JP5289256B2 - Vehicle position detection system - Google Patents

Description

  The present invention relates to a vehicle position detection system that detects an illegally parked vehicle using an aerial image and a map, and more particularly, to a new and improved technique for detecting an illegally parked vehicle by comparing an aerial image and map data. Is.

In general, illegal parking is a nuisance to others and must be detected and eliminated immediately.
Conventionally, as a vehicle position detection system for detecting illegal parking, roughly two methods have been proposed.

First, as a first technique, there is a technique in which position measurement means such as GPS is mounted on a target vehicle and the measurement result is used (for example, see Patent Document 1).
In addition, as a second method, there is a method in which a camera or the like is installed at a specific place and the determination is performed by image processing of a reflected image (for example, see Patent Document 2).

  However, in the first method, unless position information from the target vehicle is acquired, it cannot be detected by a third party, and in the second method, only a specific place where a camera or the like is installed is determined. I can't.

As a variation of the second method, a third method is also conceivable in which a camera is mounted on the vehicle and an image is acquired and determined while traveling. However, it is necessary to drive the vehicle in all the monitoring areas. It is not practical because it requires a lot of cost and labor.
That is, according to the conventional vehicle position detection system, it is not possible to discriminate an unspecified number of vehicles from an unspecified number of areas using a practical method.

Therefore, in a vehicle position detection system that detects illegally parked vehicles by the police, as a fourth method, a policeman moves around the city and illegally parks by visual inspection in order to obtain the positions of monitored vehicles in many areas. Is judged.
In this case, there is a problem that (1) the policeman cannot handle illegal parking unless he / she finds it, and (2) the policeman cannot handle illegal parking if he / she moves the vehicle while looking around.

JP 2002-243455 A JP 2008-152736 A

  In the conventional vehicle position detection system, when applied to detection of illegally parked vehicles, the vehicle position is detected by the own vehicle as in the first method in order to detect whether or not the own vehicle is illegally parked. However, in order for a third party such as the police to detect the vehicle position, position information from the vehicle is not always available, and eventually the vehicle position cannot be detected. was there.

  In addition, according to the second method, although the above-mentioned problem is solved by fixing the determination area to a part, it is not possible to determine a vehicle located outside the determination area. Therefore, it is necessary to install a large number of vehicle position detection systems in order to cover a large number of areas, which causes a problem of increasing costs.

  Furthermore, in the fourth method of detecting illegally parked vehicles by the police, it is desirable to make the detection process as invisible as possible to the target person (the owner of the illegally parked vehicle), but the target person is aware of the detection work. In such a case, as in (2) above, the avoidance action of moving the vehicle before being found by the police officer and parking again after the detection work is completed can be selected, and as a result, the illegally parked vehicle can be excluded. There was a problem that it was not possible.

  The present invention has been made to solve the above-described problems, and an object of the present invention is to provide a vehicle position detection system that can easily and automatically detect illegally parked vehicles and reliably eliminate them. .

A vehicle position detection system according to the present invention is a vehicle position detection system that acquires an aerial image obtained by photographing the ground from above, and detects the position of a vehicle subject to illegal parking from the aerial image. A processing unit comprising: a photographing unit having a camera for photographing a captured image; a map data file storing map data corresponding to the aerial image; and a processing device for detecting a vehicle from the aerial image and the map data. Is a map alignment means for aligning an aerial image and map data, a vehicle detection means for detecting a vehicle present in the aerial image, and an illegal to determine whether or not the vehicle is in a parking prohibited area. parking determination means possess a, the map data includes the no-parking area information, the vehicle detecting means detects only the vehicle present in the aerial image and the no-parking area information and overlap portions, The illegal parking determination means determines that only the vehicle detected by the vehicle detection means is an illegal parking vehicle, and the vehicle detection means and the illegal parking determination means are within a parking prohibited area determined from the parking prohibited area information. In addition, the area corresponding to the size of the vehicle is sequentially set so that all the parking prohibition areas are examined, and the maximum value and the minimum value among the RGB information values for each pixel in the area are set. And the number of pixels whose difference is smaller than a predetermined threshold is counted, and when the number counted in the area is smaller than a predetermined number, the area is defined as a parked vehicle. Judgment .

  According to the present invention, by comparing the aerial image and the map data, it is possible to easily and automatically detect the vehicle existing in the parking prohibited area.

1 is a configuration diagram schematically illustrating an entire vehicle position detection system according to a first embodiment of the present invention. FIG. It is explanatory drawing which shows the detection function of the parking prohibition area | region by the processing apparatus in FIG. It is a flowchart which shows the vehicle detection operation | movement of the processing apparatus in FIG. It is a flowchart which shows specifically the area | region determination process in FIG. It is explanatory drawing which shows the detection function of the illegally parked vehicle by the processing apparatus in FIG. It is a block diagram which shows roughly the whole vehicle position detection system which concerns on Embodiment 2 of this invention.

Embodiment 1 FIG.
1 is a block diagram schematically showing an entire vehicle position detection system according to Embodiment 1 of the present invention.
In FIG. 1, the vehicle position detection system includes a helicopter (hereinafter abbreviated as “heli”) 101 that captures an aerial image 501 and a processing device 102 that detects the position of the vehicle from the aerial image 501. .

The helicopter 101 includes a camera that captures the ground imaging region 500 and acquires an aerial image 501 and a communication unit (not shown) that transmits the aerial image 501 together with helicopter position / orientation information.
The processing device 102 stores an image / position receiver 110 that receives transmission information from the helicopter 101, a map data file 510 that stores map data M corresponding to the aerial image 501, and various parameters related to the camera 100. A camera parameter file 511.

  Further, the processing device 102 has, as program functions, a map alignment unit that performs alignment between the aerial image 501 and the map data M, a vehicle detection unit that detects a vehicle existing in the aerial image 501, and a vehicle parked. Illegal parking determination means for determining whether or not the vehicle is in the prohibited area.

Thus, it is possible to detect a monitored vehicle that is illegally parked using the reception information from the image / position receiver 110 and the map data M and camera parameters C stored in the files 510 and 511 held in advance. it can.
Furthermore, the processing apparatus 102 also includes output means (not shown) for displaying the illegally parked vehicle (detection result).

Next, the information acquisition operation and map alignment operation according to the first embodiment of the present invention shown in FIG. 1 will be described with reference to FIG.
FIG. 2 is an explanatory view showing a detection function of the parking prohibited area 600 by the processing device 102.
The helicopter 101 captures the imaging region 500 from above and transmits the aerial image 501 to the processing apparatus 102 as real time information as appropriate. At this time, helicopter position / orientation information 502 at the time of shooting is transmitted together with the aerial image 501.

The processing apparatus 102 receives the aerial image 501 and the helicopter position / orientation information 502 via the image / position receiver 110.
In addition, the processing device 102, from the map data file 510, the range represented by the map (latitude and longitude, etc.), the parking prohibited area information, and characteristic feature information that can be easily distinguished from the sky (position and size of a crossover, etc.) Are taken in as map data M.
Similarly, the processing apparatus 102 captures the shooting specification information of the camera 100 as the camera parameter C from the camera parameter file 511.

In the processing device 102, the map alignment means detects the approximate feature on the map from the helicopter position / orientation information 502, and then processes the aerial image 501 to detect characteristic feature information such as a crossover, Overlay processing of the aerial image 501 and the map data M is performed.
At this time, the positional relationship between the aerial image 501 and the map data M is determined from the position and size in the image of the characteristic feature information and the position and size of the feature in the map data M.

  In addition, the range indicated by the aerial image 501 is determined, and the parking prohibited area 600 (see FIG. 2) in the aerial image 501 is determined by overlapping with the parking prohibited area information included in the map data M.

Next, the vehicle detection operation and illegal parking determination operation after determination of the parking prohibited area 600 will be described with reference to FIGS. 3 to 5 together with FIGS.
The vehicle detection means and the illegal parking determination means in the processing device 102 determine whether or not a illegally parked vehicle exists in the parking prohibited area 600.

  FIG. 3 is a flowchart showing the vehicle position detection operation by the processing device 102, and FIG. 4 is a flowchart specifically showing the color determination process (step S13) in FIG. FIG. 5 is an explanatory diagram showing a function for detecting illegally parked vehicles by the processing device 102.

3 and 4, the processing device 102 performs image processing on the parking prohibited area 600 and detects a vehicle subject to illegal parking.
At the start of the processing in FIG. 3, the aerial image 501, the survey target area (parking prohibited area 600), and the vehicle determination thresholds α and β stored in advance in the processing apparatus 102 are input information. It is assumed that

  The vehicle position detection flow in FIGS. 3 and 4 is a process assuming that “if there is no vehicle in the parking prohibited area 600, this indicates the color of the road (gray of asphalt)”.

In FIG. 3, first, the brightness of each pixel of the aerial image 501 is calculated, the compensation of the threshold α is calculated using the brightness of the aerial image 501, and the correction threshold α ′ used in the subsequent processing is calculated ( Step S11).
Step S11 is a compensation process for absorbing the difference because the image state (brightness) varies depending on the shooting situation.

  Specifically, a value corresponding to “lightness = 255” is initially set as the threshold value α, and if the maximum lightness value Lm of the pixels of the aerial image 501 is “0 <Lm ≦ 255”, The correction threshold value α ′ is calculated as in equation (1).

  α ′ = α × (Lm / 255) (1)

  Note that when the maximum brightness value Lm is “0”, the entire aerial image 501 indicates a completely dark state, so that the vehicle cannot be identified, and is excluded from the vehicle position detection processing execution conditions. .

Subsequently, the region X to be investigated is selected and determined from the aerial image 501 (step S12).
In step S12, it is assumed that portions where the vehicle position detection process (described later) is not executed are sequentially selected from the parking prohibited area 600.

  The size of the region X to be selected is calculated by calculating the relationship between the number of pixels in the aerial image 501 and the actual size on the ground using the camera parameter C. Is set to a size large enough to contain the

Next, in order to determine the presence or absence of a vehicle in the region X, the color determination process in the region X (step S13) using the region X determined in step S12 and the RGB information of each pixel of the aerial image 501. ).
As described above, step S13 is processing based on the assumption that “if there is no vehicle in the parking prohibited area 600, the color of the road is indicated”. Specifically, it is determined how much gray (asphalt color) is included in the color (RGB information) of the pixels in the region X.

Hereinafter, the color determination process (step S13) in FIG. 3 will be described in detail with reference to FIG. In this case, the correction threshold value α ′ is captured as input information at the start of the flow.
In FIG. 4, the processing apparatus 102 first initializes “0 clear” of the color determination counter value pc (step S <b> 21). The color determination counter value pc finally becomes the color determination result (return value) shown in FIG. 4 (step S13).

  Subsequently, one pixel (one point) is selected from the unexamined points in the region X (step S22), and the RGB value (color information) of the selected point is acquired (step S23), and R, G, Among the values of B, the maximum value pmax and the minimum value pmin are acquired (step S24).

  Next, the value of pmax−pmin is calculated and compared with the correction threshold value α ′ (for example, α ′ = 20), and the selection in the region X is determined depending on whether the relationship of the following expression (2) is satisfied. It is determined whether or not the point is gray (step S25).

  pmax−pmin <α ′ (2)

If it is determined in step S25 that pmax−pmin <α ′ (that is, YES), the selected pixel is regarded as being close to gray (the difference in RGB values is small), and the color determination counter value pc is set to “1”. (Step S26), and the process proceeds to step S27.
On the other hand, if it is determined in step S25 that pmax−pmin ≧ α ′ (that is, NO), the process proceeds to step S27 without executing step S26.

Next, it is determined whether or not the investigation process (steps S22 to S26) has been executed for all the pixels (points) in the region X (step S27).
If it is determined in step S27 that all the pixels in the region X have been checked (that is, YES), the final value of the color determination counter value pc is output as a processing result (return value), and the processing of FIG. The flow is terminated, and the process proceeds to step S14 in FIG.

  On the other hand, if it is determined in step S27 that an unexamined pixel exists in the region X (that is, NO), the process returns to the pixel selection process (step S22) and the investigation process (steps S22 to S26) is repeatedly executed. .

Returning to FIG. 3, the vehicle detection means and illegal parking determination means in the processing apparatus 102 have a ratio (pc / N) between the color determination counter value pc, which is the determination result in step S13, and the number N of pixels in the region X. It is determined whether or not the threshold value β (for example, β = 0.7) or more (step S14).
If it is determined in step S14 that pc / N ≧ β (that is, YES), the current investigation target area X is determined to be a road (step S15), and pc / N <β (that is, NO) is determined. If so, the region X to be investigated this time is determined as a parked vehicle (step S16).

Next, it is determined whether or not the vehicle position detection process and the illegal parking determination process (steps S12 to S16) have been executed for all areas in the parking prohibited area 600 (step S17).
If it is determined in step S17 that the vehicle position detection process has been completed for all the areas in the parking prohibited area 600 (that is, YES), the area X in which illegally parked vehicles exist is output, and FIG. The processing flow ends.

  On the other hand, if it is determined in step S17 that an unprocessed area exists in the parking prohibited area 600 (that is, NO), the process returns to the area selection process (step S12), and the vehicle position detection process (steps S12 to S16). Repeatedly.

From the vehicle position detection information (region X) obtained by the processing of FIG. 3, as shown in FIG. 5, it can be determined where the illegally parked vehicle exists.
Thereby, after confirming the absolute position of the vehicle from FIG. 5, the police can immediately mobilize the vehicle to the parking position, check the actual vehicle, and detect all illegally parked vehicles.

  At this time, there is no guarantee that the vehicle detected by the above process (FIG. 3) and the vehicle actually cleared by the policeman on the spot are the same, but the cleared vehicle is illegal parking. So there will be no trouble.

  Although not mentioned here, for the purpose of improving the accuracy of the vehicle position detection result, redundancy is added to the vehicle position detection process, and it is obtained from a plurality of aerial images 501 taken in time series. 3 may be executed for each of the “same parking prohibited areas 600 existing in different aerial images 501”.

  That is, the helicopter 101 captures a plurality of aerial images 501 in time series, and the vehicle detection means and the illegal parking determination means in the processing device 102 are placed in the same parking prohibited area 600 in the plurality of aerial images 501. On the other hand, if the same area X in the same parking prohibited area 600 is determined to be “parked vehicle” by conducting the investigations of FIG. 3 and FIG. Is determined.

In this case, regarding the same parking prohibited area 600 in the plurality of aerial images 501, the final vehicle position detection result is output only when the vehicle position is detected in all the processes, and only a part is detected. Is added as a redundant method of considering it as a false detection and excluding it (the final vehicle position detection result is not output).
This is because it is assumed that the same vehicle is always detected in the same region X if the parked vehicle is continuously parked in a short time and the vehicle is parked.

  As described above, the vehicle position detection system according to the first embodiment (FIGS. 1 to 5) of the present invention acquires the aerial image 501 obtained by photographing the ground from above, and illegally parks from the aerial image 501. Map data file for storing a helicopter 101 (photographing means) having a camera 100 for photographing an aerial image 501 and map data M corresponding to the aerial image 501 in order to detect the position of the target vehicle. 510, and a processing device 102 that detects a vehicle from the aerial image 501 and the map data M.

  The processing device 102 includes a map alignment unit that aligns the aerial image 501 and the map data M, a vehicle detection unit that detects a vehicle present in the aerial image 501, and the vehicle is in the parking prohibited area 600. Illegal parking determination means for determining whether or not.

  The map data M in the map data file 510 includes characteristic feature information (such as a crossover position), and the map alignment means in the processing device 102 uses the characteristic feature information in the aerial image 501 and the map data M. Alignment is performed by collating with the characteristic feature information inside.

  Further, the map data M includes parking prohibition area information (parking prohibition area 600), and the vehicle detection means in the processing device 102 detects only vehicles that exist in a portion where the aerial image 501 and the parking prohibition area 600 overlap. The illegal parking determination means determines that only the vehicle detected by the vehicle detection means is illegally parked.

  Further, the vehicle detection means and the illegal parking determination means sequentially set the area X corresponding to the size of the vehicle so that all the parking prohibited areas are examined in the parking prohibited area 600 determined from the parking prohibited area information. The difference between the maximum value pmax and the minimum value pmin of each value of RGB information is calculated for each pixel in the region X, and the number of pixels (pmax−pmin) is smaller than the correction threshold α ′ ( The color determination counter value pc) is counted, and when the number counted in the area X is smaller than a predetermined number (pc / N <β), the area is determined as a parked vehicle.

As described above, by comparing the aerial image 501 and the map data M, the vehicle existing in the parking prohibited area 600 can be detected easily and automatically.
In addition, since the helicopter 101 or the like captures the ground from the sky and detects the illegally parked vehicle using the aerial image 501 and the map data M, the processing can be performed without noticing the vehicle to be investigated.

  Further, since the map data M includes not only position information (such as latitude and longitude information of the range indicated by the map) but also information on the parking prohibited area 600 in advance, the processing device 102 uses the aerial image 501 and the map. By taking matching with the data M, the parking prohibited area 600 in the aerial image 501 can be easily determined and determined.

  In addition, since the parked vehicle is determined by performing image processing only on the parking prohibited area 600 in the aerial image 501, it is possible to immediately determine that the determined parked vehicle is an illegally parked vehicle. it can.

Also, as shown in FIG. 5, the absolute position where the vehicle actually exists can be known by superimposing the position where the vehicle is identified on the map.
As a result, if a policeman arrives at the place at the same time after discriminating the illegally parked vehicle and the absolute position, the illegally parked vehicle can be reliably detected without being escaped.

  Furthermore, since the helicopter 101 includes transmission means (communication means) and the processing apparatus 102 includes an image / position receiver 110 (communication means), an aerial image 501 is transmitted from the helicopter 101 to the processing apparatus 102 in real time. The vehicle position detection process can be realized quickly.

Embodiment 2. FIG.
In the first embodiment (FIG. 1), the aerial image 501 and the helicopter position / orientation information 502 are appropriately transmitted from the helicopter 101 to the processing device 102 via the communication unit. However, as shown in FIG. Alternatively, the recording medium 200 may be mounted on the helicopter 101, a plurality of aerial images 501 may be stored in the recording medium 200, and processed together in the processing apparatus 102A after the photographing is completed.

  FIG. 6 is a block diagram schematically showing the entire vehicle position detection system according to Embodiment 2 of the present invention. Components similar to those described above (see FIG. 1) are denoted by the same reference numerals as described above, Alternatively, “A” is appended after the reference numerals and detailed description is omitted.

  In this case, in the helicopter 101, the aerial image 501 and the helicopter position / orientation information 502 are once collectively recorded on the recording medium 200. Together with the map data M and the camera parameter C).

  According to the second embodiment (FIG. 6) of the present invention, communication means (such as the image / position receiver 110) is not required in the helicopter 101 and the processing apparatus 102A, but the overall system configuration is simplified. Since the real-time property is lowered, it is desirable to execute the above-described vehicle position detection process immediately after the end of shooting.

Embodiment 3 FIG.
In Embodiments 1 and 2 (FIGS. 1 and 6), the processing device 102 is installed at a location (ground side) different from the helicopter 101 (imaging means for image acquisition) on the sky side. The processing apparatus 102 may be mounted on the helicopter 101 together with the files 510 and 511, and only the final vehicle position detection result may be transmitted to the ground side.

  In this case, although the equipment mounted on the helicopter 101 side increases, there is no need to transmit the aerial image 501 consisting of a huge amount of data to the ground side, so transmission information from the helicopter 101 side to the ground side is reduced only to the measurement results. There is an effect that.

Embodiment 4 FIG.
In the first embodiment, an overpass or the like is used as “characteristic feature information easy to distinguish from the sky” for matching the aerial image 501 and the map data M, but other feature information is used. May be used. It is also possible to perform matching not using single feature information but using structures such as the arrangement of buildings and the width of roads.

Embodiment 5 FIG.
Furthermore, in the first embodiment (FIGS. 3 and 4), the color of each pixel in the region X is used to detect the vehicle from the aerial image 501. You may detect a parked vehicle by acquiring target information.
In this case, it is difficult to directly acquire the three-dimensional information from the aerial image 501, but it can be applied by using a determination unit based on learning data.

  100 camera, 101 helicopter (photographing means), 102, 102A processing device, 110 image / position receiver, 200 recording medium, 501 aerial image, 502 helicopter position / orientation information, 510 map data file, 511 camera parameter file, 600 Parking prohibited area, C camera parameter, M map data, pc color determination counter value, pmax maximum value, pmin minimum value, X area, α ′ color determination correction threshold, β parked vehicle determination threshold.

Claims (5)

A vehicle position detection system that acquires an aerial image obtained by photographing the ground from the sky side, and detects the position of a vehicle subject to illegal parking from the aerial image,
Photographing means having a camera for photographing the aerial image;
A map data file storing map data corresponding to the aerial image;
A processing device for detecting the vehicle from the aerial image and the map data;
The processor is
Map alignment means for aligning the aerial image and the map data;
Vehicle detection means for detecting a vehicle present in the aerial image;
It possesses the illegal parking determination means for determining whether the vehicle is present in the no-parking area,
The map data includes the parking prohibited area information,
The vehicle detection means detects only a vehicle existing in a portion where the aerial image and the parking prohibited area information overlap,
The illegal parking determination means determines that only the vehicle detected by the vehicle detection means is an illegal parking vehicle,
The vehicle detection means and the illegal parking determination means are:
In order to investigate all the parking prohibition areas within the parking prohibition area determined from the parking prohibition area information, sequentially set an area corresponding to the size of the vehicle,
For each pixel in the region, calculate the difference between the maximum value and the minimum value of each value of RGB information, and count the number of pixels where the difference is smaller than a predetermined threshold,
The vehicle position detection system , wherein when the number counted in the area is smaller than a predetermined number, the area is determined as a parked vehicle . The map data includes characteristic feature information,
The vehicle position according to claim 1, wherein the map alignment unit performs alignment by comparing characteristic feature information in the aerial image with characteristic feature information in the map data. Detection system. The imaging means captures a plurality of aerial images in time series,
The vehicle detection means and the illegal parking determination means are:
For each of the same parking prohibited areas in the plurality of aerial images, the investigation is performed,
2. The vehicle position according to claim 1 , wherein the vehicle position is finally determined to be an illegally parked vehicle only when it is determined that the same area in the same parking prohibited area is a parked vehicle. Detection system. Each of the photographing means and the processing device has a communication means,
The vehicle position detection system according to any one of claims 1 to 3 , wherein the aerial image is transmitted from the imaging unit to the processing device in real time. The photographing means has a storage medium for storing a plurality of aerial images,
The vehicle position detection system according to any one of claims 1 to 3 , wherein the processing device collectively processes a plurality of aerial images stored in the storage medium. .

Images