JP3384526B2 - Logistics measurement equipment - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a physical distribution measuring device used for a traffic flow measuring device or the like using stereo image measurement which is useful for traffic control.

[0002]

2. Description of the Related Art First, the principle of stereo image measurement
explain. The stereo image measurement technology uses two left and right imaging devices.
The same object is photographed by the table, and the object point is detected by the triangulation principle.
The distance to p is measured. In Figure 7, the optical axis is water
Imaging planes 70 of two imaging devices arranged flat and parallel
1, 702 and the object point p are shown. In reality,
The imaging planes 701 and 702 exist in front of the x-axis, but
For the sake of clarity, I will show it like this. Now up to object point p
Suppose that the distance zp is measured. The object point p is on the left and right images
Each point P_L, P_RImage at point P_L, P_ROn the image
X coordinate in x direction_L, X_RThen parallax
S is calculated as in Expression (1).         Parallax S = XL-XR (1) From the left and right images, point P_L(X_L, Y_L), P
_R(X_R, Y_R) {However, Y_L= Y _R} Process to find
Is referred to as association processing. Correlation is done from the left and right images
This is done by finding the pair of areas with the highest seki.
In addition, from the parallax S to the object p from the image pickup device according to Expression (2).
At distance z_pCan be calculated.           Distance z to object point p_p= (B × f) / s (2) However, B: arrangement interval of the imaging device f: lens focal length

Next, as a technique for estimating the three-dimensional position of the observation surface by using stereo image measurement, Japanese Patent Application No. 7-23.
The "plane estimation method" described in Japanese Patent No. 4019 will be described. In the following description, estimating the three-dimensional position of the road surface will be described as an example by replacing the object with the vehicle and the observation surface with the road surface.

A plurality of image pickup devices installed at a high place on the road capture a stereo image of the road surface where no vehicle exists. Of the stereo images, one reference image (left image) is shown in FIG.
A total of M × N blocks BK (X, Y) in the vertical Y direction
Then, the parallax is measured for each block. The parallax data measured in each block is described as S (X, Y). This is plotted on the XYS coordinate shown in FIG. Since the parallax S (X, Y) when the index Y (1 ≦ Y ≦ N) is fixed should be on the same plane and on one straight line, the straight line is approximated using the Hough transform.
Since this straight line passes through the road surface, it will be referred to as a plane passing straight line 802. As the straight line parameters of the plane passing straight line obtained by the Hough transformation, the length ρ of the perpendicular line drawn from the straight line to the Y axis and the inclination θ of the perpendicular line are used, and each plane passing straight line is expressed by the equation (3). be able to. ρ _Y = _X sin θ _Y + _Y sin θ _Y (3) where ρ _Y : Distance from Y axis to plane passing straight line θ _Y : Angle formed by perpendicular line drawn from plane passing straight line and XY plane (subscript Y is each Indicates index Y)

The plane passing straight line is approximated for each index Y as described above. As a result, the road surface 803 can be approximated by a plane passing straight line group, and in a block where parallax cannot be acquired due to insufficient features or the like, the parallax RS (X, Y) on the road surface is obtained from the plane passing straight line group by estimation. be able to.

Next, a "vehicle detection device" described in Japanese Patent Application No. 8-113897 will be described with reference to FIG. 9 as a physical distribution measurement device using conventional stereo image measurement.
A stereo image of the road is captured by the two imaging devices CU1 and CU2 arranged on the left and right so that the optical axes are parallel to each other, and the left image of the stereo images is used as the reference image.

In the associating unit ST1, the left image is divided into a total of M × N blocks BK (X,
Y), find a similar region in the right image for each block, and obtain the parallax S (X, Y). The acquired parallax S (X, Y) {1 ≦ X ≦ M, 1 ≦ Y ≦ N} will be referred to as a parallax image. However, since a region having no feature as an image on the road surface or the like cannot be associated, parallax cannot be acquired in all blocks.

In the plane estimation unit ST2, the correspondence unit ST1
The position of the road surface is three-dimensionally estimated using the Hough transform as described in the related art using the parallax of each block output from the parallax, and the parallax of the entire road plane in the absence of a vehicle is calculated. To win.

In the coordinate conversion unit ST3, as shown in FIG. 10, the parallax S in the coordinate system co1 with the image pickup surface as a reference.
(X, Y) is converted into three-dimensional data H (U, V) in the coordinate system co2 with the road surface as a reference. The coordinate system co2 based on the road surface is composed of two dimensions U and V indicating the position on the road surface and one dimension H indicating the height of the object on the road. The coordinate conversion is performed in three phases, and each phase will be described below. 1. Phase 1 (conversion in U direction) First, U (X, Y) is calculated. When the imaging device is installed at a high place on the road, the imaged image is as shown in FIG. This is converted in the horizontal direction so that the data array intervals are at regular intervals on the actual road surface. Specifically, the image is divided into small lanes 1101 and the data is rearranged so that the small lanes 1101 are parallel to each other, so that U (X, Y) is first determined. If the above process is schematically explained, the data arrangement of FIG.
(B) This is a process for allocating data. 2. Phase 2 (conversion in H direction) In Phase 2, using the parallax S (X, Y) acquired,
The height H (X, Y) of the photographed vehicle from the road surface is calculated. The height H (X, Y) of the vehicle from the road surface is shown in FIG.
Image pickup devices CU1 and CU2 and a road surface 1201 as shown in FIG.
(4) in the geometrical relationship with the vehicle TR to be detected
Can be calculated using

[0010]

[Equation 1] (4) where z _K (X, Y): distance from the image pickup device to the vehicle h: height of the image pickup device from the road surface z _RK (X, Y): distance from the image pickup device to the road surface As described above, the parallax S (X, Y) can be easily converted into the distance z _K (X, Y) by the equation (2), and the distance z _RK (X, Y) from the image pickup device to the road surface can be obtained. In the same manner, the same can be calculated from the parallax RS (X, Y) of the road surface. In addition, the calculation of the height H (X, Y) is performed according to FIG.
In, the value in the normal direction of the paper is calculated. 3. Phase 3 (V-direction conversion) In phase 3, V (X, Y) is calculated. Specifically V
(X, Y) are image pickup devices CU1 and C as shown in FIG.
The geometrical relationship between U2, the road surface 1201 and the detected vehicle TR is calculated using equation (5).

[0011]

[Equation 2] (5) where z _K (X, Y): distance from the imaging device to the vehicle h: height of the imaging device from the road surface z _RK (X, Y): distance from the imaging device to the road surface θ: Convert the parallax image S (X, Y) into H (U (X, Y), V (X, Y)) for the three-dimensional data as described above, which is the angle formed by the perpendicular and the optical axis of the imaging device. Can be obtained at.

Returning to FIG. 9, the lane detector ST5 projects the three-dimensional data U, V, H obtained by the coordinate conversion in the traveling direction of the vehicle or the opposite direction. For example, in FIG. 8, the three-dimensional data of the vehicle and the road is projected on the UH plane. In order to detect the lane position of the road from the projected image, for example, a position where the height H of the vehicle is equal to or larger than a threshold value may be found on the U axis. Note that the lane detection unit outputs only the lane position that is stable over a certain period of time, thereby detecting the lane position without being affected by the vehicle or the like that changes the lane.

In the projection vehicle detection unit ST6, first, the three-dimensional data is divided into lanes based on the lane position obtained by the lane detection unit ST5, and the three-dimensional data of each lane is taken as HV in FIG.
Project on a plane. For example, the space RL on the right lane in FIG.
If the three-dimensional data of is projected on the HV plane, it becomes projection data PR. However, the projected image is affected by the error,
It is unlikely that the vehicle will look like it is seen from the side. Next, the vehicle is detected by scanning the window WN along the V axis with respect to the projection data, and when the number of projection data whose height H is equal to or larger than the threshold value reaches the number equal to or larger than the threshold value within the window, the vehicle exists. The position where the projection data rises (the head position or the rear end position) is detected as the position of the vehicle.

In the vehicle type discriminating section ST7, the length of the vehicle in the V-axis direction and the height H of the vehicle are measured from the projected image as shown in FIG. 2 and compared with the data of each vehicle type which is known in advance. Distinguish large vehicles, small vehicles, motorcycles, etc.

Next, detection of a stationary object in a conventional physical distribution measuring device will be described by taking an example of "illegal parking monitoring device" (Technical Report No. 512 of the Institute of Electrical Engineers of Japan, Intelligent Traffic Measurement Section 96). As shown in FIG. 14, the luminance data of the measurement sample points 1402 in the measurement area 1401 set in advance on the roadside band is ternarized and extracted as the characteristic points of the vehicle. The parked vehicle 1403 is detected by measuring the time when the vehicle is stopped in the measurement area. The flow of the detection algorithm is shown in FIG. 15, and each phase will be described below in order.

Step 1501 (setting of measurement area) The area for measuring the vehicle is set, and the points for extracting the brightness data (the above measurement sample points 1402) are determined so as to be evenly spaced on the road surface. Step 1502 (Ternary processing) Based on the brightness (luminance) of the road surface 1404, the measurement area is classified into an area brighter than the road surface, an area darker than the road surface, and a road surface. Give a value and make it three-valued. Step 1503 (Detection of Vehicle) Continuously ternized screens are overlapped at a fixed time interval to classify them into a moving object and a stationary object. A stopped vehicle is identified based on the size of the stopped object. Step 1504 (determination of parking) It is determined whether or not the existence range of the vehicle detected in each constant cycle changes with time, and the vehicle that has been stationary for a certain time or more is set as "parking".

Next, as an occupancy rate measuring method in the conventional physical distribution measuring apparatus, a road occupancy rate measuring method will be described with reference to FIG. An ultrasonic vehicle detector 1601 is installed above the road, and this is the vehicle 1 below the detector.
The presence of 602 is sensed. The time occupancy is obtained by calculating the percentage of time occupied by the vehicle 1602 passing under the sensor per unit time. For example, if a vehicle is detected for 70 seconds in 100 seconds, the time occupancy rate becomes 70%. In the conventional method using the ultrasonic type vehicle sensor, the space occupancy (the ratio of the space occupied by the vehicles existing in a certain section of the road at a certain moment) cannot be measured.

[0018]

However, in the above-mentioned conventional physical distribution measuring device, although it is relatively easy to detect a moving object, the object is stopped and the objects are front, rear, left and right on the image. When it seems to overlap
It was difficult to detect each object separately. Further, although the stationary object is detected using the brightness information of the image, it is conceivable that erroneous detection or detection omission may occur due to the influence of the shadow generated on the observation surface and the reflection of light. Further, there is a space occupancy rate as information for judging the degree of congestion of roads, and there is a demand for a method of measuring the space occupancy rate, which is more advanced information, in place of the time occupancy rate used conventionally.

Therefore, in the physical distribution measuring device of the present invention,
A first object of the present invention is to enable objects to be accurately separated and detected even when they are stopped and appear to overlap each other. A second object of the present invention is to detect a stopped object such as a parked vehicle, a broken vehicle, or an accident vehicle with higher accuracy using three-dimensional information. Further, the third object of the present invention is to measure the space occupancy rate, which cannot be measured by the conventional ultrasonic type vehicle sensor.

[0020]

In order to achieve the first object, according to the present invention, a vehicle detection method suitable for stopping when it can be determined that the vehicles are stopped in a state where they appear to overlap each other in an image. It is intended to switch to. As a result, even object vehicles that appear to overlap in the image can be detected separately.

In order to achieve the second object, according to the present invention, the three-dimensional data, which is the three-dimensional information of the object, is integrated to stably stop without being affected by disturbance such as shadow or light reflection. It is intended to detect an object. Thereby, the stationary object can be detected with high accuracy.

In order to achieve the third object, according to the present invention, the road environment including a vehicle is measured by stereo image measurement.
The measurement is performed dimensionally, and coordinate conversion and projection processing are performed. Thereby, the space occupancy rate can be measured.

[0023]

BEST MODE FOR CARRYING OUT THE INVENTION The invention according to claim 1 of the present invention associates a plurality of image pickup devices for obtaining a plurality of images by photographing an observation surface with a region having a high correlation between the plurality of images. a mapping unit for acquiring a parallax, and a plane estimating section for estimating the position of the observation surface three-dimensionally from the parallax, a coordinate transformation unit for acquiring 3-dimensional data from the position of the parallax between the observation surface, the The difference between three-dimensional data frames and the space occupancy
A stop determination section which calculates to an object, it is determined whether the stop by a traffic jam, a distribution measuring apparatus equipped with an adaptive object detecting unit that detects a position of an object from the determined to the three-dimensional data, a plurality Using the three-dimensional data obtained by the single imaging device, the associating unit, the plane estimation unit, and the coordinate conversion unit, the stop determination unit calculates the inter-frame difference of the three-dimensional data.
The space occupancy rate is calculated to determine whether or not the objects are stopped in a state where the objects appear to overlap each other on the image. When the stop determination unit determines that the traffic is congested , the adaptive object detection unit switches to an object detection method suitable for the stop , and the three-dimensional data is detected.
In Rukoto detecting the position of an object by integrating the data has the effect of improving the accuracy of separating and detecting each stop objects appear overlapped as a result individually.

[0024]

The invention according to claim 2 of the present invention is characterized in that the observation surface is a road surface and the object is a vehicle.
The physical distribution measuring device according to Item 1, which has an action of accurately measuring a stopped vehicle on a road.

[0026]

[0027]

According to a third aspect of the present invention , the parallax is obtained by associating a plurality of image pickup devices for taking a plurality of images by photographing the observation surface with regions having a high correlation between the plurality of images. An associating unit, a plane estimation unit that three-dimensionally estimates the position of the observation plane from the parallax, a coordinate conversion unit that obtains three-dimensional data from the parallax and the position of the observation plane, and a specification from the three-dimensional data A three-dimensional data obtained by a physical distribution measuring device including an occupancy measuring unit that measures a space occupying ratio in a space, and a plurality of imaging devices, an associating unit, a plane estimating unit, and a coordinate converting unit. The occupancy measuring unit creates the projection data of the observation surface by using, and measures the space occupancy by calculating "space occupancy = number of projection data in measurement area / maximum number of projection data in measurement area". By measuring the space occupancy rate in a specific space It has the effect that.

The invention according to claim 4 of the present invention is characterized in that the observation surface is a road surface, the object is a vehicle, and the space occupancy in a specific space is the space occupancy in each vehicle.
The physical distribution measuring device described in 3, which has an action of accurately measuring the space occupancy of the vehicle on the road.

Embodiments of the present invention will be described below with reference to the drawings. (Embodiment 1) Embodiment 1 corresponding to claims 1, 2 and 3 of the present invention will be described with reference to FIGS. Regarding the operations of the imaging devices CU1 and CU2, the associating unit ST1, the plane estimating unit ST2, and the coordinate converting unit ST3,
This is as described in the section of the related art. In the following embodiments, separation detection in the case where vehicles appear to overlap in a road environment is taken as an example, and the observation surface in the claims is replaced with the road surface and the object is replaced with the vehicle.

In the stop determination unit 101, first, the coordinate conversion unit S
Using the three-dimensional data obtained by T3, the projection data is created for each lane by the method described in the section of the related art. Next, the difference amount and the space occupancy rate of the three-dimensional data between the frames are calculated for each lane using the three-dimensional data and the projection data. The difference amount of the three-dimensional data between the frames is calculated from the three-dimensional data at the current time TN.
It is the one obtained by subtracting the three-dimensional data at the past time TP, taking the absolute value, and summing up in the lane. Specifically, the calculation is performed using the following formula (6).

[0032]

[Equation 3] (6) However, HTN (U, V): Current three-dimensional data HTP (U, V): Past three-dimensional data set Φn: All (U, V) in lane n

Generally, the difference amount is large when the vehicle in the lane is moving, and the difference amount is small when the vehicle does not exist or is stopped. Current time TN and past time TP
The time interval between and is not particularly limited and is 1 frame interval or more.

The space occupancy is a certain section on the road,
It is the sum of the total lengths of vehicles in a section divided by the total distance of the section. The space occupancy is generally obtained by the following equation (7). For example, when two vehicles having a total length of 5 m are present in the measurement area 40 m in the image, the space occupancy is 25%. Space occupancy = sum of the total length of vehicles in the section / total distance in the section (7)

Since it is possible to calculate the above equation (7) by approximating it to the following equation (8), the following equation (8) is used in the present invention. Space occupancy rate = number of projection data in measurement area / maximum number of projection data in measurement area (8) However, the number of projection data is the total number of projection data PR having a height H equal to or higher than a threshold in FIG. The maximum number of projection data is the maximum number of projection data that can be taken in the measurement range. Normally, when vehicles are concentrated in a section such as a traffic jam, the space occupancy rate becomes high.

Using the three-dimensional data difference between the frames calculated as described above and the space occupancy rate, when the space occupancy rate is high and the three-dimensional data difference between the frames is sufficiently small, the vehicle runs in the lane. It is determined that the vehicle is stopped due to traffic congestion and the vehicle interval is short.

In the adaptive object detection unit 102, when the stop determination unit 101 determines that the vehicle is stopped due to congestion, the vehicle detection system is switched to a system suitable for the stop.
In other cases, that is, when the vehicle is traveling smoothly, the vehicle is detected separately. The vehicle detection method in each case will be described below.

First, a method for detecting a vehicle when the vehicle is running smoothly will be described with reference to FIG. Scan the window WN for the projection data PR for each lane,
When the projection data PR having a height equal to or higher than the threshold value α is concentrated on a part, it is determined that the part is a vehicle, and the end of the part is detected as the head position TP or the rear end position of the vehicle. The specific method of detection is shown below. Window WN
If there are n cells inside, the projection data value in each cell is expressed as CL (i) {0 ≦ i <n}. When the condition of the following expression (9) is satisfied, the position of CL (2) is detected as the head position TP or the rear end position of the vehicle. {CL (0) <α} and {CL (1) <α} and {CL (2)> α} and {m ≦ 3 CL (i) with α or more in 3 ≦ i <n}.・ ・ (9)

Next, will be described with reference to FIG detection method when the vehicle is stopped, but the same as the case is basically flowing smoothly, integrating the processed projection data
Different only be used. When the distance between the front and rear vehicles is small as shown in FIG. 3, the threshold α is set between the projection data of the front and rear vehicles.
When the false projection data is generated due to the random error exceeding, the projection data of the preceding and following vehicles are connected and the leading position of the following vehicle cannot be specified, resulting in detection failure. Therefore, the integration process is performed on the projection data. The integration process is a process of adding projection data of the past several frames when the vehicle is stopped and dividing by the added number of frames. That is, the projection data of the past several frames are smoothed and used in the time direction. The window WN is scanned with respect to the projection data subjected to the integration processing, and the head position TP or the rear end position of the vehicle is detected by the above-mentioned method. By performing integration processing on the projection data, the influence of random errors can be reduced, and as a result, stable vehicle detection during traffic congestion can be performed.

Hereinafter, the characteristics of the physical distribution measuring device according to the first embodiment of the present invention and the conventional traffic flow measuring device will be compared using (Table 1).

[0041]

[Table 1] As is clear from Table 1, the first embodiment of the present invention
The physical distribution measuring device according to the above can obtain a good effect in detecting a stopped vehicle at the time of traffic jam. In the above description, the detection of the stopped vehicle on the road has been described, but the road surface may be replaced with the observation surface and the vehicle may be replaced with an object to detect a stopped object.

(Embodiment 2) Next, claim 4 of the present invention
Embodiment 2 corresponding to 5 and 5 will be described. FIG. 4 shows a physical distribution measuring device for detecting a stationary object according to the second embodiment, and the processes up to the image pickup devices CU1 and CU2, the associating unit ST1, the plane estimating unit ST2, and the coordinate converting unit ST3 are the same as those in the related art. Same as the technical section. In the following, the detection of a stopped vehicle such as a parked vehicle in a road environment will be described as an example by replacing the observation surface in the claims with the road surface and the object with the vehicle.

The stopped object detection unit 402 detects stopped vehicles on the road such as parked vehicles, broken vehicles, and accident vehicles. The stopped vehicle detection unit first integrates the three-dimensional data H (U, V). The integration process is 3 on the road surface that is the measurement range.
This is a process of integrating the dimensional data for a certain period of time and dividing by the number of integrated frames. The three-dimensional data that has been subjected to the integration process will be called three-dimensional integrated data. Below, concrete 3
A method of calculating the dimensional integral data will be described. The operation of integrating the sequentially obtained three-dimensional data from time Ts to Te is performed by the following formula (10) for each position (U, V) on the road.
Is performed using.

[0044]

[Equation 4] However, HS (U, V): three-dimensional data subjected to integration processing Ht (U.V): three-dimensional data at time t Fn: number of frames from time Ts to Te ... (10)

Next, a method of detecting a stopped vehicle using three-dimensional integral data will be described. As schematically shown in FIG. 5, in the three-dimensional integration data, if the integration process is performed from time T1 to time Tn, the area of the vehicle parked during that time is higher in the height direction than the area that is not. growing. Detecting a parked vehicle by determining an area of an object having a height in the three-dimensional integration data HS (U, V) that is equal to or greater than a threshold value and having an appropriate shape and size as a vehicle to be an area of the parked vehicle. You can

In the above description, it is assumed that the three-dimensional data is integrated, but the projection data of each lane is calculated and the projection data is integrated, or H (X, Y) in the following equation is equal to or larger than the threshold β. Alternatively, a method of integrating the function R (X, Y) {R (X, Y) = 0 or 1} such that 1 is set to 0 and 0 is set to γ or less may be used. if (H (U, V)> γ) R (U, V) = 1 else R (U, V) = 0 (11) In the above description, detection of a stopped vehicle on the road has been described. The road surface can be replaced with an observation surface, the vehicle can be replaced with an object, and a stationary object can be detected.

(Third Embodiment) Next, a sixth aspect of the present invention will be described.
A third embodiment corresponding to Nos. 7 and 7 will be described. FIG. 6 shows a physical distribution measuring apparatus for measuring the space occupancy rate in the third embodiment, and the correspondence unit ST1, the plane estimating unit ST2, and the coordinate converting unit ST3 are the same as in the first and second embodiments. Is. The occupancy rate measuring unit 601 measures the space occupancy rate. First, the 3 obtained in the coordinate conversion unit ST3
The projection data of each lane is created from the dimension data. The method of creating projection data from three-dimensional data is the same as that described above. Equation (8) is used to calculate the space occupancy from the projection data.

From the above, the space occupancy rate can be grasped for each lane. In the above embodiments, the space occupancy ratio measuring method for each lane in the road environment has been described, but the lane may be replaced with the specific space.

[0049]

As is apparent from the first embodiment, the present invention detects that an object has stopped with a stop determination unit, and switches the object detection method to a method suitable for stopping the vehicles. It is possible to accurately detect the vehicle even when the vehicle is stopped due to a large number of overlapping points.

As is apparent from the second embodiment, the present invention integrates three-dimensional data such as shadows and headlights that are not affected by changes in brightness, so that it can be used in an outdoor environment with many disturbances. Even if there is, a stopped object such as a parked vehicle, a broken vehicle, or an accident vehicle can be accurately detected.

As is clear from the third embodiment, the present invention measures the three-dimensional shape of the road including the vehicle and is useful as data for grasping the traffic situation in traffic control. Can be measured by image processing.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a traffic flow measuring device according to a first embodiment of the present invention.

FIG. 2 is a schematic diagram illustrating vehicle detection according to the first embodiment of the present invention.

FIG. 3 is a schematic diagram illustrating vehicle detection according to the first embodiment of the present invention.

FIG. 4 is a block diagram showing a configuration of a traffic flow measuring device having a stationary object detection unit according to a second embodiment of the present invention.

FIG. 5 is a schematic diagram illustrating a principle of detecting a stopped vehicle according to the second embodiment of the present invention.

FIG. 6 is a block diagram showing a configuration of a traffic flow measuring device having an occupancy rate measuring unit according to a third embodiment of the present invention.

FIG. 7 is a schematic diagram showing the principle of stereo image measurement in a conventional technique.

FIG. 8 is a schematic diagram of a plane estimation method in a conventional technique.

FIG. 9 is a block diagram showing a configuration of a traffic flow measuring device according to a conventional technique.

FIG. 10 is a schematic diagram illustrating coordinate conversion in a conventional technique.

FIG. 11 is a schematic diagram illustrating coordinate conversion in a conventional technique.

FIG. 12 is a schematic diagram showing a geometrical relationship among an image pickup device, a road surface, and a vehicle in a conventional technique.

FIG. 13 is a schematic diagram showing the relationship between three-dimensional data and projection data in the conventional technique.

FIG. 14 is a schematic diagram illustrating detection of a parked vehicle in the related art.

FIG. 15 is a flowchart showing a flow of processing for detecting a parked vehicle in the related art.

FIG. 16 is a schematic diagram illustrating occupancy rate measurement in the conventional technique.

[Explanation of symbols]

CU1, CU2 imaging device ST1 associating unit ST2 Plane estimation unit ST3 Coordinate conversion unit 101 Stop judgment unit 102 Adaptive vehicle detector 401 Stopped vehicle detection unit HS integrated 3D data 601 Occupancy rate measurement unit 701 Imaging surface (left image) 702 Imaging surface (right image) B Baseline length (camera interval) f Focal length P object 801 Image side 802 plane passing straight line 803 road plane BK block ST5 Lane detection unit ST6 Projection vehicle detector ST7 Vehicle type determination section co1 Coordinate system based on the imaging surface co2 Coordinate system based on the road surface 1101 small lane Distance from RK imager to road surface RS Road surface parallax K Distance from imaging device to vehicle TR vehicle 1201 road surface RL Space on the right lane PR projection data WN window 1401 measurement area 1402 luminance sample points 1403 Parking vehicle 1404 Road surface 1601 Ultrasonic vehicle detector 1602 vehicle TP vehicle start position

─────────────────────────────────────────────────── ─── Continued Front Page (58) Fields surveyed (Int.Cl. ⁷ , DB name) G06T 1/00 330 G01B 11/00 G01C 3/06 G08G 1/04 G09B 29/10

Claims (4)

(57) [Claims] 1. A plurality of imaging devices that capture a plurality of images by imaging an observation surface, an associating unit that obtains parallax by associating regions having a high correlation between the plurality of images, and the observation from the parallax. A plane estimation unit that three-dimensionally estimates the position of a plane, a coordinate conversion unit that obtains three-dimensional data from the parallax and the position of the observation plane, an inter-frame difference of the three-dimensional data, and a space occupancy rate are calculated. A stop determination unit that determines whether or not the object has stopped due to congestion, and an adaptive object detection unit that detects the position of the object from the three-dimensional data and the determination, and the adaptive object detection unit is the A physical distribution measuring device characterized by integrating the three-dimensional data and detecting the position of an object when the stop determination unit determines that the traffic is congested. 2. The physical distribution measuring device according to claim 1, wherein the observation surface is a road surface and the object is a vehicle. 3. A plurality of image pickup devices for capturing a plurality of images by photographing an observation plane, an associating unit for obtaining a parallax by associating regions having a high correlation between the plurality of images, and an observation plane from the parallax. Plane estimation unit that three-dimensionally estimates the position, a coordinate conversion unit that obtains three-dimensional data from the parallax and the position of the observation plane, and an occupancy that measures a space occupancy rate in a specific space from the three-dimensional data. A ratio measuring unit, wherein the occupation ratio measuring unit uses the three-dimensional data obtained by the coordinate conversion unit,
A physical distribution measuring device characterized in that projection data of an observation surface is created, and space occupancy in a specific space is measured by calculating space occupancy = number of projection data in measurement area / maximum number of projection data in measurement area. . 4. An observation surface is a road surface and an object is a vehicle,
The physical distribution measuring device according to claim 3 , wherein the space occupancy rate in the specific space is set as a space occupancy rate in each vehicle.