JP6439571B2 - Traffic information collecting device, traffic information collecting method, and traffic information collecting program - Google Patents

Description

  The present invention relates to a technique for collecting traffic information such as travel time of a vehicle between two points.

  Conventionally, the travel time of a vehicle between two points on a road network is acquired as traffic information.

  For example, there is one that reads a license plate of a vehicle at two points and acquires a travel time from a time difference between passage times of the same vehicle at the two points (see Non-Patent Document 1). Reading the license plate of the vehicle is performed by image processing such as character recognition on the captured image of the vehicle captured by the camera.

  In addition, the captured image of the vehicle imaged by the camera installed at two points is processed, the vehicle width, the vehicle height, the vehicle length, the hue, etc. are extracted as the feature amount of the vehicle, the vehicle is collated with the extracted feature amount, A vehicle detected at one point and a vehicle detected at the other point are identified. There is also a proposal for acquiring travel time from the time difference between the passage times of two points in the identified vehicle (see Non-Patent Document 2).

http://www.utms.or.jp/japanese/cont/syusyuu/index6.html http://library.jsce.or.jp/jsce/open/00039/200906_no39/pdf/93.pdf

  However, since the system according to Non-Patent Document 1 is configured to perform image processing for character recognition and the like, it requires a computer with high processing capability and is expensive.

  In addition, the system according to Patent Document 2 is a vehicle having the same feature value (the same model and the same color) between two points where travel time is acquired, or a vehicle having a similar feature value (different models but different shapes, As the number of vehicles with similar colors increases, the vehicle identification accuracy decreases (the frequency of occurrence of a situation in which a vehicle detected at one point is mistakenly identified as another vehicle detected at the other point). Higher.) The possibility of an increase in the number of vehicles having the same feature quantity between two points where travel time is acquired or vehicles having similar feature quantities increases as the number of vehicles existing between the two points increases. Accordingly, as the degree of congestion between the two points for acquiring the travel time increases, the vehicle identification accuracy decreases, and as a result, the accuracy of the travel time acquired for the two points also decreases.

  An object of the present invention is to provide a technique for accurately obtaining a travel time between two points while suppressing a decrease in identification accuracy for identifying a vehicle detected at one point as another vehicle detected at the other point. It is in.

  In order to achieve the above object, the traffic information collecting apparatus of the present invention is configured as follows.

  An inflow vehicle detection part processes the image which imaged the entrance of the link, and detects the vehicle which flows into this link. The outflow vehicle detection unit processes an image obtained by imaging the exit of the link, and detects a vehicle flowing out from the link. The travel time acquisition unit collates the vehicle detected by the inflow vehicle detection unit with the vehicle detected by the outflow vehicle detection unit, and the time detected by the inflow vehicle detection unit for the identified vehicle, and the outflow vehicle detection unit detects Get travel time of link from time. That is, the travel time acquisition unit acquires a time difference between the time when the identified vehicle passes the link entrance and the time when the identified vehicle passes the link exit as the travel time of the link.

  The staying number estimating unit estimates the staying number of vehicles staying on the link. The number of vehicles staying at the link (the number of staying vehicles) may be a number obtained by subtracting the number of vehicles detected at the exit of the link from the number of vehicles detected at the entrance of the link. In this case, if an inflow path through which the vehicle flows in or an outflow path through which the vehicle flows out from this link is connected between the entrance and the exit of the link, the inflow path flows into the link. The difference between the number of vehicles and the number of vehicles that have flowed out of the link to the outflow path is an estimation error of the staying number. As a technique for suppressing this estimation error, there is a technique (Patent No. 5692323) using a correction coefficient correspondence table in which a correction coefficient for correcting the number of stays is associated with each time zone. In addition, the number of vehicles staying on the link may be estimated using the travel time of the link.

  The travel time prediction unit predicts a predicted value of the travel time of the vehicle on the link using the staying number estimated by the staying number estimation unit. Generally, as the number of vehicles staying on the link increases, the travel time of the link becomes longer. That is, the travel time of this link can be roughly predicted by using the number of vehicles staying on the link.

  The detection time range determination unit is a travel time acquisition unit that matches a vehicle detected by one of the inflow vehicle detection unit or the outflow vehicle detection unit and detects the vehicle detected by the other of the inflow vehicle detection unit or the outflow vehicle detection unit. The range of the detection time of the vehicle in the other of the inflow vehicle detection unit or the outflow vehicle detection unit is determined based on the predicted value of the travel time of the vehicle in the link predicted by the travel time prediction unit. That is, the travel time acquisition unit detects the vehicle detected by one of the inflow vehicle detection unit or the outflow vehicle detection unit within the detection time range determined by the detection time range determination unit, or detects the inflow vehicle detection unit. Check with the vehicle detected on the other side.

  Since the detection time range determination unit determines the detection time range based on the predicted value of the travel time of the vehicle on the link predicted by the travel time prediction unit, it is detected by either the inflow vehicle detection unit or the outflow vehicle detection unit. The possibility that the detected vehicle is detected by the other of the inflow vehicle detection unit or the outflow vehicle detection unit in the detection time range determined by the detection time range determination unit can be ensured. Therefore, the inflow vehicle detection unit or the outflow vehicle detection unit collates the vehicle detected by one of the inflow vehicle detection unit or the outflow vehicle detection unit and efficiently limits the vehicle detected by the other. A decrease in identification accuracy between the detected vehicle and the vehicle detected by the outflow vehicle detection unit is suppressed.

  A first vehicle extraction unit that extracts a vehicle that satisfies a predetermined extraction condition from vehicles detected by the inflow vehicle detection unit as a vehicle that the travel time acquisition unit uses for verification, and an outflow vehicle detection unit detects You may make it the structure provided with the 2nd vehicle extraction part which extracts the vehicle which satisfies extraction conditions from a vehicle as a vehicle which a travel time acquisition part uses for collation. If comprised in this way, the vehicle detected by the other of the inflow vehicle detection part or the outflow vehicle detection part which collates the vehicle detected by one of the inflow vehicle detection part or the outflow vehicle detection part is limited more efficiently. Further, a decrease in identification accuracy between the vehicle detected by the inflow vehicle detection unit and the vehicle detected by the outflow vehicle detection unit is further suppressed.

In addition , a state determination unit that determines whether the link exit is in a non-saturated state or a super-saturated state, and the detection time range determination unit determines that the link exit is in a super-saturated state. When the travel time acquisition unit compares the vehicle detected by one of the inflow vehicle detection unit or the outflow vehicle detection unit with the vehicle detected by the other of the inflow vehicle detection unit or the outflow vehicle detection unit, the inflow vehicle detection unit or it increases the range of detection time of the vehicle in the other outflow vehicle detection unit.

Thus , the vehicle detected by one of the inflow vehicle detection unit or the outflow vehicle detection unit is detected by the other of the inflow vehicle detection unit or the outflow vehicle detection unit within the detection time range determined by the detection time range determination unit. Can improve the possibility of being.

  Furthermore, the travel time acquisition unit is configured to check the vehicle in the link predicted by the travel time prediction unit when the vehicle cannot be identified in the comparison between the vehicle detected by the inflow vehicle detection unit and the vehicle detected by the outflow vehicle detection unit. The predicted travel time value may be the travel time of the link.

  According to the present invention, it is possible to suppress a decrease in identification accuracy for identifying a vehicle detected at one point as another vehicle detected at the other point, and to accurately acquire the travel time between the two points.

It is the schematic which shows a traffic control system. It is a block diagram which shows the structure of the principal part of a traffic control center. It is a block diagram which shows the structure of the principal part of a traffic information collection unit. It is a block diagram which shows the structure of the principal part of a vehicle sensor. It is a flowchart which shows operation | movement of a vehicle sensor. It is a flowchart which shows a detection time range determination process. It is a figure explaining a detection time range. It is a flowchart which shows a collation process. It is a flowchart which shows a staying number estimation process. It is a figure which shows the fluctuation | variation of the travel time of the object link acquired with the traffic control system concerning this example.

  Embodiments of the present invention will be described below.

  FIG. 1 is a schematic diagram showing a traffic control system according to this example. Here, a description will be given by taking as an example a traffic control system that acquires (estimates) the travel time of the target link and the number of vehicles staying on the target link shown in FIG. 1 as traffic information. The target link is a road that connects intersection B and intersection A as shown in FIG. 1, and is a road on which a vehicle that passes through intersection B and heads for intersection A travels. Although not shown in FIG. 1, in the middle of the target link (between the intersection A and the intersection B), an inflow path through which the vehicle flows into the target link and an outflow path (hereinafter referred to as the inflow path) from which the vehicle flows out. , And the outflow path are collectively referred to as a branch path). There may be one or more branch paths connected to the target link. Therefore, among the vehicles that have passed through the intersection B and entered the target link, there are vehicles that flow out from the target link (vehicles that do not pass through the intersection A) on the branch road that is connected on the way. In addition, there is a vehicle that does not pass through the intersection B and flows into the target link from a branch road that is connected on the way and passes through the intersection A.

  This traffic control system includes a traffic control center 1, a signal controller 2 (2a, 2b), and a vehicle detector 3 (3a, 3b). At the intersection A, a signal lamp 5a is installed for each approach direction of the vehicle. At the intersection B, a signal lamp 5b is installed for each approach direction of the vehicle.

  The traffic control center 1 inputs a signal control parameter for displaying the signal lamp 5a to the signal controller 2a. In addition, the traffic control center 1 inputs a signal control parameter related to the display of the signal lamp 5b to the signal controller 2b. Further, the traffic control center 1 includes a traffic information collection unit that acquires the travel time of the target link and the number of vehicles staying on the target link as traffic information. Details of the traffic information collection unit will be described later.

  The signal controller 2a controls the display display of the signal lamp 5a installed at the intersection A based on the signal control parameter input from the traffic control center 1. The signal controller 2b controls the display indication of the signal lamp 5b installed at the intersection B based on the signal control parameter input from the traffic control center 1. The signal control parameter indicates a cycle indicating the time of one cycle of the signal lamp 5 (5a, 5b), a split indicating the ratio of each indication closed in one cycle (one cycle), and a difference in cycle start timing between the intersections. Offset etc. are included.

  The vehicle sensor 3 is an image type vehicle sensor. The vehicle detector 3a processes an image obtained by imaging the exit of the target link, and detects a vehicle that has flowed out of the target link to the intersection A. The vehicle detector 3b processes an image obtained by imaging the entrance of the target link, and detects a vehicle that has passed through the intersection B and entered the target link. In this example, the vehicle detector 3a is connected to the traffic control center 1 via a signal controller 2a installed near the intersection A. The vehicle detector 3b is connected to the traffic control center 1 via a signal controller 2b installed near the intersection B.

  The vehicle detector 3 (3a, 3b) may be configured to be directly connected to the traffic control center 1 without passing through the signal controller 2 (2a, 2b), or a device other than the signal controller 2 It may be configured to connect to the traffic control center 1 via

  FIG. 2 is a block diagram showing the configuration of the main part of the traffic control center. The traffic control center 1 includes a control unit 11, a traffic information collection unit 12, a signal control parameter generation unit 13, and an input / output unit 14.

  The control unit 11 controls the operation of each part of the traffic control center 1 main body.

  The traffic information collection unit 12 acquires the travel time of the target link shown in FIG. 1 and the number of vehicles staying in the target link. In this example, the traffic information collecting unit 12 and the vehicle detector 3 (3a, 3b) constitute a traffic information collecting device according to the present invention.

  The signal control parameter generation unit 13 generates signal control parameters for the signal lamps 5a and 5b installed at the intersections A and B based on the number of vehicles staying on the target link estimated by the traffic information collection unit 12. .

  The input / output unit 14 inputs and outputs data with the signal controllers 2a and 2b. The input / output unit 14 receives the output of the vehicle detector 3a (details will be described later) via the signal controller 2a and the output of the vehicle detector 3b (details will be described later). It is input via the signal controller 2b. The input / output unit 14 outputs the signal control parameters of the signal lamps 5a and 5b to the signal controllers 2a and 2b.

  FIG. 3 is a block diagram showing the configuration of the main part of the traffic information collection unit. The traffic information collection unit 12 includes a control unit 21, an inflow number counting unit 22, an outflow number counting unit 23, a non-saturated state determination unit 24, a travel time acquisition unit 25, a detection time range determination unit 26, and a stay A number estimating unit 27.

  The control unit 21 controls the operation of each part of the main body of the traffic information collection unit 12. The control unit 21 performs communication (input / output) with the control unit 11.

  The inflowing number counting unit 22 counts the number of vehicles that have passed through the intersection B and entered the target link based on the output of the vehicle detector 3b input through the signal controller 2b. The outflow number counting unit 23 counts the number of vehicles that have flowed out to the intersection A from the target link based on the output of the vehicle detector 3a input via the signal controller 2a.

  The non-saturated state determination unit 24 determines whether the state of the target link is a non-saturated state or a super-saturated state. The non-saturated state is a state where there is no traffic jam, and the super-saturated state is a state where there is a traffic jam.

  The travel time acquisition unit 25 collates the vehicle that has flowed into the target link detected by the vehicle sensor 3b with the vehicle that has flowed out of the target link detected by the vehicle sensor 3a, and is detected by the vehicle sensor 3b. By identifying the vehicle and the vehicle detected by the vehicle detector 3a, the travel time of the vehicle on the target link is acquired. Specifically, the travel time acquisition unit 25 includes a time when the vehicle flows into the target link (time detected by the vehicle sensor 3b) and a time when the vehicle flows out from the target link (time detected by the vehicle sensor 3a). Is obtained as the travel time of the vehicle on the target link.

  The detection time range determination unit 26 determines a time zone (detection time range) where the vehicle detector 3a is likely to detect the vehicle detected by the vehicle detector 3b. The detection time range determination unit 26 may determine the detection time range for all the vehicles detected by the vehicle sensor 3b. In this example, the detection time range determination unit 26 is a vehicle detected by the vehicle sensor 3b, and is a predetermined extraction. A range of detection times is determined for vehicles that satisfy the conditions. In other words, even if the vehicle is detected by the vehicle sensor 3b, the detection time range is not determined for a vehicle that does not satisfy the predetermined extraction condition. The detection time range determination unit 26 includes a memory that stores a vehicle detection signal, which will be described later. This memory may be provided in the control unit 21.

  The staying number estimating unit 27 passes the intersection B counted by the inflowing number counting unit 22 and flows into the target link, and the number of vehicles flowing out to the intersection A from the target link counted by the outflowing number counting unit 23. Is used to estimate the number of vehicles staying at the target link.

  FIG. 4 is a block diagram showing a configuration of a main part of the vehicle detector. In this example, the vehicle detectors 3a and 3b have the same configuration and will be described together. The vehicle detector 3 includes a control unit 31, an imaging unit 32, an image processing unit 33, a vehicle detection signal generation unit 34, and an output unit 35.

  The control unit 31 controls the operation of each part of the vehicle detector 3 main body.

  The imaging unit 32 is a video camera that captures a moving image in the imaging area. The imaging unit 32 captures a color image. The frame rate of the imaging unit 32 may be about 10 frames / sec. The vehicle detector 3a is installed so that the imaging area of the imaging unit 32 is the exit of the target link. The vehicle detector 3b is installed so that the imaging area of the imaging unit 32 becomes the entrance of the target link.

  The image processing unit 33 processes the frame image of the imaging area imaged by the imaging unit 32 and detects the imaged vehicle. The vehicle is detected by image processing such as a known background difference. The image processing unit 33 performs a conversion process from RGB to HSV for each detected vehicle. The image processing unit 33 obtains a distribution of hue (H), saturation (S), and brightness (V) for the detected vehicle.

  The vehicle detection signal generation unit 34 generates a detection signal for each vehicle detected by the image processing unit 33. The vehicle detection signal includes the vehicle ID, the vehicle detection time (the imaging time of the frame image in which the vehicle was detected), the hue (H) distribution of the vehicle, and the saturation (S) of the vehicle. Is associated with the distribution of brightness (V) of the vehicle.

  The output unit 35 outputs a vehicle detection signal each time the vehicle detection signal generation unit 34 generates a vehicle detection signal. The vehicle detection signal output from the output unit 35 is input to the traffic control center 1 via the signal controller 2.

  Next, the operation of the traffic control system according to this example will be described. Here, an operation for obtaining the travel time of the target link and an operation for estimating the number of vehicles staying in the target link (the number of staying) will be described.

  First, the operation of the vehicle detector 3 will be described. FIG. 5 is a flowchart showing the operation of the vehicle detector. The vehicle detectors 3a and 3b differ only in whether the vehicle to be detected is a vehicle that flows out of the target link or a vehicle that flows into the target link, and repeatedly executes the processing shown in FIG.

  When the image processing unit 33 detects the vehicle imaged in the frame image of the imaging area imaged by the imaging unit 32 (s1), the vehicle sensor 3 performs a conversion process from RGB to HSV for this vehicle (s2). ). The image processing unit 33 detects the vehicle related to s1 by image processing such as a known background difference. Also, the conversion process from RGB to HSV for s2 is performed by a calculation using a known conversion formula.

  In the vehicle detector 3, the vehicle detection signal generation unit 34 generates a vehicle detection signal detected this time (s3). In s3, the ID of the vehicle detected this time, the imaging time of the frame image in which this vehicle was detected (the detection time of this vehicle), the distribution of the hue (H) of this vehicle, and the saturation (S) of this vehicle The vehicle detection signal is generated by associating this distribution with the distribution of the vehicle brightness (V).

  The vehicle detector 3 outputs the vehicle detection signal generated in s3 at the output unit 35 (s4), and returns to s1. The vehicle detection signal output in s4 is input to the traffic control center 1 via the signal controller 2 (2a or 2b).

  Thus, the vehicle detector 3 inputs the vehicle detection signal generated for the detected vehicle to the traffic control center 1 for each detected vehicle.

  Next, the operation of the traffic information collection unit 12 will be described. The traffic information collecting unit 12 counts up the number of vehicles that have flowed into the target link by 1 when the inflow number counting unit 22 receives a vehicle detection signal from the vehicle detector 3b, and the outflow number counting unit 23 detects the vehicle. When a vehicle detection signal is input from the device 3a, the number of vehicles that have flowed out of the target link is counted up by one.

  Further, the traffic information collecting unit 12 obtains the travel time of the vehicle on the target link by comparing the vehicle detected by the vehicle detector 3a with the vehicle detected by the vehicle detector 3b. The process for acquiring the travel time is performed by the following detection time range determination process and collation process. FIG. 6 is a flowchart showing the detection time range determination process.

  The detection time range determination process will be described with reference to FIG. The detection time range determination unit 26 performs this detection time range determination process. When the detection signal of the vehicle is input from the vehicle detector 3b (s11), the detection time range determination unit 26 determines whether the vehicle according to the input detection signal satisfies the predetermined extraction condition. (S12). If the detection time range determination unit 26 determines that the extraction condition is not satisfied in s12, the detection time range determination unit 26 returns to s11.

  In step s12, it is determined whether the vehicle that has flowed into the target link satisfies the extraction condition at the entrance of the target link, and the vehicle that has flowed out of the target link satisfies the extraction condition at the exit of the target link. It does not mean whether or not to do.

  The extraction condition is, for example, a distribution of hue (H), saturation (S), and brightness (V), and is a condition for extracting a red vehicle. The extraction condition may be a condition defined for at least one of the distribution of hue (H), the distribution of saturation (S), and the distribution of lightness (V). The extraction condition is preferably such that about 10% (about 10%) of the vehicles detected by the vehicle detector 3b are satisfied. The reason is that if the rate extracted from the vehicle detected by the vehicle detector 3b is too low, the travel time acquisition frequency described later decreases. On the other hand, if the frequency extracted from the vehicle detected by the vehicle detector 3b is too high, the processing load after s13 increases and the processing load of collation processing described later also increases. In addition, the number of matching combinations increases, and the probability of erroneous matching occurring increases.

Moreover, if the detection time range determination part 26 determines with satisfy | filling extraction conditions by s12, it will estimate the travel time of this vehicle in an object link (s13). The travel time predicted in s13 is referred to as predicted travel time. Estimated travel time is
Predicted travel time = travel time at non-saturation + number of staying vehicles × passage time of intersection A per vehicle.

The travel time at the time of non-saturation in the formula for calculating the predicted travel time is determined in advance. Specifically, the travel time at the time of non-saturation is a time calculated from the length of the target link and the average speed of the vehicle traveling on the target link at the time of non-saturation. The staying number is the number of vehicles staying on the target link estimated by the staying number estimation process described later. Furthermore, the passing time of intersection A per vehicle is
Passing time of intersection A per vehicle = (time of one cycle of signal lamp 5a) / ((presenting time for vehicle group of target link) × (saturated traffic flow rate of intersection A))
Is calculated by

  The time of one cycle of the signal lamp 5a and the current display time for the vehicle group of the target link are obtained from the signal control parameters of the signal lamp 5a output to the signal controller 2a. The saturation traffic flow rate at the intersection A may be set in advance, or a value measured by this system may be used. Various methods for measuring the saturated traffic flow rate have already been proposed, and a description thereof will be omitted here.

  The detection time range determination unit 26 determines whether the target link is currently non-saturated or super-saturated (s14). The traffic information collecting unit 12 repeatedly determines whether the target link is non-saturated or super-saturated in the non-saturated state determination unit 24 for each cycle of the signal lamp 5a. Specifically, the non-saturated state determination unit 24 calculates the blue hour usage rate of the vehicle that has flowed out from the target link to the intersection A for each cycle of the signal lamp 5a, and the calculated blue time usage rate is predetermined. If it is equal to or greater than the threshold, it is determined that the target link is in a supersaturated state. On the other hand, if the calculated blue hour usage rate is less than the threshold, it is determined that the target link is in a non-saturated state.

The blue hour utilization rate is
Blue hour utilization rate = (Number of vehicles that have flowed out from the target link to the intersection A) / ((Display time for the vehicle group of the target link during one cycle of the signal lamp 5a) × (saturated traffic flow rate at the intersection A))
Is calculated by

When the detection time range determination unit 26 determines that the target link is in a non-saturated state in s14, the detection time range in the case of non-saturation based on the predicted travel time calculated in s13 for the vehicle determined to satisfy the extraction condition in s12. Is determined (s15). In s15, from the time when this vehicle is detected by the vehicle sensor 3b, the time range of one cycle of the ± signal lamp 5a is determined as the detection time range centering on the time when the estimated travel time calculated in s13 has elapsed. Specifically, as shown in FIG. 7A, when the time when this vehicle is detected by the vehicle detector 3b is Tb, the detection time range is (Tb + predicted travel time−one cycle of the signal lamp 5a). From ( time), ( Tb + predicted travel time + time of one cycle of the signal lamp 5a) is determined.

  When the detection time range determination unit 26 determines that the target link is in a supersaturated state in s14, the detection time range for the supersaturation based on the predicted travel time calculated in s13 for the vehicle determined to satisfy the extraction condition in s12. Is determined (s16). In s16, from the time when the vehicle is detected by the vehicle sensor 3b, the time range of the 1.2 cycle of the ± signal lamp 5a is determined as the detection time range around the time when the estimated travel time calculated in s13 has elapsed. Specifically, as shown in FIG. 7B, when the time when the vehicle is detected by the vehicle detector 3b is Tb, the detection time range is (Tb + predicted travel time−1.2 cycles of the signal lamp 5a). From (time), it is determined as (Tb + predicted travel time + 1.2 cycle time of the signal lamp 5a).

  As described above, the detection time range determination unit 26 makes the detection time range when the target link is in a supersaturated state wider than the detection time range when the target link is in a non-saturated state.

  Here, the detection time range at the time of non-saturation in s15 and the detection time range at the time of over-saturation in s16 are determined using the time of one cycle of the signal lamp 5a, but one cycle of the signal lamp 5a The time may be replaced with a predetermined time or the like. Further, although the detection time range at the time of oversaturation is set to 1.2 times the detection time range at the time of non-saturation, this magnification may be other magnifications.

  The detection time range determination unit 26 adds and stores the detection time range determined in s15 or s16 to the detection signal of the vehicle determined to satisfy the extraction condition in s12 (s17), and returns to s11.

  That is, the vehicle detection signal stored in the detection time range determination unit 26 is a detection signal for a vehicle that is determined to satisfy the extraction condition in s12.

  Next, the matching process will be described with reference to FIG. The travel time acquisition unit 25 performs this verification process. When the vehicle detection signal is input from the vehicle detector 3a (s21), the travel time acquisition unit 25 determines whether the vehicle related to the input detection signal is a vehicle that satisfies a predetermined extraction condition. (S22). If the travel time acquisition unit 25 determines that the extraction condition is not satisfied in s22, the travel time acquisition unit 25 returns to s21.

  In s22, it is determined whether the vehicle that has flowed out of the target link satisfies the extraction condition at the exit of the target link, and the vehicle that has flowed into the target link satisfies the extraction condition at the entrance of the target link. It does not mean whether or not to do.

  The determination relating to s22 is the same as s12 described above, and the extraction conditions are also the same as the extraction conditions used in s12.

  If the travel time acquisition unit 25 determines that the extraction condition is satisfied in s22, the travel time acquisition unit 25 reads the detection time when the vehicle is detected by the vehicle detector 3a from the detection signal of the vehicle (s23).

  The travel time acquisition unit 25 extracts all the vehicles in which the detection time of the vehicle read in s23 is within the detection time range from the vehicles for which the detection time range determination unit 26 stores the detection signal (s24).

  For each vehicle extracted in s24, the travel time acquisition unit 25 determines the distribution of hue (H), saturation (S), and brightness (V) included in the detection signal of this vehicle, and the extraction condition in s22. The hue (H), saturation (S), and lightness (V) distributions included in the detection signal of the vehicle determined to be satisfied are collated, and the similarity is calculated (s25).

  The travel time acquisition unit 25 determines whether or not there is a set of vehicles whose similarity is equal to or higher than a predetermined similarity threshold by collation according to s25 (s26). The travel time acquisition unit 25 identifies the vehicle determined to satisfy the extraction condition in 22 as the vehicle set having the maximum similarity if there is a vehicle set having a similarity greater than or equal to a predetermined similarity threshold. (S27). The travel time acquisition unit 25 acquires the time difference between the detection time of the vehicle sensor 3b and the detection time of the vehicle sensor 3a for the vehicle identified in s27 as the travel time of the target link (s28).

  Further, if there is no set of vehicles whose similarity is equal to or higher than a predetermined similarity threshold in the collation related to s25, the travel time acquisition unit 25 is a fixed time from when the travel time of the target link was previously acquired in s28. It is determined whether or not (travel time update cycle, for example, 5 minutes or 10 minutes) has elapsed (s29). The travel time acquisition unit 25 returns to s21 if a predetermined time has not elapsed since the travel time of the target link was previously acquired in s28. In addition, when a certain time has elapsed since the travel time of the target link was last acquired in s28, the travel time acquisition unit 25 acquires the predicted travel time calculated last time in s13 as the travel time of the target link (s30). .

  When the travel time acquisition unit 25 acquires the travel time in s28 or s30, the travel time acquisition unit 25 updates the travel time of the target link to the travel time acquired this time (s31), and returns to s21.

  As described above, the traffic information collecting unit 12 is configured to limit the vehicles detected at the entrance of the target link to be compared with the vehicles detected at the exit of the target link within the detection time range. It is possible to prevent the vehicle from being identified as another vehicle detected at the entrance of the target link. Therefore, the travel time of the target link can be obtained with high accuracy.

  Further, since the vehicle detected at the entrance of the target link to be compared with the vehicle detected at the exit of the target link is limited within the detection time range, the load on the verification process can be suppressed.

  In addition, since the vehicles to be collated are limited to those that satisfy the predetermined extraction condition, the load on the collation process can be further suppressed.

  In the verification process in the above example, the vehicle detected by the vehicle sensor 3a is checked against one or more vehicles detected by the vehicle sensor 3b. However, the vehicle detected by the vehicle sensor 3b is You may make it collate with the 1 or several vehicle which the container 3a detected. That is, for each vehicle extracted as satisfying the extraction condition in s12, this vehicle may be collated with one or a plurality of vehicles detected by the vehicle detector 3a within the detection time range determined for this vehicle. . Further, a detection time range (a time range expected to be detected by the vehicle sensor 3b) may be determined for the vehicle detected by the vehicle sensor 3a.

  Next, the staying number estimation process for estimating the number of vehicles staying at the target link will be described. FIG. 9 is a flowchart showing the staying number estimation process. This staying number estimation process is performed for each cycle of the signal lamp 5a. The staying number estimating unit 27 determines whether the target link is currently unsaturated or supersaturated (s41). As described above, the traffic information collection unit 12 repeatedly determines in the non-saturated state determination unit 24 whether the target link is in a non-saturated state or a super-saturated state. The process according to s41 is the same as the process according to s13 described above. However, s13 is performed in the detection time range determination unit 26, and s41 is performed in the staying number estimation unit 27.

  When the staying number estimating unit 27 determines that the target link is in a non-saturated state, the staying number estimating unit 27 updates the number of vehicles staying in the target link to 0 (s42).

On the other hand, if the staying number estimating unit 27 determines that the target link is in a supersaturated state, the staying number estimating unit 27 calculates the number of vehicles staying in the target link (s43). The number of vehicles staying at the target link (number of staying vehicles)
Number of stays = (delay time due to stay) / (passing time of intersection A per vehicle)
Calculated by

As described above, the passing time of intersection A per vehicle is
(Passing time of intersection A per vehicle) = (time of one cycle of the signal lamp 5a) / ((presenting time for vehicle group of target link) × (saturated traffic flow rate of intersection A))
Is calculated by

Also, the delay time due to residence is
Delay time due to stay = (Travel time at that time)-(Travel time at non-saturation)
Is calculated by The travel time during non-saturation is determined in advance as described above.

  The staying number estimating unit 27 sets the number of vehicles staying at the target link to 0 when the delay time due to staying is a negative value.

  The staying number estimation unit 27 updates the number of vehicles staying in the target link to the staying number calculated in s43 (s44), and returns to s41.

  Moreover, the estimation of the staying number in the staying number estimating unit 27 may be performed by the process described in Japanese Patent No. 569323. However, if the travel time is calculated in s28 described above in the previous cycle of the signal lamp 5a, it is preferable to calculate the number of staying in s43.

  FIG. 10 is a diagram showing a change in travel time of the target link acquired by the traffic control system according to this example. In FIG. 10, times t1 and t2 are timings when travel time is acquired in s28.

  As shown in FIG. 10, when the travel time is calculated based on the number of staying vehicles staying on the target link, an error in the travel time to be acquired occurs due to an error in the estimated number. On the other hand, in the system according to this example, the travel time becomes substantially true at the timing when the travel time is calculated in s28. Further, since the travel time becomes substantially true, an error in which the staying number of the target links estimated by the staying number estimating unit 27 is accumulated is suppressed. Therefore, since the accuracy of the predicted travel time of the target link is improved and the detection time range is appropriately determined, it is possible to sufficiently suppress a decrease in the identification accuracy in the matching process.

DESCRIPTION OF SYMBOLS 1 ... Traffic control center 2 (2a, 2b) ... Signal controller 3 (3a, 3b) ... Vehicle detector 5 (5a, 5b) ... Signal lamp 11 ... Control unit 12 ... Traffic information collection unit 13 ... Signal control parameter generation Unit 14 ... Input / output unit 21 ... Control unit 22 ... Inflowing number counting unit 23 ... Outflowing number counting unit 24 ... Unsaturated state determination unit 25 ... Travel time acquisition unit 26 ... Detection time range determination unit 27 ... Remaining number estimation unit 31 ... Control unit 32 ... imaging unit 33 ... image processing unit 34 ... vehicle detection signal generation unit 35 ... output unit

Claims (6)

An inflow vehicle detection unit that processes an image obtained by imaging the entrance of the link and detects a vehicle flowing into the link;
An outflow vehicle detection unit that processes an image obtained by imaging the exit of the link and detects a vehicle flowing out from the link;
The vehicle detected by the inflowing vehicle detection unit is compared with the vehicle detected by the outflowing vehicle detection unit, and the time detected by the inflowing vehicle detection unit for the identified vehicle and the time detected by the outflowing vehicle detection unit A travel time acquisition unit for acquiring the travel time of the link;
A staying number estimating unit for estimating a staying number of vehicles staying at the link;
A travel time prediction unit for predicting a predicted value of the travel time of the vehicle in the link using the staying number estimated by the staying number estimation unit;
In the travel time acquisition unit, the vehicle detected by the inflowing vehicle detection unit or the outflowing vehicle detection unit, which is compared with the vehicle detected by one of the inflowing vehicle detection unit or the outflowing vehicle detection unit, A detection time range determination unit that determines a vehicle detection time range at the other of the inflow vehicle detection unit or the outflow vehicle detection unit based on a predicted value of the vehicle travel time at the link predicted by the travel time prediction unit. When,
A state determination unit for determining whether the exit of the link is in a non-saturated state or a super-saturated state,
When the state determination unit determines that the exit of the link is in a supersaturated state, the detection time range determination unit detects one of the inflow vehicle detection unit or the outflow vehicle detection unit in the travel time acquisition unit. For the vehicle detected by the other of the inflow vehicle detection unit or the outflow vehicle detection unit to be compared with the vehicle that has been detected, the vehicle detection time range in the other of the inflow vehicle detection unit or the outflow vehicle detection unit is increased. , traffic information collection device.   The traffic information collection device according to claim 1, wherein the staying number estimation unit estimates the staying number of vehicles staying at the link using the travel time of the link acquired by the travel time acquisition unit. A first vehicle extraction unit that extracts, from the vehicle detected by the inflow vehicle detection unit, a vehicle that satisfies a predetermined extraction condition as a vehicle that the travel time acquisition unit uses for verification;
3. A second vehicle extraction unit that extracts, from the vehicle detected by the outflow vehicle detection unit, a vehicle that satisfies the extraction condition as a vehicle used by the travel time acquisition unit for verification. Traffic information collection device. The travel time acquisition unit predicts the travel time prediction unit when the vehicle is not identified in the comparison between the vehicle detected by the inflow vehicle detection unit and the vehicle detected by the outflow vehicle detection unit. the predicted value of the travel time of the vehicle in the link travel time of the link, the traffic information collecting device according to any of claims 1-3. An inflow vehicle detection step of processing an image obtained by imaging the entrance of the link and detecting a vehicle flowing into the link;
An outflow vehicle detection step for processing an image obtained by imaging the exit of the link and detecting a vehicle flowing out from the link;
The vehicle detected in the inflowing vehicle detection step and the vehicle detected in the outflowing vehicle detection step are collated, and the identified vehicle is detected from the time detected in the inflowing vehicle detection step and the time detected in the outflowing vehicle detection step. A travel time acquisition step of acquiring the travel time of the link;
A staying number estimating step for estimating the staying number of vehicles staying at the link;
A travel time prediction step of predicting a predicted value of the travel time of the vehicle in the link using the staying number estimated by the staying number estimating step;
In the travel time acquisition step, the vehicle detected in the inflowing vehicle detection step or the outflowing vehicle detection step is compared with the vehicle detected in the inflowing vehicle detection step or the outflowing vehicle detection step. A detection time range determination step for determining a vehicle detection time range in the other of the inflow vehicle detection step or the outflow vehicle detection step based on the predicted value of the travel time of the vehicle in the link predicted in the travel time prediction step. When,
A state determination step of determining whether the exit of the link is non-saturated or super-saturated,
In the detection time range determination step, when it is determined in the state determination step that the exit of the link is in a supersaturated state, in the travel time acquisition step, one of the inflow vehicle detection step or the outflow vehicle detection step is detected. For the vehicle detected at the other of the inflowing vehicle detection step or the outflowing vehicle detection step to be compared with the detected vehicle, the range of the vehicle detection time at the other of the inflowing vehicle detection step or the outflowing vehicle detection step is increased. , Traffic information collection method. An inflow vehicle detection step of processing an image obtained by imaging the entrance of the link and detecting a vehicle flowing into the link;
An outflow vehicle detection step for processing an image obtained by imaging the exit of the link and detecting a vehicle flowing out from the link;
The vehicle detected in the inflowing vehicle detection step and the vehicle detected in the outflowing vehicle detection step are collated, and the identified vehicle is detected from the time detected in the inflowing vehicle detection step and the time detected in the outflowing vehicle detection step. A travel time acquisition step of acquiring the travel time of the link;
A staying number estimating step for estimating the staying number of vehicles staying at the link;
A travel time prediction step of predicting a predicted value of the travel time of the vehicle in the link using the staying number estimated by the staying number estimating step;
In the travel time acquisition step, the vehicle detected in the inflowing vehicle detection step or the outflowing vehicle detection step is compared with the vehicle detected in the inflowing vehicle detection step or the outflowing vehicle detection step. A detection time range determination step for determining a vehicle detection time range in the other of the inflow vehicle detection step or the outflow vehicle detection step based on the predicted value of the travel time of the vehicle in the link predicted in the travel time prediction step. When,
A state determination step of determining whether the exit of the link is in a non-saturated state or a super-saturated state ,
In the detection time range determination step, when it is determined in the state determination step that the exit of the link is in a supersaturated state, in the travel time acquisition step, one of the inflow vehicle detection step or the outflow vehicle detection step is detected. For the vehicle detected at the other of the inflowing vehicle detection step or the outflowing vehicle detection step to be compared with the detected vehicle, the range of the vehicle detection time at the other of the inflowing vehicle detection step or the outflowing vehicle detection step is increased. , traffic information collection program.

Images