JP2005056186A - Traffic condition observation system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To grasp traffic conditions in a predetermined area quantitatively and planary in a simple manner. <P>SOLUTION: A traffic flow measurement program 42 executes a first storage step that stores a first image imaged by remotely sensing a scope including the predetermined area at a first time, and a second image imaged by remotely sensing the scope including the predetermined area at a second time after a predetermined time from the first time in a memory 4, a second storage step that stores geographical information including at least traffic control information for a road in the prescribed area in the memory 4, an extraction step that extracts a scope of a vehicle based on the first and second images stored in a memory 4 respectively, a search step that searches the second image for each vehicle in the first image based on the result of extraction in the extraction step, and a presentation step that presents a prescribed observation result regarding the traffic condition based on the search result in the search step. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a traffic condition observation system for supporting quantitative measurement of traffic conditions.

  It is a request from each direction to accurately grasp the actual situation of traffic conditions. For example, traffic congestion is a serious social problem. In order to eliminate the traffic jam, it is necessary to take measures such as identifying the cause of the traffic jam and changing the traffic control or street structure based on the cause. For this purpose, it is desired to perform a detailed behavior analysis of the vehicle. For example, in order to estimate the demand for parking lots, it is required to accurately know the situation such as street parking. In this way, it is important that the actual state of traffic conditions be grasped “in a plane” for the wide area road network.

  By the way, the current traffic situation survey is generally a traffic survey by fixed point observation, and a vehicle detector installed on the road is used to support this. Many apparatuses have been proposed in which a monitoring camera is installed on a road and traffic flow is measured based on an image taken by the monitoring camera (see, for example, Patent Document 1).

Japanese Patent Laid-Open No. 7-21488

  However, the conventional methods as described above are merely “spot” observations of traffic volume. In order to grasp the plane traffic situation by such a conventional method, it is necessary to perform fixed-point observation at many points at the same time, and it is very expensive to implement this. Even if such simultaneous observation is performed, it is actually difficult to accurately grasp traffic characteristics (for example, traffic density) of individual roads in a wide area.

  According to one aspect of the present invention, there is provided a traffic condition observation system for quantitatively observing a traffic condition in a predetermined area, wherein a region including the predetermined area is captured by remote sensing at a first time. First storage means for storing one image and a second image obtained by remote sensing of an area including the predetermined area at a second time after a predetermined time from the first time; Based on second storage means for storing geographical information including traffic control information of roads within a predetermined area, and each of the first and second images stored in the first storage means, a vehicle area is determined. An extraction means for extracting, a search means for searching each vehicle in the first image from the second image based on an extraction result by the extraction means, and a traffic based on a search result by the search means Traffic observation system characterized by having a presentation means for presenting predetermined observations relating to status is provided.

  According to the present invention, the traffic situation in a predetermined area can be easily grasped quantitatively and in a plane.

  DESCRIPTION OF EMBODIMENTS Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a diagram illustrating a configuration of a traffic condition observation system according to an embodiment. This system can be realized by a general personal computer or workstation. As shown in the figure, a CPU 1 that controls the entire system, a ROM 2 that stores a boot program, etc., and a RAM 3 that functions as a main storage device. In addition, the following configuration is provided. 4 is a hard disk device (HDD) for storing various programs and data, and 5 is a memory (VRAM) for developing image data to be displayed, and 6 is a display device (CRT) by developing image data and the like. Can be displayed. Reference numerals 7 and 8 denote a keyboard and a mouse for performing various settings, respectively.

  FIG. 2 is a schematic diagram showing programs and data stored in the HDD 4. As shown in the figure, the HDD 4 is installed with an OS 41, a traffic flow measurement program 42 for measuring a traffic situation (traffic flow) in a survey target area, and a GIS software 43 for realizing a geographic information system (GIS). GIS intends to provide various geographic information by combining other attribute information based on digitized geographic information (map data). Specifically, as shown in the figure, the GIS includes map data 432 and attribute data 433 covering the survey target area as geographic information in addition to the GIS control program 431. Here, the attribute data 433 includes at least traffic control information of each road (for example, information such as speed limit and one-way traffic). In addition to this, the HDD 4 has an area for storing satellite image data obtained by remote sensing from an artificial satellite, for example.

  Next, the outline of the processing of the traffic condition observation system in the embodiment will be described with reference to the flowchart of FIG. The process shown in the flowchart of FIG. 3 is realized by the linkage of the traffic flow measurement program 42 and the GIS 43. Both programs are loaded into the RAM 3 and executed by the CPU 1.

  First, in step S1, satellite image data captured by, for example, an artificial satellite is input. An example of a suitable artificial satellite is the QuickBird satellite. Earth observation satellites such as the QuickBird satellite remotely sense the ground surface with a line sensor while moving at about 8km / s in the satellite orbit. A line sensor mounted on a satellite projects an electromagnetic wave received from the ground surface onto an image plane and stores it as digital data. The digital data subjected to radiation correction and sensor correction is input in step S1. In the case of the QuickBird satellite, the spatial resolution of remotely sensed image data is about 61 cm (directly below) to 72 cm (25 degrees off nadir angle), and it can capture an image of the ground surface of about 16.5 km square at a time. it can.

  The advantage of using satellite images is exactly this specification. If the resolution of the satellite image, the vehicle can be automatically recognized and tracked, and it is possible to analyze the vehicle behavior at the same time in a wide area network by obtaining an image of about 16.5 km square at a time. become. In normal aerial photography, the area captured at one time is extremely narrow compared to satellite images. For example, to obtain a geographical area of 16.5km square, about 100 photographs are taken and each is corrected individually. Need to be joined together. Since aerial photography is generally silver halide photography, this splicing work will be very labor intensive. On the other hand, since the satellite image is digital data, it is suitable for processing by software. In addition, in the case of aerial photographs, since a large amount of shooting as described above takes several hours, there is a problem that it is impossible to grasp the traffic situation at the same time. As described above, the satellite image is optimal for a traffic condition investigation.

  The satellite image is picked up as described above. In step S1, a plurality (for example, two pieces) of image data obtained by picking up an area including the investigation target area at a predetermined time interval (Δt) is input. For example, the QuickBird satellite captures two areas including the same area from two directions by changing the direction of the sensor while the satellite moves from one north to the south in one orbit, and acquires this as a stereo pair image. Therefore, a plurality of image data can be acquired using this function. Alternatively, when a plurality of satellites can be used in cooperation, images captured from the plurality of satellites at predetermined time intervals may be input. In this case, it is possible to freely set the imaging time interval, which is advantageous in that an optimal imaging time interval can be set.

  In step S1, not only image data but also image support data (ISD) is input. The image support data includes at least information on the position and time of the satellite that captured the image. Image support data includes, for example, attitude data (first data point time, points, point interval and attitude information), satellite orbital calendar data (first data point time, points, point interval and satellite orbit information), geometric Correction data (parameters for photogrammetry of a virtual camera model that models satellite sensors and optical systems: focal length, central axis coordinates, etc.), image metadata (product level, coordinate values of four corners of the image (latitude, longitude) ), Main attributes such as image products including map projection information, image acquisition time), RPC (Rapid Positioning Capability Extension Format) data (coordinate values of four corners of space and four corners of image) Mathematical data).

  In this embodiment, imaging data from an artificial satellite is used in this way, but the present invention is not limited to using an artificial satellite. An image obtained by other means may be used as long as the geographical area necessary for the area survey can be captured at a time and the obtained image has a resolution that allows automatic recognition and tracking of the vehicle. For example, imaging data from a currently planned stratospheric platform may be used.

In subsequent step S2, an ortho-transform is performed on each satellite image data input in step S1. The ortho conversion is a process of converting the center projection image into a vertical parallel projection image. In this specification, the satellite image after the ortho-transform is called an ortho image, and the ortho image of the image taken at time t ₁ is A ₀ , and the ortho image of the image taken at time t ₁ + Δt is B ₀ . Hereinafter, vehicle behavior observation processing is performed based on these ortho images A and B.

FIG. 4 shows an example of the ortho images A ₀ and B ₀ . In the figure, the rectangle on the road indicates the vehicle. The accuracy of vehicle behavior observation processing depends mainly on how accurately vehicles on the road can be extracted and tracked. In that respect, it is also conceivable that structures in areas other than the road cause misrecognition. Therefore, in this embodiment, as an option, images A ₁ and B ₁ in which areas other than the road are masked are created by superimposing map data on the ortho images A ₀ and B _{0 in} step S3. As described above, since the coordinate values of the four corners of the image are also input as the image support data, the map data can be easily superimposed by using this. An example of the ortho images A ₁ and B ₁ in which areas other than the roads of the ortho images A ₀ and B ₀ are masked is shown in FIG.

  Next, in step S4, only the vehicles that have moved during Δt are extracted. For example, after extracting vehicles from both images, a vehicle that has been stopped for Δt from among them is identified, and vehicles that have been stopped during Δt are excluded, so that Can be extracted. Thereby, the vehicle that has been stopped during the imaging time interval Δt can be excluded from the target of the vehicle tracking process performed in the following steps.

Such a technique is a known technique. If a more specific example is shown, for example, first, a vehicle on the road is detected from each of the ortho images A ₁ and B ₁ shown in FIG. 5 by a known template matching technique. Extract. Thereafter, binary images A ′ ₁ and B ′ ₁ are generated in which the extracted vehicle area is 1 and the other areas are 0. Next, the generated binary images A ′ ₁ and B ′ ₁ are subjected to a logical product of the same pixel positions, and the resulting binary image is extracted again by template matching. It is recognized that the vehicle extracted here has been stopped for Δt.

  Here, from the stop position on the road or the like, the vehicle was stopped for Δt as shown in FIG. 6 by analyzing whether the vehicle was stopped due to a signal or whether it was parked on the road for another reason. An image obtained by extracting only the vehicle may be separately generated and provided to the user. Such an image may be useful for, for example, examination of traffic light control and grasping the actual situation of parking and stopping on the road.

Then, from each of the ortho images A ₁ and B ₁ , ortho images A ₂ and B ₂ are created by deleting the vehicle that has been stopped for Δt. FIG. 7 is a diagram illustrating an example of the ortho images A ₂ and B ₂ from which the vehicle that has been stopped for Δt is deleted.

In subsequent steps S5 and S6, a search process for the travel route of each vehicle is performed based on the ortho images A ₂ and B ₂ from which the vehicles that have been stopped for Δt are deleted.

First, in step S5, candidates for the same vehicle are searched from the ortho image B ₂ for each vehicle of the ortho image A ₂ .

Figure 8 is a vehicle a10 of interest in orthorectified image A _2, is a diagram conceptually showing the process of searching from the orthorectified image B _2. By exploring the entire orthoimage B ₂ to find the vehicle a10 is wasteful. In the present embodiment, a range that is impossible in traffic control is excluded from the search range using GIS, and the search range is limited to a realistic range in which the vehicle of interest can move within the time Δt. .

For example, interest vehicle a10 in orthoimage A ₂ in FIG. 8, as indicated by the arrow in the figure, as the route pattern, the next node (i.e., intersection) straight, the left, or it is conceivable to turn right. Here, referring to GIS, for example, it is assumed that a road in a left turn direction is prohibited from entering. Then, since it is impossible in traffic control that the vehicle of interest a10 makes a left turn at the next intersection, this left-hand link (a road turned left) is excluded from the search range. Further, the distance L that the vehicle of interest a10 can move during Δt is expressed by the following equation.

            L ≦ S · Δt

  However, S represents a virtual value of the travelable speed of a10.

Based on these, the search range as shown in orthoimage B ₂ in FIG. 8 are set, the candidate of the same vehicle and focused vehicle a10 from this is searched. A known tracking algorithm can be used for this search. However, in this embodiment, one candidate is not specified at this stage, and a plurality of candidates are selected based on the matching distance, for example. In the illustrated example, the vehicle b4, b10 of the orthoimage B ₂ is selected as a candidate.

In a step S5 performs such processing for each vehicle in the orthorectified image A _2. FIG. 9 is a diagram conceptually showing processing for searching the ortho image B ₂ for the vehicle a 9 (vehicle in front of a 10) focused on in the ortho image A ₂ . In this case as well as the vehicle a10 described above, the link in the left direction (left-turned road) is excluded from the search range, and the distance that the vehicle of interest a10 can move during Δt is estimated as shown in FIG. A search range as shown in the ortho image B ₂ is set. Note that the search range is different from the search range shown in the ortho image B ₂ in FIG. In the example of FIG. 9, the vehicle b2 of orthoimage B _2, b3, and b5 are selected as candidates.

Thus, in step S5, for each vehicle in the ortho-image A _2, refine each moving range in consideration of the network structure and the moving distance of the vehicle of the road within the range to search for a candidate of the same vehicle .

Next, in step S6, the most likely identical vehicle candidate in the entire road network is identified. Here, for example, the candidates for each vehicle are narrowed down to one so as to minimize the matching distance of the entire road network, not the matching distance for each individual vehicle used in step S5. This is a so-called delayed decision. FIG. 10 is a diagram showing an example of a result of this delayed decision, in which the vehicles in the ortho image A ₂ are arranged in a line and the same vehicles in the ortho image B ₂ are associated with each other. Here, (see Figure 8) the same vehicle candidate b4 or where a was the b10 in the orthoimage B ₂ of the vehicle a10, in this example, the vehicle b4 is the entire road network better it is judged to be the same as the vehicle a8 since the matching distance is reduced, accordingly, the same vehicle in the orthoimage B ₂ of the vehicle a10 is specified as b10. That is, it is determined that the vehicle a10 has made a right turn.

In this way, in step S6, all the same vehicles are specified between the ortho images A ₂ and B ₂ .

  In step S7, a moving vector diagram of the vehicle as shown in FIG. 11, for example, is created and displayed on the CRT 6 as the observation result. According to such a moving vector diagram, the vehicle behavior at a predetermined time (Δt) becomes clear at a glance.

  Further, the speed V, specifically, for example, the spatial average speed indicating the movement distance of the vehicle group per unit time may be calculated and displayed.

  It is also possible to obtain and display the traffic density K indicating the number of vehicles present on a unit section of the road at a certain moment. This traffic density can be determined directly from satellite images. This is in contrast to conventional fixed-point observation, where traffic density can only be obtained as an estimated value.

  In addition, it is also possible to arbitrarily create and display observation results in various modes. For example, the traffic volume Q can be calculated and displayed. The traffic volume is an index indicating the number of vehicles passing through one section of the road per unit time, and [units / h], [units / day], etc. are used as units.

  In step S7, at least one of the traffic volume Q, the traffic density K, and the speed V may be displayed, or a KV curve, a QV curve, or a correlation between these pieces of information may be displayed. The KQ curve may be estimated, created, and displayed by combining with other fixed point observation data.

  The traffic volume Q, traffic density K, and speed V data are data representing the characteristics of traffic flow having a wide area consistency at the same time. If these are given as initial values and input into a network-type traffic flow simulator such as AVENUE or SOUND together with conventional fixed-point observation data having temporal continuity, the actual situation of traffic conditions is simulated more accurately than conventional methods. It becomes possible.

  As a result, it will contribute to detailed traffic planning, including traffic management and parking management of not only trunk roads but also district roads (community roads).

  In this way, the traffic situation in a predetermined area can be easily grasped quantitatively and in a plane using the computer system.

  The embodiment of the present invention has been described in detail above. However, as described above, the present invention can be realized by a general computer system. Therefore, the program itself installed in the computer system in order to implement the functional processing of the present invention on a computer, or a recording medium storing the program is also within the scope of the present invention.

  The recording medium for supplying the program includes a hard disk, a magneto-optical disk, a memory card, and the like. In addition, the program can be supplied via a network such as the Internet.

It is a figure which shows the structure of the traffic condition observation system in embodiment. It is a schematic diagram which shows the program and data which are memorize | stored in the hard-disk apparatus of the traffic condition observation system in embodiment. It is a flowchart which shows the outline | summary of the process of the traffic condition observation system in embodiment. It is a figure which shows an example of the ortho image obtained by performing ortho transformation with respect to the input satellite image. It is a figure which shows an example of the ortho image by which areas other than the road were masked. It is a figure which shows an example of the image which extracted only the vehicle stopped during (DELTA) t. It is a figure which shows an example of the ortho image from which the vehicle stopped during (DELTA) t was deleted. It is a figure explaining the search process of the same vehicle between two ortho images. It is a figure explaining the search process of the same vehicle between two ortho images. It is a figure explaining the search process of the same vehicle by a delayed decision. It is a figure which shows an example of the movement vector figure of a vehicle.

Claims (11)

A traffic condition observation system for quantitatively observing traffic conditions in a predetermined area,
A first image obtained by imaging a region including the predetermined region at a first time by remote sensing, and a region including the predetermined region by a remote sensing at a second time after a predetermined time from the first time. First storage means for storing the second image that has been recorded;
Second storage means for storing geographical information including at least traffic control information of roads in the predetermined area;
Extraction means for extracting a region of the vehicle based on each of the first and second images stored in the first storage means;
Search means for searching each vehicle in the first image from the second image based on the extraction result by the extraction means;
Presenting means for presenting a predetermined observation result relating to traffic conditions based on a search result by the search means;
A traffic condition observation system characterized by comprising:   The traffic condition observation system according to claim 1, wherein the first and second images are captured by remote sensing from a satellite.   The traffic condition observation system according to claim 1 or 2, wherein the search means includes means for excluding a vehicle that has been stopped during the predetermined time from a search target.   The search means limits a search range in the second image for each vehicle in the first image based on a road network structure and a distance that the vehicle can move within the predetermined time. The traffic condition observation system according to any one of claims 1 to 3, wherein candidates for the same vehicle are searched from.   The search means obtains a plurality of candidates for the same vehicle in the second image for each vehicle in the first image, and then specifies the most likely candidate for the same vehicle in the entire road network. The traffic condition observation system according to any one of claims 1 to 4. The geographical information includes map data including the predetermined area,
Means for creating first and second images in which areas other than roads are masked by superimposing the map data on each of the first and second images stored in the first storage means; Have
The traffic condition observation system according to claim 1, wherein the extraction unit extracts a region of the vehicle based on each of the first and second images in which regions other than the road are masked.   The traffic condition observation system according to claim 1, wherein the presenting means presents a moving vector diagram of a vehicle as the predetermined observation result.   The traffic condition observation system according to claim 1, wherein the presenting unit presents at least one of traffic volume, traffic density, and vehicle speed as the predetermined observation result. A data processing method using a computer system for quantitatively observing traffic conditions in a predetermined area,
A first image obtained by imaging a region including the predetermined region at a first time by remote sensing, and a region including the predetermined region by a remote sensing at a second time after a predetermined time from the first time. A first storage step for storing the second image thus obtained in a memory;
A second storage step of storing, in a memory, geographical information including at least traffic control information of a road in the predetermined area;
An extraction step of extracting a region of the vehicle based on each of the first and second images stored in the memory;
A search step of searching each vehicle in the first image from the second image based on the extraction result of the extraction step;
Based on the search result of the search step, a presentation step for presenting a predetermined observation result relating to traffic conditions;
A data processing method characterized by comprising: In order to quantitatively observe traffic conditions within a given area,
A first image obtained by imaging a region including the predetermined region at a first time by remote sensing, and a region including the predetermined region by a remote sensing at a second time after a predetermined time from the first time. A first storage step for storing the second image thus obtained in a memory;
A second storage step of storing, in a memory, geographical information including at least traffic control information of roads in the predetermined area;
An extraction step of extracting a region of the vehicle based on each of the first and second images stored in the memory;
A search step of searching each vehicle in the first image from the second image based on the extraction result of the extraction step;
A presenting step of presenting a predetermined observation result relating to traffic conditions based on the search result of the search step;
A program for running In order to quantitatively observe traffic conditions within a given area,
A first image obtained by imaging a region including the predetermined region at a first time by remote sensing, and a region including the predetermined region by a remote sensing at a second time after a predetermined time from the first time. A first storage step for storing the second image thus obtained in a memory;
A second storage step of storing, in a memory, geographical information including at least traffic control information of roads in the predetermined area;
An extraction step of extracting a region of the vehicle based on each of the first and second images stored in the memory;
A search step of searching each vehicle in the first image from the second image based on the extraction result of the extraction step;
A presenting step of presenting a predetermined observation result relating to traffic conditions based on the search result of the search step;
A computer-readable storage medium storing a program for executing the program.