techniques can be used to extract the vehicle V1. By incorporating pattern recognition software trained by machine learning, such as deep learning, into the image processing device 31 as a trained model for vehicle extraction, the image region of the vehicle V1 can be determined from the preliminary image P0 or the target image OI, and representative coordinates, such as the center of gravity coordinates, can be calculated for the front region or the lower front end of the vehicle V1. Any type of pattern recognition software can be used as long as it can recognize the vehicle V1. However, distinguishing attributes such as the vehicle type and color makes it easier to check the identity of the extracted vehicle images. A specific example of pattern recognition software that can be used is one that uses, for example, HOG (Histograms of Oriented Gradients) to extract features and a support vector machine as a learning model for identifying the features. The method of extracting vehicle images is not limited to the machine learning method described above, and may also be one that extracts artificially set local features using a statistical learning method, such as optical flow or template matching.

[0021] The image processing device 31 receives the captured images PI as successive frames (e.g., 10 fps, 30 fsp) from the camera 10, and sequentially extracts vehicle images from each captured image PI. The image processing device 31 determines representative coordinates (pixel position or screen coordinates) of the vehicle image for each of the successive captured images PI over time. At night, the image from the camera 10 mainly includes headlights, but even in such a case, the image processing device 31 can identify and track the vehicle V1 using pattern recognition technology.

[0022] The image processing device 31 outputs tracking information ST for the learning target and tracking information OT for the judgment target. Here, the tracking information ST and OT relate to a large number of vehicles that are learning targets and judgment targets, and represent the results of vehicle detection and tracking. Specifically, the tracking information ST and OT includes the vehicle ID of each vehicle extracted from the screen IM of the captured image PI, the coordinate position (X, Y) of each vehicle on the screen IM of the captured image PI, and the frame number. Here, the coordinate position (X, Y) corresponds to the pixel position, and the origin can be a corner of the captured image PI or another reference point such as the vanishing point EX. By using the tracking information ST and OT, the movement vector (ΔX, ΔY) of the vehicle V1 can be determined as the positional deviation of the vehicle image between the captured images PI, and the velocity vector (Vx, Vy) can be determined based on this. In this case, the movement vector and velocity vector can be determined not only between the captured images PI corresponding to two temporally adjacent frames, but also between the captured images PI corresponding to two frames separated from each other by one or more frames.

[0023] The section speed recording device 32a of the learning processing unit 33 will be described with reference to Figures 1(B), 3(A) and 3(B), 4, etc. The section speed recording device 32a classifies the movement amount of a vehicle detected on the screen IM of the captured image PI into one of multiple zones and acquires measurement information MI that includes the classified zone and speed and is linked to the vehicle ID. The section speed recording device 32a outputs this measurement information MI to the compilation device 32b and stores it in the memory unit 32d. Here, dividing into zones corresponds to grasping the approximate position by reducing the position resolution to an appropriate coarseness, which means reducing the amount of information processing. When determining the movement amount on the screen IM, the section speed recording device 32a uses one coordinate axis Y of two coordinate axes X and Y that identify the position on the screen IM. That is, the sectional speed recording device 32a classifies the movement amount ΔY, which is the Y component of the movement amount vector of the vehicle V1 extracted from the screen IM, into sectional zones ZO (see FIG. 4), and determines the speed ZV, which is the Y component of the speed vector, with the resolution of the zone ZO. This speed ZV is stored in the storage unit 32d together with information about the zone ZO and the vehicle ID. Here, each sectional zone ZO is an area having the same width in pixels, for example. In the example shown in FIG. 4, the first zone is in the range of ΔY = 1 to 10, the second zone is in the range of ΔY = 11 to 20, the third zone is in the range of ΔY = 21 to 30, the fourth zone is in the range of ΔY = 31 to 40, the fifth zone is in the range of ΔY = 41 to 50, the sixth zone is in the range of ΔY = 51 to 60, ... and the nth zone is in the range of ΔY = 10 × (n - 1) + 1 to 10n (where n is a natural number). The representative value ZY of each zone is considered to be, for example, the maximum value or the median value. In the example shown in Fig. 4, for the n-th zone, the representative value ZY(n) = 10n.

[0024] As shown in FIG. 3A, when a series of movement amounts ΔY exist for the detected vehicle V1, the sectional speed recording device 32a defines the starting point Y1 (or Y2, Y3, ...) of the series of movement amounts ΔY of the detected vehicle V1 as a base point BP, and classifies the movement amounts ΔY from the base point BP as zones ZO. That is, for data with a matching vehicle ID, the movement amounts ΔY are measured for elapsed times that are natural number multiples of the frame period from the common base point BP, and each measurement value is classified and sorted into a zone ZO. Specifically, as shown in FIG. 3B, for example, when the starting point Y1 is the base point BP, the movement amounts of the vehicle V1 classified into the zones ZO are obtained as representative values ​​ZY = 10, 40, and 60 for the zones ZO in four consecutive frames. In the above description, it is assumed that the starting points Y1, Y2, Y3, ... are Y coordinate positions (i.e., Y pixel positions), but they may also be specified by the zones ZO. In this case, the positions and movement amounts of the vehicle V1 are all classified into zones ZO.

[0025] When a series of movement amounts ΔY exist for the detected vehicle V1, the start point Y1 of the series of movement amounts ΔY for the detected vehicle V1 and subsequent points Y2 and Y3, which are successively shifted from the start point Y1, are set as base points BP, and the movement amounts ΔY from the multiple base points BP (i.e., the start point Y1 and the subsequent points Y2 and Y3) are each divided into zones ZO. Specifically, as shown in FIG. 3B, for example, for the movement amounts of the vehicle V1 divided into zones ZO when the start point Y1 is set as the base point BP, representative values ​​ZY=10, 40, and 60 are obtained for four consecutive frames. Furthermore, for the movement amounts when the subsequent point Y2 is set as the base point BP, representative values ​​ZY=20 and 50 are obtained for three consecutive frames. Furthermore, for the movement amounts when the subsequent point Y3 is set as the base point BP, a representative value ZY=30 is obtained for two consecutive frames. In the sorting table shown in Figure 4, the area HA indicated by cross-pattern hatching corresponds to the above specific example, and indicates that when the starting point Y1 is the base point BP, movement amounts are obtained in zones ZO with representative values ​​ZY = 10, 40, and 60; when the starting point Y2 is the base point BP, movement amounts are obtained in zones ZO with representative values ​​ZY = 20 and 50; and when the starting point Y3 is the base point BP, movement amounts are obtained in zone ZO with representative value ZY = 30.

[0026] The section speed recording device 32a calculates the speed ZV based on the representative value ZY corresponding to the movement amount classified as shown in Figures 3(B) and 4, i.e., the movement amount ΔY of the target vehicle V1 as determined by the zone ZO. The speed ZV is calculated by dividing the representative value ZY of the movement amount of the target vehicle V1 by the travel time of the vehicle V1. Here, the travel time of the vehicle V1 corresponds to the number of frames when the representative value ZY is acquired and is a natural number multiple of the frame period. The speed ZV corresponds to the speed of the vehicle V1 on the screen IM measured with the resolution or roughness of the zone ZO. Specifically, the speed ZV of the vehicle V1 when the starting point Y1 is the base point BP is calculated by dividing the movement amount, i.e., the representative value ZY = 10, 40, 60, by the travel time. If the frame period is PF (s), the velocity ZV of the vehicle V1 is 10 / PF (pixel / s), 40 / 2PF (pixel / s), and 60 / 3PF (pixel / s).

[0027] The counting device 32b shown in Fig. 1(B) receives the measurement information MI from the section speed recording device 32a, counts the measurement information MI obtained for multiple vehicles, and stores the counting results SD in the memory unit 32d. The counting results SD become reference data when the determination device 32c makes a determination.

[0028] FIG. 5 shows a tabulation table 37 obtained by tabulation by the tabulation device 32b. In the tabulation table 37, the speed data VD labeled V1, V2, and V3 pertain to the vehicles V1, V2, and V3 and represent the speeds ZV of the vehicles V1, V2, and V3 on the scale of the zone ZO. Each field FI, identified by the base point BP and the zone ZO, stores the speed data VD through learning; for example, approximately 10 pieces of speed data VD are stored. The number of speed data VD in each field FI depends on parameters such as the number of vehicles V1, V2, V3, ... used in learning and the width of the zone ZO, and is therefore irregular. It is desirable to set the number of speed data VD in each field FI to a predetermined number, such as at least several or several tens, in order to ensure the statistical reliability of the tabulation and learning.

[0029] In the example shown in Figure 5, the speed data VD is recorded in zone ZO corresponding to the front end of the range in which vehicles V1, V2, V3, ... are detected, but if uniform processing is performed throughout, the speed data VD may also be recorded in zone ZO corresponding to the center of the range in which vehicles V1, V2, V3, ... are detected.

[0030] The counting device 32b shown in Fig. 1(B) performs the above-described counting process on, for example, 100 vehicles to be learned, and stores the counting results SD as counted data in the storage unit 32d. The counted data can include the average value and standard deviation of the speed data VD included in each column FI, in addition to information corresponding to the counting table 37 shown in Fig. 5.

[0031] The determination device 32c of the detection processing unit 34 receives from the zone speed recording device 32a object measurement information OMI, which is information including the zone and speed of the vehicle to be determined (hereinafter also referred to as the determination object vehicle VO), and which links these with the vehicle ID. This object measurement information OMI is similar to the measurement information MI described with respect to the learning processing unit 33, and if the vehicle V1 is the object to be determined, specifically, it includes information such as the amount of movement, i.e., the representative value ZY shown in FIG. 3(B), and the speed ZV calculated from this.

[0032] 6(A) and other figures, a method for determining an abnormally moving vehicle by the determination device 32c will be described. The determination device 32c identifies an abnormally moving vehicle using the aggregation table 37 obtained in the learning stage as described above. The determination device 32c uses the average value of the speed in each column FI in the aggregation table 37 as a reference, and determines that the speed of the determination target vehicle VO is outside the allowable range if the speed of the determination target vehicle VO deviates by more than a predetermined value from the speed in the corresponding zone ZO, i.e., the average value AV of the speed data VD.

[0033] The histogram shown in FIG. 6(A) shows the learning results corresponding to a specific field FI in the aggregation table 37 illustrated in FIG. 5. The horizontal axis of the histogram indicates the vehicle speed in the Y direction on the screen IM (see FIG. 3), and the vertical axis indicates the frequency or number of vehicles. The numerous vertical bars obtained by learning correspond to the speed data VD and approximate a Gaussian distribution. As a result, the average value AV and standard deviation σ calculated from the distribution of the speed data VD can be used as criteria for determining whether or not the speed is within the allowable range. Specifically, if the speed ZV indicated by the object measurement information OMI obtained for the target vehicle VO is less than a predetermined upper limit, i.e., the sum (AV + σ) of the average value and the allowable deviation (specifically, the standard deviation), the determination device 32c determines that the speed ZV is within the allowable range. In this case, it is estimated that the speed ZV of the target vehicle VO is not in an abnormal driving state. On the other hand, if the speed ZV indicated by the object measurement information OMI obtained for the target vehicle VO is equal to or greater than the upper limit value AV+σ, the determination device 32c determines that the speed ZV is outside the allowable range. In this case, the speed ZV of the target vehicle VO is estimated to be in an abnormal driving state. In the above, the allowable range can be considered as the standard range of the speed data VD obtained by learning.

[0034] FIG. 6B is a diagram illustrating a modified example of the determination criteria used by the determination device 32c. In this case, a value obtained by multiplying the standard deviation σ by a coefficient k (where k is a value set, for example, between 0.5 and 3.0) is used. Specifically, if the speed ZV indicated by the object measurement information OMI obtained for the determination target vehicle VO is less than a predetermined upper limit value AV+k·σ, the determination device 32c determines that the speed ZV is within the allowable range. On the other hand, if the speed ZV indicated by the object measurement information OMI obtained for the determination target vehicle VO is equal to or greater than the upper limit value AV+k·σ, the determination device 32c determines that the speed ZV is outside the allowable range. In this case, too, the allowable range can be considered as the standard range of the speed data VD obtained by learning.

[0035] FIG. 6C illustrates another variation of the determination criteria used by the determination device 32c. In this case, the range is defined using the standard deviation σ multiplied by a coefficient −p (p is, for example, a value between 1.0 and 4.0). This makes it possible to identify vehicles traveling at abnormally high speeds or abnormally low speeds. Specifically, if the speed ZV indicated by the object measurement information OMI obtained for the determination target vehicle VO is less than a predetermined upper limit value AV+σ and exceeds a predetermined lower limit value AV-p·σ, the determination device 32c determines that the speed ZV is within the allowable range. On the other hand, if the speed ZV indicated by the object measurement information OMI obtained for the determination target vehicle VO is equal to or greater than the upper limit value AV+σ or equal to or less than the lower limit value AV-p·σ, the determination device 32c determines that the speed ZV is outside the allowable range.

[0036] Although not shown, in the example shown in Fig. 6(C), the upper limit value can also be set to AV+k σ, as in Fig. 6(B). In this case, if the speed ZV exceeds the lower limit value AV-p σ and is less than the upper limit value AV+k σ, the speed ZV is within the allowable range.

[0037] 7(A) to 7(C) are diagrams illustrating the determination by the determination device 32c. The target measurement information OMI obtained for the determination target vehicle VO may contain not only one value of the speed ZV, but also multiple values. For example, in the example shown in FIG. 4 or 5, when the starting point Y1 is used as the base point BP for the vehicle V1, three speed data VD are obtained. In such a case, the problem is how to utilize the multiple obtained speed data VD to reach a conclusion. In the case of FIG. 7(A), the speed ZV obtained by the sectional speed recording device 32a is equal to or greater than the upper limit of the allowable range and is outside the allowable range. In the case of FIG. 7(B), the speed ZV obtained by the sectional speed recording device 32a is less than the upper limit of the allowable range and is within the allowable range. In the case of FIG. 7(C), the speed ZV obtained by the sectional speed recording device 32a is also less than the upper limit of the allowable range and is within the allowable range. In other words, two determinations are obtained for the same determination target vehicle VO: one determination that the speed is outside the allowable range and the other that the speed is within the allowable range. In this case, if the ratio of speed data VD determined to be outside the allowable range to the speed data VD obtained for the target vehicle VO is equal to or greater than a predetermined value, the speed of the target vehicle VO is determined to be outside the allowable range. Here, a ratio equal to or greater than a predetermined value means, for example, 0.5 or greater, but is not limited to this. In the example illustrated in Figures 7(A) to 7(C), the ratio of speed data VD outside the allowable range is approximately 0.33, which is less than 0.5, so the target vehicle VO is determined to be an abnormally moving vehicle. In other words, if the results shown in Figures 7(A) to 7(C) are obtained, the target vehicle VO is determined to not be an abnormally moving vehicle.

[0038] Although not explained here, in this embodiment, starting points Y1, Y2, Y3, ... are used, so in the example shown in Figure 4 or 5, strictly speaking, six speed data VD are obtained for vehicle V1, and it is determined whether the proportion of speed data VD that is outside the allowable range among these six speed data VD is greater than or equal to a predetermined value.

[0039] FIG. 8 is a conceptual diagram illustrating the average value AV of the speed data VD in the vertical column FI in the aggregation table 37 shown in FIG. 5, where the base point BP is the starting point Y1. The average value AV of the speed data VD is indicated by circles. The average value AV may vary due to various error factors, and in extreme cases, missing values ​​may occur. In such cases, the average speed value AV used for judgment can be set more statistically precisely using an approximation curve LL obtained by fitting a quadratic, cubic, or higher order function to the actual measurement points indicated by circles. In this case, columns FI where the average speed value AV is missing can also be supplemented by interpolation or extrapolation. Although a detailed explanation is omitted, the standard deviation σ may also vary and be missing, but the standard deviation σ used for judgment can be set more statistically precisely using an approximation curve similar to that shown in FIG. 8.

[0040] Returning to FIG. 1(B), the memory unit 32d stores the measurement information MI to be learned, the aggregation result SD corresponding to the aggregation table 37 that aggregates the measurement information MI, and information regarding the allowable range including the average value AV and the standard deviation σ.

[0041] 9 is a flowchart illustrating the operation of the abnormal driving detection device 100 shown in FIGS. 1(A) and 1(B). The information processing device 20 of the abnormal driving detection device 100 operates the camera 10 and first starts video input as a setting processing stage (step S01). That is, a prior image P0 for learning, which is an image PI captured by the camera 10, is input to the image processing device 31. The image processing device 31 starts a moving object detection process for the prior images P0 that are sequentially input (step S02). That is, the image processing device 31 starts extracting images of the vehicle V1, etc. from the prior image P0.

[0042] If the image processing device 31 cannot detect a vehicle (NO in step S03), it repeatedly checks whether a vehicle has been detected, and if it detects a vehicle (YES in step S03), it outputs tracking information ST about the detected vehicle, i.e., the learning target vehicle, to the sectional speed recording device 32a of the learning processing unit 33. The sectional speed recording device 32a checks whether a movement vector about the detected vehicle has been acquired or extracted from the tracking information ST (step S04). If a movement vector has been acquired, the sectional speed recording device 32a calculates measurement information MI and outputs it to the counting device 32b (step S05).

[0043] Thereafter, the information processing device 20 checks whether the learning stage or the setting processing stage has ended (step S06). If the setting processing stage has not ended (NO in step S06), the counting device 32b performs counting on the measurement information MI received from the section speed recording device 32a, thereby updating the counting results SD including the counting table 37 and the like that have been obtained up to that point, and stores the updated counting results SD in the memory unit 32d (step S07). Note that if the setting processing stage has ended (YES in step S06), the information processing device 20 proceeds to the determination processing stage in which a determination is made on the determination target vehicle VO.

[0044] In addition, if the process has already moved from the setting processing stage to the judgment processing stage (if the process has proceeded after YES in step S06), the image processing device 31 receives a target image OI for judgment or actual use, which is an image PI captured by the camera 10, and the image processing device 31 performs moving object detection processing on the target images OI that are input sequentially.

[0045] When the setting process stage is completed and the process proceeds to the determination process stage (YES in step S06), the determination device 32c of the detection processing unit 34 takes in the measurement information MI obtained immediately before by the section speed recording device 32a as the target measurement information OMI, and performs a determination process on the speed data included in the target measurement information OMI by referring to the aggregation result SD obtained in step S07 (step S08). Specifically, for example, if the speed ZV indicated by the target measurement information OMI obtained for the target vehicle VO is less than the upper limit value AV+σ or AV+k·σ, the determination device 32c determines that the speed ZV is within the allowable range and that the target vehicle VO is not an abnormally moving vehicle. Conversely, if the speed ZV indicated by the target measurement information OMI obtained for the target vehicle VO is equal to or greater than the upper limit value AV+σ or AV+k·σ, the determination device 32c determines that the speed ZV is outside the allowable range and that the target vehicle VO is an abnormally moving vehicle. Furthermore, when multiple speed data VD are obtained, if the proportion of speed data VD that is judged to be outside the acceptable range is greater than or equal to a predetermined value, the speed of the vehicle VO to be judged is judged to be outside the acceptable range, and if the proportion of speed data VD that is judged to be outside the acceptable range is less than a predetermined value, the speed of the vehicle VO to be judged is judged to be within the acceptable range.

[0046] When the determination device 32c determines that the target vehicle VO is an abnormally driving vehicle (YES in step S09), it outputs the detection of the abnormally driving vehicle via the user interface unit 35 and notifies the operator of detection information indicating the presence of the abnormally driving vehicle (step S11). The abnormally driving vehicle notification includes vehicle images and audio. The detection information may include the abnormal driving determination result, vehicle information including the vehicle's movement trajectory and the degree of deviation from the average value, the time of abnormality detection, etc. The determination device 32c can also output the detection information and commands to deal with it to an external device. This makes it possible to cause the display, which is the sign 3 of the marked vehicle WV shown in FIG. 2, to display a warning such as "Abnormally Driving Vehicle Present" to the abnormally driving vehicle and surrounding vehicles.

[0047] If the determination device 32c determines that the vehicle VO to be determined is not an abnormally driving vehicle (NO in step S09), it outputs the detection of a normally driving vehicle via the user interface unit 35 and notifies the operator of the detection information indicating the presence of a normally driving vehicle (step S10).

[0048] Thereafter, the aggregation device 32b aggregates the object measurement information OMI used in the determination process of step S08 to update the aggregation result SD consisting of the aggregation table 37 and the like obtained up to that point, and stores the updated aggregation result SD in the storage unit 32d (step S12). That is, even in the determination process stage, the aggregation result SD is updated using the object measurement information OMI obtained about the determination target vehicle VO. Note that it is not necessary to update the aggregation result SD using the object measurement information OMI obtained about the determination target vehicle VO in the determination process stage, and it is also possible to process the aggregation result SD obtained in the setting process stage without updating it in the determination process stage.

[0049] The information processing device 20 determines whether it is time to end the measurement, as when an end command is received, and if the measurement is not to be ended (NO in step S13), returns to step S03 and repeats the same processing up to step S12.

[0050] The abnormal driving detection device 100 of the first embodiment described above includes a zone speed recording device 32a that classifies the movement amount of a vehicle V1 detected on a screen IM into one of a plurality of zones ZO and stores measurement information MI including the classified zone ZO and the speed ZV in a memory unit 32d, a compilation device 32b that compiles the measurement information MI obtained for the plurality of vehicles V1, and a determination device 32c that compares the measurement information MI obtained for a target vehicle VO with the compilation result SD by the compilation device 32b to determine whether the speed ZV of the target vehicle VO is within an allowable range. In other words, the zone speed recording device 32a monitors or classifies not only a limited area on the screen IM but also various areas on the road RO.

[0051] In the abnormal driving detection device 100, the zone speed recording device 32a classifies the movement amount of the vehicle V1 detected on the screen IM into one of the zones ZO and stores measurement information MI including the zone ZO and the speed ZV in the memory unit 32d. Therefore, even while treating the movement amount of the vehicle V1 as discrete, the movement amount can be grasped over the entire route of the vehicle V1, and the speed ZV on the screen IM can be obtained for each zone ZO. The measurement information MI obtained for multiple vehicles V1 to be learned is compiled by the compilation device 32b and becomes statistically processed reference data. The determination device 32c compares the measurement information MI obtained for the target vehicle VO with the compilation result SD by the compilation device 32b to determine whether the speed ZV of the target vehicle VO is within an allowable range. Such a determination result can be consistent regardless of the distance or proximity of the vehicle detection position on the screen IM.

[0052] [Second embodiment] The abnormal driving detection device and method according to the second embodiment will be described below. In the second embodiment, the same matters as those in the first embodiment will not be described.

[0053] 10, the zones ZO in the aggregation table 37 are grouped into a first range R1 where the representative value ZY is between 10 and 30, a second range R2 where the representative value ZY is between 40 and 50, and a third range R3 where the representative value ZY is 60 or greater. In this embodiment, the determination device 32c weights the results for each of the ranges R1, R2, and R3, and prioritizes the result for the third range R3, which is closest to the camera. In other words, the determination device 32c prioritizes the speed obtained for a zone with a relatively large amount of movement (e.g., the third range R3) over the speed obtained for a zone with a relatively small amount of movement (e.g., the first range R1).

[0054] The number of data items in the column FI belonging to the first range R1 that are determined to be outside the allowable range for the speed ZV is N1(+), and the number of data items in the column FI belonging to the first range R1 that are determined to be within the allowable range for the speed ZV is N1(-). The number of data items in the column FI belonging to the second range R2 that are determined to be outside the allowable range for the speed ZV is N2(+), and the number of data items in the column FI belonging to the second range R2 that are determined to be within the allowable range for the speed ZV is N2(-). The number of data items in the column FI belonging to the third range R3 that are determined to be outside the allowable range for the speed ZV is N3(+), and the number of data items in the column FI belonging to the third range R3 that are determined to be within the allowable range for the speed ZV is N3(-). Furthermore, the coefficient for the number of data N1(+) and N1(-) in the first range R1 is a positive number c1, the coefficient for the number of data N2(+) and N2(-) in the second range R2 is a positive number c2, and the coefficient for the number of data N3(+) and N3(-) in the third range R3 is a positive number c3, and c1 <c2<c3とする。

[0055] In this case, the numerical value D(-) within the allowable range weighted according to the region is D(-) = c1 × N1(-) + c2 × N2(-) + c3 × N3(-), and the numerical value D(+) outside the allowable range weighted according to the region is D(+) = c1 × N1(+) + c2 × N2(+) + c3 × N3(+). Based on this premise, the determination device 32c determines that the speed of the determination target vehicle VO is outside the allowable range if the ratio of the numerical value D(+) outside the allowable range to the total TA = D(-) + D(+) is equal to or greater than a predetermined value, and determines that the speed of the determination target vehicle VO is within the allowable range if the ratio of the numerical value D(+) outside the allowable range is less than the predetermined value.

[0056] In the above, the division into ranges R1, R2, and R3 is merely an example, and the number of divisions and the distribution of weights can be set in various ways depending on the application.

[0057] [Third embodiment] The abnormal driving detection device and method of the third embodiment will be described below. In the third embodiment, the same matters as those in the first embodiment will not be described. In this case, the determination device 32c determines whether the speed ZV obtained for the zone ZO with the largest movement amount is outside the allowable range based on the speed ZV.

[0058] In the aggregation table 37 shown in FIG. 11 , for example, for the speed ZV of vehicle V1, the data in the column for the starting point Y1 that is classified into zone LZ1 with a representative value ZY=60 is the data with the longest travel distance. Data classified into zone LZ1 with a long travel distance like this generally has high speed detection accuracy. Therefore, by using the speed ZV classified into this longest zone LZ1 as the judgment criterion, the accuracy of the judgment can be improved without significantly reducing the reliability of the judgment. Using the longest zone LZ1 also has the advantage that the number of speed data VD included in the column FI does not need to be very large. For the speed ZV of vehicle V2, the data classified into zone LZ2 (which coincides with zone LZ1) is the data with the longest travel distance, and the speed ZV classified into this longest zone LZ2 is used as the judgment criterion. For the speed ZV of vehicle V3, the data classified into zone LZ3 is the data with the longest travel distance, and the speed ZV classified into this longest zone LZ3 is used as the judgment criterion. Use of such a determination method can reduce the amount of calculation and the processing load of the information processing device 20. Whether to select the processing of the first embodiment or the processing of the third embodiment depends on the circumstances such as the hardware constituting the information processing device 20, the detection response of the image processing device 31, and the tolerance for erroneous detection.

[0059] FIG. 12 is a diagram illustrating a modification related to FIG. 11. In this case, data is not collected for the columns after the starting point Y3. The speed ZV obtained with the successor points Y2 and Y3 as the base points BP is generally small. Therefore, when making a determination using the longest zones LZ1 and LZ2, it is unlikely that the speed ZV with the successor points Y2 and Y3 as the base points BP will be used, and data is not collected, particularly when the base points BP are points after the successor point Y3. By using such a counting table 37, the processing load on the information processing device 20 can be further reduced.

[0060] [others] The present invention is not limited to the above-described embodiment, and can be embodied in various forms without departing from the spirit and scope of the present invention.

[0061] For example, in the above description, the lanes RL1 and RL2 of the road RO are not distinguished from each other. However, as shown in FIG. 13, the vehicle speed ZV can be measured independently in the regions LA1 and LA2 corresponding to the lanes RL1 and RL2, and a separate aggregation table 37 can be prepared for each lane RL1 and RL2. In this case, even if there is a large difference in the slope of the lanes RL1 and RL2, an aggregation table 37 adapted to the slope of each lane RL1 and RL2 can be prepared, thereby improving the reliability of the determination of abnormal driving. While FIG. 13 illustrates two lanes RL1 and RL2, a separate aggregation table 37 can be prepared for each lane even for a road RO with three or more lanes. Note that when preparing an aggregation table 37 for each lane, handling vehicles that change lanes becomes an issue. In this case, data is incorporated into the aggregation table 37 based on the lane in which the vehicle that changed lanes was last measured. The determination can also be made with emphasis on the measurement results for the lane in which the vehicle was last measured.

[0062] In the above embodiment, the movement amount ΔY, which is the Y component of the extracted movement amount vector of the vehicle, is sorted into zones ZO to create the tally table 37, but when the movement amount ΔX is larger than the movement amount ΔY, such as when the vehicle is traveling horizontally within the screen IM, the movement amount ΔX, which is the X component, may be sorted into zones ZO to create the tally table 37. Furthermore, when there is little difference in magnitude between the movement amounts ΔY and ΔX, the tally table 37 may be created by sorting into zones ZO the movement amount in the direction of the road RO, which is inclined midway between the X and Y axes.

[0063] The abnormal driving detection device 100 and the camera 10 do not have to be mounted on the sign-marked vehicle WV. For example, one or more cameras 10 can be installed near or in a remote location from the sign-marked vehicle WV to detect abnormal driving vehicles remotely or simultaneously in parallel. Note that if the camera 10 is installed at a location remote from the information processing device 20, a communication circuit will be required to remotely operate the camera 10 and its associated drive circuitry, and a support stand will also be required to set the camera 10 at a relatively high viewpoint position.

[0064] The width of the zone ZO is not limited to 10 and can be changed as appropriate depending on the resolution of the camera 10, the photographing angle of the road RO, and the like.

[0065] In the above explanation, it is assumed that the distribution of the speed data VD obtained by learning is a Gaussian distribution, but other types of probability density functions can be used to determine abnormal driving. Furthermore, when a Gaussian distribution type probability density function is used, abnormal driving determination is not limited to those using the mean value or standard deviation, and the median or other parameters can also be used. [Explanation of symbols]

[0066] 2a...loading platform, 2b...driver's seat, 3...sign, 4...warning sign, 4a...warning light, 10...camera, 20...information processing device, 20a...information processing device main body, 21...main control device, 22...storage device, 23...user interface, 24...communication device, 31...image processing device, 32a...section speed recording device, 32b...collecting device, 32c...determination device, 32d...storage unit, 33...learning processing unit, 34...detection processing unit, 35...user interface unit, 37...collecting table, 100...abnormal driving detection device, BP...base point, FI...column, HA...area, IM...screen, LA1,LA2...area, LL...approximate curve, LZ1,LZ2,LZ3...zone, MI...measurement information, OI...target image, OMI...target measurement information, P0...preliminary image, PA...shooting range, PI...captured image, R1, R2, R3... range, RL1, RL2... lane, RO... road, SD... aggregation result, SI... prior image, ST, OT... tracking information, V1, V2, V3... vehicle, VD... speed data, VO... vehicle to be judged, WV... vehicle marked with a sign, Y1, Y2, Y3... starting point, Y2... subsequent point for starting point Y1, Y3... subsequent point for starting point Y2, etc., ZO... zone, ZV... speed, ZY... representative value

Claims

1. a zone speed recording device that classifies the amount of movement of the vehicle detected on the screen into one of a plurality of zones and stores measurement information including the zone and the speed in a memory unit; a collection device that collects the measurement information obtained for a plurality of vehicles; a determination device that compares the measurement information obtained regarding the vehicle to be determined with the tabulation result by the tabulation device and determines whether the speed of the vehicle to be determined is within an allowable range; An abnormal driving detection device comprising:

2. the zone speed recording device classifies the movement amounts from a plurality of base points into the zones, each of which is a starting point of the series of movement amounts of the detected vehicle and a subsequent point obtained by sequentially shifting the starting point as a base point; The abnormal driving detection device according to claim 1 .

3. the zone speed recording device uses a starting point of the detected series of vehicle movement amounts as a base point and divides the movement amounts from the base point into the zones; The abnormal driving detection device according to claim 1 .

4. the determination device determines that the speed of the vehicle to be determined is outside the allowable range when the speed of the vehicle to be determined deviates from the average speed of the corresponding zone by a predetermined value or more, using an average value of the speeds in each zone obtained for a plurality of vehicles as a reference; The abnormal driving detection device according to claim 1 .

5. the predetermined value is set based on the standard deviation of the speed distribution in the zone obtained by the counting device. The abnormal driving detection device according to claim 4.

6. the determination device determines that the speed of the vehicle to be determined is outside the allowable range when a proportion of speed data obtained about the vehicle to be determined that is outside a standard range corresponding to the standard deviation of the distribution of the speeds in the corresponding zone is equal to or greater than a predetermined value; The abnormal driving detection device according to claim 5.

7. the determination device determines the speed obtained for a zone having a relatively large amount of movement with priority over the speed obtained for a zone having a relatively small amount of movement; The abnormal driving detection device according to claim 1 .

8. the determination device makes a determination based on the speed obtained for the zone with the largest amount of movement; The abnormal driving detection device according to claim 7.

9. the section speed recording device uses one of two coordinate axes that identify a position on the screen when determining the amount of movement on the screen; The abnormal driving detection device according to claim 1 .

10. A speed zone recording device is used to classify the amount of movement of the vehicle detected on the screen into one of a plurality of zones, and to store measurement information including the classified zone and the speed in a storage unit, Acquiring and aggregating the measurement information for a plurality of vehicles; comparing the measurement information obtained for the vehicle to be determined with a summary of the measurement information obtained for the plurality of vehicles to determine whether the speed of the vehicle to be determined is within an allowable range; Abnormal driving detection method.

Images