JP2019046209A - Vehicle position acquisition system and vehicle position acquisition method - Google Patents

Abstract

To solve a problem in a prior art, and provide a vehicle position acquisition system and vehicle position acquisition method capable of more precisely obtaining the position of a vehicle, as compared with the prior art, by using a motion picture taken by imaging an intersection.SOLUTION: A vehicle position acquisition system of the present invention acquires the position of a vehicle traveling on a road, and includes orientation element calculation means and planar position calculation means. The orientation element calculation means extracts a motion picture component image, collates the motion picture component image with a base orthographic projection image, and calculates an orientation element employed when the motion picture component image is taken. The planar position calculation means obtains the planar position of the vehicle contained in the motion picture component image on the basis of the orientation element and the base orthographic projection image.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a technology for acquiring the position of a moving vehicle, and more specifically to a vehicle position acquisition system and method for acquiring the position of a vehicle using an image forming a moving image. is there.

  With the enforcement of the revised Road Traffic Act of 2014, planar intersections in the form of roundabouts have been introduced in various places. In principle, this roundabout is a roundabout which is controlled according to the rule that vehicles entering the roundabout take a slow pace, giving priority to traffic on the roundabout among circular flat junctions without traffic control by traffic lights. The UK and other foreign countries have had many achievements in the past because they have the advantage of reducing accidents caused by encounters and being able to run with confidence because they do not install traffic lights.

  On the other hand, in Japan, it is not long after introduction, and knowledge about geometrical structure design about roundabout is not sufficient, and it is related to vehicle behavior (planar trajectory, temporal change of speed, acceleration, etc.) and geometrical structure. Investigation and analysis are being conducted in various places.

  Not only roundabouts but also intersections are places where traffic in different directions is overlapped, and places where collisions or contact accidents are likely to occur. In addition, at many intersections, traffic signals are used to control the traffic of both intersections, causing bottlenecks in traffic and contributing to traffic congestion in some places.

  In any case, at intersections including roundabouts, since the traffic form is significantly different from that of a general traveling section, monitoring and investigation have been carried out so far to analyze the behavior of vehicles in the intersection. For example, Patent Document 1 proposes a technique for acquiring oblique images of a vehicle passing an intersection by a plurality of imaging means, and monitoring the situation of the intersection with an orthographic projection image created from these oblique images.

JP 2001-118183 A

  For monitoring and surveys conducted to understand the behavior of vehicles at intersections, methods for acquiring still images and moving pictures are mainstream because the results can be viewed. In particular, in the case of focusing on the movement of the vehicle, it is conceivable to use a moving image from which continuous images can be obtained. In addition, in order to grasp the behavior of the vehicle, it is desirable to acquire the position of the vehicle, and for this purpose, it is effective to use an orthographic projection image including the vehicle in the intersection as described in Patent Document 1 . However, since it is extremely difficult to always take a moving vehicle as an image from above (that is, an orthographic projection image), it is usually difficult to convert an image taken from an oblique direction to obtain an orthographic projection image. There is.

  When using a moving image, a still image extracted from the moving image is used to obtain an orthographic projection image. However, it is known that the resolution of an image extracted from a moving image is generally inferior to that of an image acquired as a still image. In order to understand the behavior of the vehicle, it is necessary to track the motion of the vehicle, and for that purpose it is necessary to connect the individual orthographic projection images, but if the image with poor resolution is used, the orthographic projection image As a result, it is possible to point out the problem that the position of the vehicle can not be accurately grasped and, therefore, the behavior of the vehicle can not be grasped accurately.

  An object of the present invention is to solve the problems of the prior art, that is, a vehicle position acquisition system capable of obtaining the position of the vehicle more accurately than the prior art using a moving image obtained by photographing an intersection, and a vehicle position It is to provide an acquisition method.

  The invention of the present application focuses on the point of calculating an external determination element of a moving image obtained by photographing a vehicle using an orthographic projection image (base orthographic projection image) serving as a reference of position, which has not heretofore been available. It is an invention made on the basis of ideas.

  The vehicle position acquisition system according to the present invention is a vehicle position acquisition system for acquiring the position of a vehicle traveling on a road, and includes a positioning element calculation unit and a plane position calculation unit. Among them, the localization element calculation means extracts a moving image constituent image (an image constituting a moving image) and compares the moving image constituent image with the base orthographic projection image to obtain a positioning element when the moving image constituent image is photographed ( It is a means to calculate the photographing position and the photographing direction). One plane position calculation means is a means for obtaining the plane position of the vehicle included in the moving image constituting image based on the positioning element and the base orthographic projection image. The base orthographic projection image is an orthographic projection image of a ground surface including a road, and is an image to which coordinates are given.

  The vehicle position acquisition system of the present invention may further include traveling vector calculation means. The travel vector calculation means is a means for obtaining the travel vector of the vehicle based on the two-time plane positions obtained by the plane position calculation means for the same vehicle.

  The vehicle position acquisition system of the present invention may further include vehicle selection means. The vehicle selection means is a means for selecting a specific vehicle from a plurality of different vehicles, and selects the specific vehicle based on the travel vector of the specific vehicle obtained by the travel vector calculation means.

The vehicle position acquisition method of the present invention is a method of confirming the position of a vehicle traveling on a road from a moving image, and is a method including a moving image shooting step, a positioning element calculation step, and a plane position calculation step. Among them, in the moving image shooting step, a moving object including a vehicle is shot by the flying object, and in the orientation factor calculating step,
By extracting the moving image constituting image and comparing the moving image constituting image with the base orthographic projection image, a positioning element when the moving image constituting image is photographed is calculated. The plane position of the vehicle included in the moving image configuration image is determined based on

The vehicle position acquisition system and the vehicle position acquisition method of the present invention have the following effects.
(1) Since the moving image is used, the position of the traveling vehicle can be acquired continuously or intermittently.
(2) Although the image extracted from the moving image is used, the position of the vehicle can be determined more accurately than before by comparing it with the base orthographic projection image whose position is clear.

BRIEF DESCRIPTION OF THE DRAWINGS The block diagram which shows the main structures of the vehicle position acquisition system of this invention. The model figure which shows the synthetic | combination image which synthesize | combined the roundabout and the vehicle typically. FIG. 6 is a model diagram schematically showing a composite image in which a vehicle image is cut out for each of a plurality of moving image constituent images and these vehicle images are superimposed on one base orthographic projection image. FIG. 3 is a model diagram schematically showing a composite image in which a plurality of different vehicle images are superimposed on one base orthographic projection image. The flow figure showing the flow of the main processes of the vehicle position acquisition method of the present invention. FIG. 10 is a partial flow chart showing a flow of steps of calculating a traveling vector and selecting a specific vehicle.

  An example of an embodiment of a vehicle position acquisition system and a vehicle position acquisition method of the present invention will be described based on the drawings.

1. Definitions In describing examples of embodiments of the present invention, the definitions of terms used herein are first presented.

(Video composition image)
A moving image is composed of a large number of images (still images). Here, each of the images constituting the moving image is referred to as a "moving image constituting image". The present invention is to obtain the position of a vehicle traveling on a road, and thus the moving image configuration image includes a vehicle traveling.

(Positioning element)
The parameters obtained as a result of orienting (external orientation) of the camera at the time of shooting are three-dimensional coordinates of the camera position at the time of shooting (hereinafter referred to as "shooting position") and inclinations of three axes (hereinafter referred to as "shooting direction"). ). Here, the shooting position (X, Y, Z) and the shooting direction (ω, φ,)) are collectively referred to as “alignment elements”.

(Base orthographic projection image)
Here, an orthographic projection image (orthophoto) of the ground surface including the road is referred to as a "base orthographic projection image". However, three-dimensional coordinates are given to this base orthographic projection image. For example, an orthographic projection image in which a three-dimensional terrain model created from a plurality of photographs is displayed (two-dimensionally) on a plane (two-dimensionally), or a surface model DTM (Digital Terrain Model) or a surface model DSM (Digital Surface Model) An orthographic projection image etc. which displayed such a numerical terrain model on a plane etc. are mentioned as a base orthographic projection image.

(Traveling vector)
Here, the moving direction of the moving vehicle is referred to as a “traveling vector”. Specifically, a traveling vector is obtained by determining the traveling direction from the coordinates of two or more points through which a traveling vehicle has passed. Also, when the passing time of the vehicle is known, the traveling speed between points can be obtained, and in this case, the traveling speed can be included in the traveling vector.

2. Vehicle Position Acquisition System Next, a vehicle position acquisition system of the present invention will be described with reference to FIG. FIG. 1 is a block diagram showing the main configuration of a vehicle position acquisition system 100 according to the present invention. In addition, although the example of the vehicle which travels the roundabout shown in FIG. 2 is demonstrated here for convenience, the present invention can be applied to a vehicle traveling on any road such as a general traveling section besides an intersection. it can.

  As shown in FIG. 1, the vehicle position acquisition system 100 is configured to include orientation element calculation means 110 and plane position calculation means 120, and can also be configured to include each means shown in FIG. Hereinafter, the main elements of the vehicle position acquisition system 100 will be described in detail.

(Base orthographic projection image creation means)
The base orthographic projection image creation means 131 is a means for literally creating a base orthographic projection image, and is appropriately created corresponding to the type of input data to be prepared. For example, FIG. 1 shows an example of creating a base orthographic projection image based on a plurality of still images. Hereinafter, the procedure of creating a base orthographic projection image will be specifically described.

  First, a plurality of still images are photographed by the still image photographing means 132. At this time, the image is taken so as to include the road (in this case, a roundabout), and to minimize the inclusion of a vehicle. The still image photographing means 132 may use a camera for acquiring a still image, or may use a moving image photographing means 141 which will be described later. However, in the case of using the moving image shooting means 141, it is desirable to obtain a still image after adjustment (for example, interval shooting etc.) so as to obtain a relatively high resolution. When multiple still images are obtained, a three-dimensional numerical terrain model of the target terrain is created using photogrammetry technology such as bundle adjustment or SFM (Structure from Motion). It is the 3D model creation means 133 shown in FIG. 1 that creates this 3D numerical terrain model. The 3D model creating means 133 may be executed by a computer.

  Since the 3D numerical terrain model is given 3D coordinates, an orthographic projection image (orthophoto) can be created based on this. Specifically, the base orthographic projection image creation unit 131 performs a spatial operation to create a base orthographic projection image from the three-dimensional numerical terrain model. The base orthographic projection image creation means 131 may be executed by a computer. The created base orthographic projection image is stored in the base orthographic projection image storage means 134. When the base orthographic projection image is prepared in advance, the base orthographic projection image creating unit 131, the still image capturing unit 132, the 3D model creating unit 133, and the base regular projection image storage unit 134 are not necessarily vehicle positions. There is no need to include it in the acquisition system 100.

(Video shooting means)
The moving image shooting means 141 takes a moving image so as to include a vehicle traveling on the road (in this case, a roundabout) along with the road. When shooting a moving image, the moving image shooting means 141 may be fixed to a predetermined facility and shot so as to include the entire target range, or even when a person holds the moving image shooting means 141 and shoots from a high position Alternatively, the moving image capturing means 141 may be attached to a flying object such as a UAV (Unmanned Air Vehicle) and the image may be taken from above. The moving image photographed by the moving image photographing unit 141 is stored in the moving image storage unit 142 by batch processing or real time processing.

(Positioning element calculation means)
The orientation element calculating unit 110 extracts a moving image constituent image, and calculates a positioning element (photographing position and photographing direction) when the extracted moving image constituent image is photographed. The orientation factor calculation means 110 may be executed by a computer. Hereinafter, the procedure for calculating orientation elements of the moving image composition image will be specifically described.

  First, while extracting a predetermined moving image constituent image from the moving image storage means 142, the base orthographic projection image is read out from the base orthographic projection image storage means 134. As described above, the base orthographic projection image is given three-dimensional coordinates, and includes many characteristic points (feature points). In other words, the base orthographic projection image has many reference points. On the other hand, although the moving image constituting image is an image (hereinafter referred to as "oblique image") which is often photographed from an oblique direction, since a part of the base orthographic projection image is photographed, There are also some reference points included. Therefore, the reference point included in the moving image configuration image is extracted by comparing the moving image configuration image with the base orthographic projection image, and the input condition is a single photo orientation, a relative orientation such as SFM, or a photogrammetry such as bundle adjustment. By utilizing the technology, it is possible to calculate orientation elements of the moving image constituent image. Among them, in single photo orientation, a single moving image configuration image is used, and from the three-dimensional coordinates at a plurality of reference points included in the image and the plane coordinates in the image (image coordinates in the image coordinate system) Calculate the orientation factor. Further, in the relative orientation and bundle adjustment, the orientation elements of the moving image configuration image are calculated while creating a three-dimensional numerical terrain model based on the plurality of moving image configuration images and the reference points.

(Plane position calculation means)
The plane position calculation unit 120 calculates the plane position of the vehicle included in the moving image configuration image. The plane position calculation means 120 may be executed by a computer. The planar position of the vehicle can be represented by an arbitrary point such as the front end or the rear end, or can be represented by an area of the vehicle surrounded by a plurality of points. Alternatively, the three-dimensional model of the vehicle may be created and then the plane position of the vehicle may be calculated. In addition, the plane position of a vehicle here is not a position (image coordinates) in an image but a relative (or absolute) position on a real road, and more specifically base orthographic projection It is the position coordinates given in the same coordinate system as the coordinates given to the image. As described above, the moving image constituting image is an oblique image, but the orientation elements of the moving image constituting image are obtained, and the moving image constituting image includes the reference point of the base orthographic projection image. The plane position can be calculated. The plane position (coordinates) of the vehicle calculated by the plane position calculation unit 120 is stored in the vehicle plane position storage unit 121.

(Composite image creation means)
The composite image creation unit 151 composites the vehicle included in the moving image constituent image and the base orthographic projection image to create a composite image. Specifically, an image of the vehicle portion (hereinafter referred to as "vehicle image") is cut out from the moving image configuration image, and the vehicle image is projected onto the base orthographic projection image based on the plane position of the vehicle calculated by the plane position calculation means 120. Create a composite image as shown in FIG. In other words, the background image (the terrain image other than the vehicle) of the moving image constituting image is replaced with the base orthographic projection image. That is, in FIG. 2, the vehicle 200 portion is formed of the moving image configuration image, but the portion excluding the vehicle is formed of the base orthographic projection image. Since the moving image constituting image is an oblique image, the vehicle image is also an oblique image, and one of the base orthographic projection images is naturally an orthographic projection image. Therefore, only the vehicle image in the composite image is an oblique image, and the other area is an orthographic projection image.

  Although it has been described that the composite image creation unit 151 cuts out a vehicle image in order to create a composite image, this can be set as a specification that the operator operates while confirming the moving image configuration image. It is also possible to automatically cut out a vehicle image. Whichever specification is adopted, the composite image creation unit 151 may be executed by a computer.

  So far, the process of extracting the moving image configuration image from the moving image, calculating the orientation elements of the image, and obtaining the plane position of the vehicle to create the composite image has been described. Repeatedly, it is possible to track the movement of the vehicle. That is, when moving image constituent images are extracted at predetermined time intervals (periodically or irregularly) and composite images are created for the respective moving image constituent images, it is possible to grasp the movement of the vehicle (a so-called trajectory). Instead of creating a composite image for each moving image configuration image, a vehicle image is cut out for each moving image configuration image, and a composite image (FIG. 3) in which these vehicle images are superimposed on one base orthographic projection image is created It is also good. The composite image created by the composite image creation unit 151 can be output by an output unit 152 such as a display or a printer.

(Travel vector calculation means)
The traveling vector calculation unit 160 is a unit that calculates a traveling vector of the vehicle. Specifically, the positions of the vehicles are calculated from two or more different moving image configuration images (each including the same vehicle), and the traveling vector is calculated based on the positions of the vehicles. At this time, if the time at which the moving image constituting image was captured is recorded, the traveling speed can be obtained and included in the traveling vector. The travel vector calculation means 160 may be executed by a computer.

(Vehicle selection means)
When a composite image in which a plurality of vehicle images are superimposed on one base orthographic projection image is created, different vehicles may be included in one composite image as shown in FIG. In order to track the movement of the vehicle from such a composite diagram, it is necessary to distinguish the vehicle. For example, in FIG. 4, the first vehicle 201 and the second vehicle 202 are included, and the first vehicle 201b before movement, the first vehicle 201a after movement, and the second vehicle 202b before movement are further included. The second vehicle 202a after movement is included, and in order to track the movement of the vehicle, the first vehicle 201 and the second vehicle 202 have to be selected from the four vehicle images.

  The vehicle selection means 170 selects a specific vehicle from the composite image including different vehicles (that is, distinguishes the vehicles). The vehicle selection means 170 may be implemented by a computer. Further, the vehicle selection means 170 may be configured to select a specific vehicle using the traveling vector. The traveling vector of the first vehicle 201b before movement shown in FIG. 4 is acquired, the next movement position is estimated from the traveling vector, and the vehicle at the closest position among the vehicles in FIG. 4 (in this case, after movement) The first vehicle 201a) is selected. Here, the traveling vector of the first vehicle 201b before movement is calculated based on the position of the vehicle in front of it. The vehicle selection means 170 may be configured to automatically select different vehicles using image recognition technology, in place of the specifications using travel vectors.

3. Vehicle Position Acquisition Method Subsequently, a vehicle position acquisition method of the present invention will be described with reference to FIG. FIG. 5 is a flow chart showing the flow of the main steps of the vehicle position acquisition method of the present invention, showing the steps to be carried out in the middle row, and the left row shows the input information necessary for the step in the right row Indicates the output information generated from the process. The vehicle position acquisition method according to the present invention is a method implemented using the vehicle position acquisition system 100 according to the present invention, and therefore the description overlapping with the contents described in “2. Vehicle position acquisition system” is avoided here. Only content specific to the vehicle position acquisition method will be described. That is, the contents not described here are the same as those described in “2. Vehicle position acquisition system”. Of course, the terms defined in advance apply to both the description here and “2. Vehicle position acquisition system”.

  As shown in FIG. 5, in order to implement the vehicle position acquisition method of the present invention, it is preferable to start by creating a base orthographic projection image. For example, a plurality of still images are photographed (Step 10), and a 3D numerical terrain model is created from the still images using a technique such as SFM (Step 20), and the space operation is performed to obtain a 3D numerical terrain model An orthographic projection image is created (Step 30). In the case where a base orthographic projection image is prepared in advance, the process of photographing a still image (Step 10) to the process of creating a base orthographic projection image (Step 30) can be omitted.

  When the base orthographic projection image is obtained (or in parallel with the creation of the base orthographic projection image), the moving image is photographed to include the road (in this case, the roundabout) and the vehicle traveling on the road (Step 40) ). Then, a necessary moving-image constituting image is extracted from the moving image (Step 50), and a positioning element of the moving-image constituting image is calculated (Step 60), and a plane position of the vehicle is calculated (Step 70). When the plane position of the vehicle is obtained, a vehicle image is cut out from the moving image configuration image (Step 80), and the composite image is created by superimposing the vehicle image on the base orthographic projection image (Step 90). By checking the composite image obtained here, the position of the target vehicle can be grasped. In addition, it is possible to grasp the movement of the vehicle by repeatedly performing a series of processes of the extraction process (Step 50) of the moving image constituent image to the creation process (Step 90) of the composite image according to the traveling of the vehicle.

  When plane positions of two or more times of the same vehicle are obtained, it is also possible to calculate a traveling vector as shown in FIG. In FIG. 6, n (n times) planar positions are acquired for the same vehicle, and the traveling vector of the vehicle is calculated based on the information (Step 100). When different vehicles are included in one composite image, the specific vehicle can be selected based on the traveling vector of the target vehicle (specific vehicle) (Step 110).

  The vehicle position acquisition system and the vehicle position acquisition method according to the present invention, for example, can grasp the behavior of a vehicle or a pedestrian at an intersection, and manage road traffic such as a country or a local government or road traffic. Especially useful for the police. By carrying out the present invention, it is possible to grasp the behavior of the vehicle etc. at the intersection, and as a result, it is possible to take appropriate measures against the accident, which leads to the prevention of traffic accidents at the intersection. Inventions can be said to be not only industrially applicable but also socially significant.

100 Vehicle Position Acquisition System 110 Targeting Element Calculation Means 120 Plane Position Calculation Means 121 Vehicle Plane Position Storage Means 131 Base Orthographic Projection Image Creation Means 132 Still Image Shooting Means 133 3D Model Creation Means 134 Base Orthographic Projection Image Storage Means 141 Movie Shooting Means 142 moving image storage means 151 composite image creation means 152 output means 160 travel vector calculation means 170 vehicle selection means 200 vehicle 201 first vehicle 201a first vehicle 201b after movement first vehicle 201b first movement 202 second vehicle 202a second vehicle after movement 202b second vehicle before movement

Claims (4)

A vehicle position acquisition system for acquiring the position of a vehicle traveling on a road from a video,
An orientation element calculating unit that calculates a shooting position and a shooting direction when the moving image configuration image is captured by extracting the moving image configuration image that configures the moving image and comparing the moving image configuration image with the base orthographic projection image When,
Plane position calculating means for obtaining a plane position of a vehicle included in the moving image constituting image based on the photographing position, the photographing direction, and the base orthographic projection image;
The base orthographic projection image is an orthographic projection image of a ground surface including a road, and coordinates are given thereto.
Vehicle position acquisition system characterized in that. Travel vector calculation means for obtaining a travel vector of a vehicle based on plane positions of two timings obtained by the plane position calculation means for the same vehicle;
The vehicle position acquisition system according to claim 1, further comprising: The vehicle further comprises vehicle selection means for selecting a specific vehicle from a plurality of different vehicles,
The vehicle selection means selects the specific vehicle based on the travel vector of the specific vehicle obtained by the travel vector calculation means.
The vehicle position acquisition system according to claim 2, characterized in that: A method of acquiring a plane position of a vehicle traveling on a road from a video,
A moving image capturing step of capturing the moving image including a vehicle by a flying object;
An orientation element calculation step of calculating a photographing position and a photographing direction when the moving image constituting image is photographed by extracting the moving image constituting image constituting the moving image and comparing the moving image constituting image with the base orthographic projection image When,
A plane position calculating step of obtaining a plane position of a vehicle included in the moving image constituting image based on the photographing position, the photographing direction, and the base orthographic projection image;
The base orthographic projection image is an orthographic projection image of a ground surface including a road, and coordinates are given thereto.
A vehicle position acquisition method characterized in that.

Images