JP2010151658A - Apparatus and method for measuring position of mobile object - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To correctly measure a position of a mobile object 8 by a simple structure, and display a correct and real two-dimensional map image or three-dimensional scenic image based on the measurement results. <P>SOLUTION: An apparatus for measuring the position of a mobile object includes a real image feature extraction unit 21 for extracting a characteristic part from a real image taken by an imaging part 4 mounted on the mobile object 8, a scenic feature specifying part 23 for detecting the position and an azimuth angle of the mobile object 8 to extract or specify scenic information or the characteristic part from the scenic information, and a feature matching part 24 for matching the characteristic part of the real image with the characteristic part specified by the scenic feature specifying part 23. The position of the mobile object 8 is accurately measured by the feature matching. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention uses a real image obtained by imaging the moving direction of a moving object, position information acquired from a GPS (Global Positioning System) satellite and the like, and landscape information stored in association with the position information. The present invention relates to a moving body position measuring device and a measuring method thereof.

  In recent years, navigation systems mounted on vehicles and the like have become widespread. This navigation system acquires position information from a GPS satellite or the like, reads map data stored in a CD-ROM or DVD based on this position information, and displays the current vehicle position on a display device. This provides convenience that allows the current vehicle position to be recognized at a glance as the position on the map.

  However, generally, the positional information acquired from the GPS satellite includes an error of 10 m or more. Therefore, when the position information from the GPS satellite is directly associated with the map data, an image in which the vehicle is not traveling on the road is displayed. In order to correct this, a map matching process is performed. The map matching process is based on position information obtained from GPS satellites and movement information generated from travel distances and angular velocity changes obtained from travel sensors and gyro sensors mounted on mobile bodies. The road having the highest characteristic is selected from the map data, and the moving body is forcibly moved to the center of the selected road for correction. Thereby, it displays so that a mobile body may not remove from a road. Further, in order to match the position of the moving body to the actual position, the road surface in the moving direction of the moving body is imaged, the white line drawn on the captured road surface is recognized, and the position accuracy of the moving body is improved by this white line information. A method has also been proposed (see, for example, Patent Document 1).

Furthermore, a navigation system that displays a three-dimensional CG image has been developed. Based on the position information of the moving body from the GPS or the like, the current position and azimuth on the map data stored on the DVD or the like are specified, and the three-dimensional CG image stored in association with the map data is read and displayed. To do. According to this type of navigation system, it is possible to display a landscape that changes every moment as the moving body moves as a CG image. If landmarks such as buildings and signs are displayed together with 3D images, the running position can be confirmed quickly even on rainy days or at night, or when a large vehicle runs ahead and the view is blocked. Has convenience.
JP 2007-220013 A

  However, the current position of the moving body obtained by the map matching process is the center of the traveling road. Therefore, a road map or a landscape image read from a recording medium such as a CD-ROM or DVD is an image in which the traveling direction is designated at the center position of the road. FIG. 9 shows an example in which map information is displayed from a position obtained by such map matching processing.

  FIG. 9A shows a display example of a monitor image in which a vehicle is displayed on a road map based on the position of the vehicle corrected by map matching. The actual vehicle 53 is traveling in the left lane 51 of the road, but is displayed on the display as if traveling on the center separation line 52 of the road. FIG. 9B is an example in which a three-dimensional CG image acquired from a storage medium is displayed on a monitor based on the vehicle position and azimuth angle corrected by map matching. The actual vehicle 53 is traveling in one of the left three lanes, but in the three-dimensional CG image displayed on the monitor, a screen traveling on the center separation line 52 of the road is displayed. In particular, when the center of the road is a right turn lane of an oncoming vehicle, it will run in the opposite lane, increasing the risk. Therefore, the display as it is cannot be applied to an actual navigation system. Also, even when recognizing the white line drawn on the road surface and correcting the position of the moving body from the white line, it is difficult to determine which lane the vehicle is traveling on when there are multiple lanes that can be driven It is.

  Therefore, an object of the present invention is to enable display of a map image and a landscape image that match an actual running state by enabling measurement of the actual position of the moving object with high accuracy.

  In the present invention, the following means have been taken in order to solve the above problems.

  In the invention, an imaging unit that is mounted on a moving body and captures a lane drawn on a road surface in a direction in which the moving body travels and a landscape in the direction to acquire a real image, and a feature portion is extracted from the real image A real image feature extraction unit, a position information detection unit that detects the position and azimuth of the moving body, a map information storage unit that stores landscape information associated with a road on the map, and the detected position The landscape information is acquired from the map information storage unit using an azimuth, and a landscape feature specifying unit that specifies a feature part included in the acquired landscape information, a feature unit of the real image, and the landscape feature specification A mobile body position measurement comprising: a feature matching processing unit that performs a matching process on the characteristic part specified by the unit; and a mobile body position estimation unit that estimates a position of the mobile body that captured the real image based on a result of the matching process. apparatus It was.

  The landscape information is composed of a landscape image captured from a plurality of positions at predetermined intervals along the road in hopes of a runway direction or a feature extracted from the landscape image. The feature specifying unit acquires the landscape information from the map information storage unit when the moving object approaches a specific point on the road, and specifies the feature unit included in the landscape information.

  Further, the points provided at predetermined intervals along the road are intersections on the map.

  The landscape information is composed of a three-dimensional landscape image capable of displaying a landscape in relation to the position and azimuth of the moving object, and the landscape feature specifying unit is based on the position and azimuth of the moving object. The 3D landscape image is acquired from the map information storage unit, and the feature portion is extracted from the 3D landscape image to identify the feature portion.

  The map information storage unit further stores road information including the number of lanes and the lane width of the road, and the landscape feature specifying unit acquires the road information from the map information storage unit and the moving body is present. A possible lane is identified, a feature included in the landscape information is identified for each lane based on a position and an azimuth of a moving body for each lane, and the feature matching processing unit is a feature of the real image And the characteristic part specified from the landscape information is subjected to a matching process for each lane in which the moving body can exist.

  The image processing apparatus further includes a white line recognition unit that recognizes a white line drawn on a road surface from the captured real image, the white line recognition unit recognizes a white line drawn on the road surface from the real image, and The position of the moving body is corrected by detecting the offset distance from the white line.

  Further, in the present invention, a real image feature extraction step for capturing a real image by capturing the front of the mobile object and extracting a feature portion of the real image, and a position for detecting the position and azimuth of the mobile object A detection step; a map information is acquired based on a position and an azimuth angle of the mobile body; a possible position estimation step for estimating a possible position of the mobile body; and The landscape information associated with the position and the azimuth is acquired, and a landscape feature specifying step for specifying a feature included in the landscape information is performed, and a matching process is performed between the feature of the real image and the feature of the landscape information. The moving body position measuring method includes a feature matching processing step and a moving body position estimating step for estimating the position of the moving body based on the result of the feature matching processing.

  The real image feature extraction step estimates a white line recognition step for recognizing a white line drawn on the road surface from the captured real image, and an offset distance between the moving object and the white line from the recognized white line. An offset distance estimating step and a moving body position correcting step for correcting the position of the moving body based on the offset distance are provided.

  Further, the landscape feature specifying step specifies, for each lane, a feature part included in the landscape information based on a position and an azimuth angle of the mobile object in a lane in which the mobile object can exist, and the feature matching processing step includes The feature matching process is performed for each lane between the feature portion of the real image and the feature portion included in the landscape information specified for each lane.

  In the present invention, an imaging unit that is mounted on a mobile body and captures a lane drawn on the road surface in the direction in which the mobile body travels and a landscape in that direction, and acquires a real image, and a feature portion is extracted from the real image. A real image feature extraction unit, a position information detection unit that detects the position and azimuth of the moving body, a map information storage unit that stores landscape information associated with roads on the map, and the detected position and azimuth A landscape feature specifying unit for acquiring landscape information from the map information storage unit and specifying a feature included in the acquired landscape information, a feature of a real image, and a feature specified by a landscape feature specifying unit And a feature matching processing unit that performs matching processing, and a moving body position estimation unit that estimates the position of the moving body that captured a real image based on the result of the matching processing. As a result, it is possible to improve the measurement accuracy of the current position of the moving body, so even when displaying a two-dimensional map image or a three-dimensional landscape image, the image is not unnatural. It can be widely applied to applications.

  Hereinafter, the present invention will be described in detail with reference to the drawings.

(Moving object position measuring apparatus according to Embodiment 1)
FIG. 1 is a functional block diagram showing the configuration of a mobile object position measuring apparatus 10 according to Embodiment 1 of the present invention. The moving body position measuring apparatus 10 includes a control unit 1 that controls the entire apparatus, a position information detection unit 3 that detects the position and azimuth angle of the moving body, and a position and azimuth angle detected by the position information detection unit 3. A position information processing unit 2 that processes position information on the basis of the position information and generates display image data of a road map image or a landscape image; an image pickup unit 4 that picks up the moving direction of the moving body; and image data of the captured real image. It comprises a real image storage unit 5 for storing, a map information storage unit 7 for storing map information and landscape information, and a display unit 6 for displaying road map images or landscape images.

  Hereinafter, each component will be described. The control unit 1 includes a CPU 11, a ROM 12 and a RAM 13. The ROM 12 stores a main program for controlling the basic operation of the moving body position measuring apparatus 10 and various application programs for processing position information. The CPU 11 reads out each program from the ROM 12 to the RAM 13, executes a calculation using the RAM 13 as its work area, and controls the operation of the moving body position measuring apparatus 10. The CPU 11 realizes each component by executing an application program corresponding to each component of the position information processing unit 2. That is, each component of the position information processing unit 2 is basically composed of software.

  The position information detection unit 3 includes a GPS unit 15, a gyro sensor 16, and a travel sensor 17. The GPS unit 15 receives a carrier wave transmitted from a plurality of GPS satellites, and detects a position including the current latitude and longitude from the difference in arrival time of the carrier waves. The gyro sensor 16 detects the azimuth angle of the moving body from the change in the angular velocity of the moving body. The travel sensor 17 detects the travel distance and travel speed of the moving body. Since the carrier wave sent from the GPS satellite is once per second, the position of the moving body is complemented by the gyro sensor 16 and the traveling sensor 17. Even when it is difficult to receive a carrier wave from GPS hygiene due to a tunnel or a building in an urban area, the position of the moving body can be estimated. In addition, a magnetic sensor can be installed together with the gyro sensor 16, and the azimuth angle of a moving body can be specified from geomagnetism. In addition, the position of the moving body detected by the GPS unit 15 can be corrected by receiving radio waves transmitted from the base station. The position information detected by the position information detection unit 3 and the position information of the azimuth are transmitted to the position information processing unit 2.

  The imaging unit 4 images a landscape and a road surface in the moving direction of the moving body. The imaging unit 4 can be configured by a CCD imaging device or a CMOS imaging device. The real image storage unit 5 temporarily stores image data of a real image captured by the imaging unit 4. The display unit 6 displays a map image and a landscape image generated by the position information processing unit 2. The display unit 6 can be configured by a liquid crystal display device or an EL display.

  The map information storage unit 7 stores map information and landscape information stored in association with the map information, and includes a large-capacity storage device such as a CD-ROM or a DVD. The landscape information is, for example, image data of a 3D landscape image or a 2D map image, or feature portion data including a landscape feature portion viewed from a traveling path of a moving body. The map information storage unit 7 can store road information associated with the map information. The road information is an attribute of a road that includes information such as a distinction between a national road, a prefectural road, a general road, the number of lanes, and a lane width.

  The position information processing unit 2 determines the position of the moving object based on the position information detected by the position information detection unit 3, the actual image captured by the imaging unit 4, and the landscape information stored in the map information storage unit 7. Is measured with high accuracy. The position information processing unit 2 stores map information based on at least the actual image feature extraction unit 21 that extracts a feature from the actual image captured by the imaging unit 4 and the position information detected by the position information detection unit 3. A landscape feature specifying unit 23 that acquires landscape information from the unit 7 and specifies a feature part included in the landscape information, and a feature matching processing unit that performs a matching process between the feature part of the real image and the feature part included in the landscape information 24, a moving object position estimating unit 25 that estimates the position of the moving object based on the result of the matching process, and an image processing unit 26 that generates a road map image or a three-dimensional landscape image based on the estimated position of the moving object. And.

  The position information processing unit 2 further includes a white line recognition unit 27 that detects an offset distance with respect to the white line of the moving body from the actual image captured by the imaging unit 4, and road information of the road on which the moving body is located from the map information storage unit. The map information acquisition unit 22 that acquires and identifies the number of lanes and the lane width, and the map matching processing unit 28 that matches the position of the moving object to the road on the map based on the position information detected by the position information detection unit 3. Can be provided.

  Hereinafter, the function of each component of the position information processing unit 2 will be specifically described. The real image feature extraction unit 21 extracts the feature portion of the real image from the real image data captured by the imaging unit 4. The characteristic part of the real image includes, for example, the length of the line segment included in the real image, its coordinates, and the coordinates of the intersection where the line segment intersects the line segment. When buildings and trees are included in the captured real image, the line segment formed by the outline of the building and the window frame and the intersection of these line segments are calculated from the brightness difference and saturation difference of the pixels that make up the actual image. Can be extracted. The length of the line segment and the interval between the intersections vary depending on the imaging position where the actual image is captured. That is, the feature portion extracted from the actual image includes position information of the imaging position on the road. Note that the feature part is not limited to the line segment or the intersection point included in the real image. For example, the characteristic arc length, the arc radius, or the arc center coordinates included in the real image can be used as the feature part. . This is because a natural scene is included in a real image, and thus it may be possible to more accurately express a feature part by extracting an arc than a straight line.

  Based on the position information detected by the position information detection unit 3, the map information acquisition unit 22 acquires map information, road attribute information, and landscape information around the current position of the mobile object. These pieces of information are repeatedly acquired one after another according to the azimuth angle and speed of the moving body.

  The landscape feature specifying unit 23 specifies the feature portion of the landscape image from the landscape information including the three-dimensional CG image read from the map information storage unit 7. For example, from the three-dimensional landscape image as the landscape information, the length of the characteristic line segment included in this image and its coordinates, and the coordinates of the intersection where the line segment and the line segment intersect are used as the feature part. In addition to line segments and intersections, characteristic arc lengths, arc radii, or center coordinates of arcs included in the three-dimensional landscape image can be used as feature portions. As will be described later, since the feature matching process between the feature part of the landscape image and the feature part of the real image is performed, the conditions for extracting the feature part from the landscape image are the same as the conditions for extracting the feature from the real image. Set to. In addition, the position of the image component which comprises a three-dimensional landscape image, and the position of the viewpoint of a landscape image are defined correctly.

  In addition, the map information storage unit 7 was picked up from a plurality of positions with a desired road direction at points provided at predetermined intervals along the road instead of or together with the 3D landscape image. A landscape image can be stored in association with a point on the road. In this way, since it is not necessary to generate a three-dimensional landscape image, the feature matching process can be performed in a shorter time. Moreover, the map information storage part 7 can memorize | store only the feature data of the feature part previously extracted from the landscape image of the point provided at the said predetermined | prescribed space | interval as landscape information. Since the data amount of the feature part of the landscape can be much smaller than the data amount of the three-dimensional landscape image, the storage capacity of the map information storage unit 7 can be greatly reduced. Moreover, since the process step of extracting the feature portion from the landscape image can be omitted, the feature matching process can be performed in a shorter time. As a result, it is possible to shorten the delay time until the map image or the landscape image is displayed based on the position information measured from the actual image capturing time.

  The feature matching processing unit 24 performs feature matching processing between the actual image feature extracted from the actual image and the landscape image feature included in the landscape information acquired from the map information storage unit 7. In the feature matching process, the feature part of the real image is compared with the feature part of the landscape image, and the difference is calculated. Specifically, assuming a plurality of positions where the vehicle can exist, feature portions are extracted from a landscape image of a landscape desired from each position. Furthermore, the feature part of a landscape image and the feature part of a real image are compared, and a difference is calculated. The position where the moving object can exist is calculated from the number of lanes and the lane width in the road attribute information acquired by the map information acquiring unit 22 based on the position of the moving object corrected by the map matching process described later. Can be set. Alternatively, without using the number of lanes or the lane width, it is possible to set a position that is close to the position of the moving body corrected by the map matching process and that is separated from each other as the position where the moving body can exist.

  The feature matching process will be specifically described with reference to FIG. By the map matching process described later, the position of the moving body 8 is corrected to the position Po at the center of the road. The map information acquisition unit 22 acquires road information on the number of lanes and the lane width at the position Po from the map information storage unit 7. FIG. 4A shows a state in which the road surface on which the moving body 8 of the road map acquired in this way can travel is viewed from above. A lane L1 is between the white line 31 and the white line 32 on the road, a lane L2 is between the white line 32 and the white line 33, and a lane L3 is between the white line 33 and the white line 34. That is, the lane where the moving body 8 can exist is defined as a position P1 in the lane L1, a position P2 in the lane L2, and a position P3 in the lane L3. The feature matching processing unit 24 acquires, from the map information storage unit 7, a three-dimensional landscape image in which the moving direction of the moving body 8 is viewed from the position P1, the position P2, and the position P3. And a feature part is extracted from each three-dimensional landscape image, the feature matching process of the feature part of a real image and the feature part extracted from the three-dimensional landscape image in each position is performed, and the difference d1 in the position P1, the position P2 And the difference d3 at the position P3 are calculated. The moving body position estimation unit 25 estimates the position where the difference is the smallest as the current position of the moving body based on the feature matching processing result. In addition, the position of the moving body can be estimated with higher accuracy using this difference.

  In addition, when the map information storage part 7 has memorize | stored the characteristic part as landscape information, instead of extracting a characteristic part from a three-dimensional landscape image, the memorize | stored characteristic part and the characteristic part of a real image The feature matching process can be performed between. In this case, it is not necessary to extract the feature from the 3D landscape image each time. For this reason, the processing speed is high, and the delay in displaying on the display unit 6 can be reduced.

  The moving body position estimation unit 25 estimates the position of the moving body at the time when the actual image is captured by evaluating the processing result of the feature matching process. Specifically, as described with reference to FIG. 4A, the position where the mobile object can be present with the smallest difference obtained by the feature matching process is estimated as the current position of the mobile object. Further, the position of the moving body can be estimated with higher accuracy from the tendency of the difference between the positions where each moving body can exist. For example, if the feature matching process is performed with three or more positions where the mobile object can exist, the position of the mobile object can be specified two-dimensionally. In addition, the position of the moving body can be estimated more accurately using the offset distance detected by the white line recognition unit 27 described next.

  The white line recognizing unit 27 recognizes a white line of the actual image captured by the image capturing unit 4. This will be described with reference to FIG. FIG. 5A represents a white line detected by edge processing of an actual image. The edge of the white line is detected by the difference processing of the real image data. Next, Hough transform (hereinafter referred to as Hough transform) and inverse transform are performed to detect the edge line segments e1 and e2 of the left white line H1 and the edge line segments e3 and e4 of the right white line H2. As a result, the offset distance Dof from the center of the left white line at the position Px that is the viewpoint of the imaging unit can be obtained. Therefore, the offset distance is obtained by the white line recognition unit 27, and the feature map matching process can be performed in consideration of the offset distance Dof. Thereby, the position of the moving body can be estimated with higher accuracy.

  The map matching processing unit 28 extracts the road candidates on the map based on the position information supplemented by the gyro sensor and the traveling sensor based on the latitude and longitude from the GPS satellite, and selects the road with the highest probability of the traveling road. A process for forcibly moving the moving object is performed. Thereby, it is possible to eliminate the unnaturalness that the moving body travels outside the road on the map.

  The image processing unit 26 performs image processing for generating an image to be displayed on the display unit 6. Based on the current estimated position of the mobile object, the map image and the three-dimensional landscape image obtained by reading the map information and the three-dimensional landscape image from the map information storage unit 7 are displayed on the display unit 6.

(Moving object position measuring method of Embodiment 2)
FIG. 2 is a flowchart showing a moving body position measuring method according to Embodiment 2 of the present invention. FIG. 3 is a flowchart showing the offset distance detection method. FIG. 4 is a schematic diagram showing a possible position of a moving object and a three-dimensional landscape image after map matching processing, and FIG. 5 is a schematic diagram for explaining an offset distance Dof detected by white line recognition processing. FIG. 6 is a schematic diagram showing the position of the moving body and the three-dimensional landscape image after estimating the position of the moving body.

  Hereinafter, the moving body 8 will be described as a vehicle such as an automobile. When the moving body position measuring device 10 starts measuring the position of the moving body 8, the imaging unit 4 images the road surface and the landscape in the moving direction of the moving body 8. The real image storage unit 5 acquires and stores image data of a real image captured by the imaging unit 4. The real image feature extraction unit 21 acquires the image data of the real image stored in the real image storage unit 5 and extracts the feature part of the real image (step S1). The feature portion of the real image can be extracted by various methods. For example, the feature portion is extracted from a line segment included in the real image or a distribution of intersections where the line segment and the line segment intersect.

  The position information detector 3 detects the position and azimuth angle of the moving body 8 and outputs position information (step S2). The GPS unit 15 receives a carrier wave from a GPS satellite and generates position information including the latitude and longitude of the moving body 8 from the phase difference and the like. The gyro sensor 16 detects a change in the angular velocity of the moving body 8 and generates azimuth angle information including the azimuth angle of the moving body. The azimuth angle may be received from a GPS satellite and corrected by a gyro sensor. Further, the azimuth angle may be detected by a geomagnetic sensor.

  The map information acquisition unit 22 acquires map information associated with the current position and the current azimuth of the moving body 8 based on the position information detected by the position information detection unit 3 from the map information storage unit 7. The map matching processing unit 28 performs map matching processing based on the map information acquired by the map information acquiring unit 22 and corrects the position of the moving body to the center of the road on the map. The map information acquisition unit 22 acquires map information based on the position on the road and the azimuth angle corrected by the map matching process, and acquires the number of lanes and the lane width of the road at the current position from the map information. The number of lanes at the acquired current position is estimated as a position where the moving object can exist (step S3).

  The state of step S3 will be described with reference to FIG. FIG. 4A shows a position where the moving body 8 can exist. The moving body 8 is corrected to the position P0 by the map matching process. Accordingly, the positions P1, P2, and P3 of the lanes L1 to L3 that cross in the direction orthogonal to the moving direction of the moving body 8 are the positions where the moving body can exist. FIG. 4B shows a state in which a three-dimensional landscape image is displayed after the map matching process. Since the moving body 8 after the map matching process is corrected to the position Po at the center of the road, it is above the central separation zone. Since the three-dimensional landscape image is acquired from the map information storage unit 7 based on the position Po of the moving body 8 and the azimuth angle in the movement direction, the viewpoint of the three-dimensional landscape image is a central separation zone. As already explained, it is not appropriate to display a 3D landscape image from the map matching process.

  The white line recognition unit 27 detects the offset distance of the moving body in the lane from the image data of the actual image captured by the imaging unit 14. The offset distance is a distance from the center of the left white line drawn on the road surface to the moving body position. By detecting this offset distance, it is possible to further improve the estimation accuracy of the position where the moving object can exist.

  FIG. 3 is a flowchart illustrating an example of a method for obtaining the offset distance. FIG. 5A shows a white line detected by the white line recognition unit 27, and FIG. 5B shows a movable body existable position corrected by the detected offset distance Dof. The white line recognition unit 27 detects the edge of the white line drawn on the road surface from the image data of the actual image (step S10). Edge detection is performed by luminance difference processing of pixels constituting an actual image. Next, an edge image in real space is generated from the detected edge of the white line (step S11). Next, an edge line segment is estimated by performing Hough transform and inverse transform in the Hough space (step S12). Next, center line segments C1 and C2 of the respective edge line segments are obtained from the estimated two edge line segments H1 and H2 (step S13). Next, an offset distance Dof between the current position Px of the moving body 8 and the center line segment C1 is obtained (step S14). The moving body position estimation unit 25 sets the position where the moving body can be corrected using the offset distance Dof. That is, the position P′1 of the lane L1, the position P′2 of the lane L2, and the position P′3 of the lane L′ 3 are set.

  Returning to FIG. 2, the map information acquisition unit 22 determines positions P ′ 1, P ′ 2, P ′ 3 and the moving object for each set movable object existence possible position P ′ 1, P ′ 2, P ′ 3. A three-dimensional landscape image as landscape information is acquired from the map information storage unit 7 using the azimuth angle. The feature matching processing unit 24 extracts a feature portion from each of the three-dimensional landscape images viewed from the set positions P′1, P′2, and P′3 (step S4). These features are different from each other because the viewpoints for viewing the landscape are different.

  Next, a matching process is performed between the feature portion of the real image and the feature portion of the three-dimensional landscape image corresponding to each of the positions P′1, P′2, and P′3. The moving body position estimation unit 25 estimates the current position Px of the moving body from the matching processing results at the three points (step S5). Since there are three positions where the mobile object can exist, three matching processing results can be obtained. Thereby, in addition to the lane in which the moving body 8 is currently traveling, the position of the moving body 8 in the moving direction can also be estimated. As a result, the position of the moving body 8 can be estimated with high accuracy.

  FIG. 6 shows an example in which a map image or a three-dimensional landscape image is displayed based on the estimated moving body position (step S6). FIG. 6A shows a state in which the moving body 8 whose position is estimated is displayed on a two-dimensional road, and FIG. 6B shows a state in which the moving direction is viewed from the estimated position of the moving body. An example of a three-dimensional landscape image is shown. As shown in FIG. 6A, the moving body 8 can be displayed on the lane where the vehicle is currently located. Further, if the offset distance is estimated using the white line recognition unit 27, the position of the moving body 8 in the lane can also be displayed with high accuracy. Furthermore, as shown in FIG. 6B, if a three-dimensional landscape image is generated from the estimated position and azimuth of the moving body 8, the landscape viewed from the traveling lane of the moving body 8 can be displayed. . As a result, a natural image without a sense of incongruity can be displayed, and driving safety can be improved.

  It should be noted that feature data obtained by extracting a feature portion in advance as landscape information is stored in the map information storage portion 7, and this feature portion data is acquired based on the position and azimuth angle of the moving body 8, and feature matching processing is performed. It can be performed. In this way, since it is not necessary to store the 3D landscape image data, the amount of data can be reduced. Furthermore, since it is not necessary to extract a feature from the three-dimensional landscape image for each position where the moving object can exist during the feature matching process, the estimation processing speed for estimating the current position of the moving object is shortened, and the display unit 6 The shift between the map information displayed on the screen and the actual position of the moving object can be reduced.

(Moving object position measuring apparatus of Embodiment 3)
FIG. 7 is a schematic diagram of a map for explaining the moving body position measuring apparatus 10 and the measuring method according to the third embodiment of the present invention. In this embodiment, the map information storage part 7 memorize | stores the characteristic part extracted from landscape information or landscape information for every intersection of a road.

  FIG. 7 shows an intersection K1 and an intersection K2 ahead. The map information storage unit 7 stores a landscape image as landscape information regarding each intersection. The landscape image is a landscape in which the X1 direction and the Y1 direction are desired from the four positions Q11, Q12, Q13, and Q14 with respect to the intersection K1. Similarly, regarding the intersection K2, it is a landscape in which the X2 direction and the Y2 direction are desired from the four positions Q21, Q22, Q23, and Q24. In addition, the coordinate (X1, Y1) and the coordinate (X2, Y2) are set to X1, X2 as the moving direction when the moving body 8 travels from the left side to the right side.

  Specifically, landscape images Wx11 and Wy11 obtained by capturing a landscape in which the + X1 direction and the -Y1 direction are desired from the position Q11 of the intersection K1 are created. Landscape images Wx12 and Wy12 obtained by capturing a landscape in which the + X1 direction and the + Y1 direction are desired from the position Q12 are created. Similarly, landscape images Wx13 and Wy13 are created from a landscape in which the -X1 direction and -Y1 direction are desired from the position Q13, and landscape images Wx14 and Wy14 are created from a landscape in which the -X1 direction and + Y1 direction are desired from the position Q14. The eight created landscape images Wx11, Wy11, Wx12, Wy12, Wx13, Wy13, Wx14, and Wy14 are stored in the map information storage unit 7 in association with the intersection K1. Similarly, at the intersection K2, a landscape that desires the + X2 direction and the -Y2 direction from the position Q21, a landscape that desires the + X2 direction and the + Y2 direction from the position Q22, and a landscape that desires the -X2 direction and the -Y2 direction from the position Q23, A landscape image obtained by capturing the landscape in which the −X2 direction of Q24 and the + Y2 direction are desired is stored in the map information storage unit 7 in association with the intersection K2. The same applies to other intersections.

  Based on the position information detected by the position information detection unit 3, the map information acquisition unit 22 of the mobile body 8 acquires, from the map information storage unit 7, map information in the vicinity where the mobile body 8 travels. When the control unit 1 of the moving body 8 detects that the moving body 8 has approached the intersection K2 based on the position and azimuth detected by the position information detection unit 3, the imaging unit 4 determines the moving direction of the moving body 8. The real image is captured and stored in the real image storage unit 5. The real image feature extraction unit 21 extracts a feature portion from the captured real image. The position of the moving body 8 is corrected to the center of the road on the map by map matching processing. Further, the white line recognition unit 27 recognizes the white line drawn on the road surface and estimates the offset distance Dof. The moving body position estimation unit 25 sets a position where the moving body can exist from these corrected position information. In FIG. 7, the moving body 8 is traveling on a two-lane road on one side. Therefore, the movable body existence possible positions to be set are two positions corrected by the offset distance Dof of the two lanes.

  On the other hand, the feature matching processing unit 24 selectively acquires the landscape images Wx11 and Wx12 from the position and azimuth angle of the moving body 8 detected by the position information detection unit 3. The feature matching processing unit 24 extracts each feature from the image data of the landscape image Wx11 and the landscape image Wx12. As already described, the feature part can be configured from, for example, a line segment or an intersection included in the landscape image. The feature matching processing unit 24 performs a feature matching process on the feature part of the real image and the feature part extracted from the landscape image for each of the movable object existence possible positions of the movable object 8 estimated by the movable object position estimation unit 25. And output the result of the feature matching process. The moving body position estimation unit 25 estimates, from the feature matching processing result, the position where the feature of the landscape image is closest to the feature of the real image as the current position of the moving body. Based on this estimation result, the image processing unit 26 displays the moving body position on the map image and displays the map information on the display unit 6. In the above process, using the feature portions extracted from the landscape images Wx11 and Wx12, a feature portion having the position coordinates between the positions Q11 and Q12 as a variable is calculated. This is because it is necessary to perform feature matching processing with the feature portion of the real image using the feature portion corresponding to the position where the moving body 8 can exist.

  Thus, since the landscape seen from the intersection is stored in the map information storage unit 7 as a landscape image, the amount of stored data can be greatly reduced compared to the case of storing a three-dimensional landscape image. For this reason, the cost which produces map information can be reduced significantly. One of the most necessary information for the driver in the navigation system of the moving body is the lane information at the intersection and the position information of the own moving body. According to the present embodiment, these pieces of information can be presented with a simple configuration.

  In addition, in the said Embodiment 3, the map information storage part 7 was taken as the landscape image imaged from the specific position of the intersection as landscape information, and extracted the characteristic part from this landscape image. A feature portion may be extracted from a landscape image captured from a specific position of an intersection, and the feature portion data may be stored as landscape information. If comprised in this way, the data amount memorize | stored in the map information storage part 7 can further be reduced. Moreover, since it is not necessary to perform the process which extracts or calculates a feature part from a landscape image for every moving body presence possible position, the position of a moving body can be estimated in a short time.

  Moreover, you may make it memorize | store the landscape information of the point set at predetermined intervals along the road as the landscape information memorize | stored in the map information storage part 7. FIG. There are many intersections in cities, but there are few intersections in areas other than cities. Then, landscape information is memorize | stored as map information for every predetermined space | interval of a road. In this way, the position of the moving body on the road can be estimated with high accuracy even in an area with few intersections.

  Moreover, in the said Embodiment 3, although the point which images a landscape was made into four positions Q11, Q12, Q13, and Q14 of the corner | angular part of the intersection K1, it is not limited to this. The four positions Q11, Q12, Q13, and Q14 may be in a road lane or may be a point straddling the lane. In addition, when a median strip is erected at the center of the road and, for example, the X1 direction is imaged from the position Q12, if the landscape on the traveling side where the moving body 8 travels is hidden by the median strip, What is necessary is just to set two points and to image the driving | running | working side scenery from two points. In this case, there are 8 points to be imaged. In short, in the feature matching processing between the feature portion of the real image and the feature portion extracted from the landscape image, it is only necessary that the location where the real image is captured can be estimated. Further, the number of landscape images captured at the intersection is not limited to eight. Six landscape images are sufficient for a three-way intersection, and ten landscape images are required for a five-way intersection.

(Moving object position measuring apparatus of Embodiment 4)
FIG. 8 is a schematic view of a road for explaining the moving body position measuring apparatus 10 and its position measuring method according to Embodiment 4 of the present invention. The moving body 8 is traveling on a two-lane road on one side. Usually, the GPS unit 15 of the position information detection unit 3 receives a carrier wave sent from a GPS satellite at intervals of 1 second to detect the current position of the moving body. When position detection is performed in synchronization with a carrier wave from a GPS satellite and a 3D landscape image is displayed on the display unit 6, a discontinuous 3D landscape image is displayed. Therefore, it is desirable to be able to display a continuous three-dimensional image that complements the timing received from GPS hygiene. In addition, in that case, if the position of the moving body 8 can be estimated with high accuracy, the traveling safety can be improved.

  In FIG. 8, timings for receiving a carrier wave from a GPS satellite are time T1, T2, and T3. Times t1 to t4 are timings for generating a three-dimensional landscape image by complementing the period from time T1 to time T2 and from the next time T2 to time T3. First, the GPS unit 15 of the position information detection unit 3 receives a carrier wave from a GPS satellite and detects the position (T1) of the moving body 8. The gyro sensor 16 of the position information detection unit 3 detects the azimuth angle (T1) of the moving body 8. Thereafter, the corrected position of the moving body 8 at time T1 is estimated by the procedure shown in the flowchart of FIG. 2, and a three-dimensional landscape image is displayed on the display unit 6 based on the estimated position.

  Next, the position information detection unit 3 uses the position (t1) and the azimuth angle (t1) of the moving body 8 at the time t1 detected by the gyro sensor 16 and the traveling sensor 17 as the azimuth angle during the period T1 to t1. It is calculated from the change in distance and travel distance. The map information acquisition unit 22 estimates the position P (t1) of the moving object at time t1 based on the position (t1) and the azimuth angle (t1). The moving body position estimation unit 25 acquires lane information at the position (t1) from the map information storage unit 7 and sets a position where the moving body can exist. In this case, there are a left lane on which the moving body 8 travels and a right lane on the center line side. Further, the white line recognition unit 27 detects the offset distance Dof from the white line drawn on the road surface, and sets the position where the mobile object can be corrected corrected by the offset distance Dof. The feature matching processing unit 24 acquires from the map information storage unit 7 a three-dimensional landscape image of a point corresponding to each possible position of the moving body, and extracts a feature portion of each three-dimensional landscape image. The feature matching processing unit 24 performs matching processing between the feature part of the real image and the feature part corresponding to each possible position of the moving object. The moving body position estimation unit 25 estimates the current moving body position (t1) from the matching processing result. The image processing unit 26 acquires a three-dimensional landscape image from the map information storage unit 7 based on the moving body position (t1). The image processing unit 26 displays the three-dimensional landscape image on the display unit 6.

  Next, processing similar to the above is performed for time t2 with reference to time t1. As a result, a feature matching process is performed between the feature portion of the real image at time t2 and the feature portion of the three-dimensional landscape image acquired from the map information, and a new three-dimensional landscape image is displayed. Thereafter, this is repeated until time t4. Further, at time T2, similarly to time T1, a carrier wave from a GPS satellite is received to detect the position of the moving body, and the same processing is executed to display a three-dimensional landscape image.

  A continuous three-dimensional landscape image can be displayed by appropriately increasing the time division between the times T1 and T2. In this case, at each time tn (n is an integer equal to or greater than 1), the feature matching process is performed between the feature portion of the real image and the feature portion extracted from the landscape information, so the position of the moving object is always corrected. ing. Therefore, it is possible to display an accurate and realistic three-dimensional landscape image.

  Instead of extracting the feature from the three-dimensional landscape image, it was extracted from a landscape image taken in hope of the traveling direction from a plurality of positions at predetermined intervals along the road or from this landscape image. As described above, the feature matching process can be executed based on the landscape image or the feature extracted from the landscape image by storing the feature in the map information storage unit 7.

It is a functional block diagram showing the structure of the mobile body position measuring apparatus which concerns on embodiment of this invention. It is a flowchart figure showing the mobile body position measuring method which concerns on embodiment of this invention. It is a flowchart figure showing the detection method of the offset distance which concerns on embodiment of this invention. It is a figure for demonstrating the feature matching process of the moving body position measuring apparatus which concerns on embodiment of this invention. It is a figure for demonstrating the correction | amendment by the detection of the offset distance of the moving body position measuring apparatus which concerns on embodiment of this invention. It is a figure which shows the example of a display after the mobile body position estimation of the mobile body position measuring apparatus which concerns on embodiment of this invention. It is a figure for demonstrating the mobile body position measuring apparatus and measuring method which concern on embodiment of this invention. It is a figure for demonstrating the mobile body position measuring apparatus and measuring method which concern on embodiment of this invention. It is a figure showing the example of a display image by a conventionally well-known navigation system.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Control part 2 Position information processing part 3 Position information detection part 4 Imaging part 5 Real image storage part 6 Display part 7 Map information storage part 8 Mobile body 10 Mobile body position measuring apparatus

Claims (9)

An imaging unit that is mounted on a moving body and captures a lane drawn on a road surface in a direction in which the moving body travels and a landscape in that direction, and acquires a real image;
A real image feature extraction unit for extracting a feature from the real image;
A position information detection unit for detecting the position and azimuth of the moving body;
A map information storage unit for storing landscape information associated with roads on the map;
A landscape feature specifying unit that acquires the landscape information from the map information storage unit using the detected position and azimuth, and that identifies a feature included in the acquired landscape information;
A feature matching processing unit that performs matching processing between the feature of the real image and the feature specified by the landscape feature specifying unit;
A moving body position measurement apparatus comprising: a moving body position estimation unit that estimates a position of a moving body that has captured the real image based on a result of the matching process. The landscape information is composed of a landscape image picked up from a plurality of positions at predetermined points along the road in hopes of a runway direction or a feature extracted from the landscape image,
The landscape feature specifying unit acquires the landscape information from the map information storage unit when the moving object approaches a specific point on the road, and specifies a feature part included in the landscape information. The moving body position measuring device according to claim 1.   The mobile body position measuring apparatus according to claim 2, wherein the points provided at predetermined intervals along the road are intersections on a map. The landscape information is composed of a three-dimensional landscape image that can display a landscape in relation to the position and azimuth of the moving object,
The landscape feature specifying unit acquires the three-dimensional landscape image from the map information storage unit based on the position and azimuth of the moving body, and extracts the feature portion from the three-dimensional landscape image to specify the feature unit The moving body position measuring apparatus according to claim 1, wherein: The map information storage unit further stores road information including the number of lanes and the lane width of the road,
The landscape feature specifying unit acquires the road information from the map information storage unit, specifies a lane in which the moving body can exist, and uses the landscape information based on the position and azimuth of the moving body for each lane. Identify the included features for each lane,
The said feature matching process part performs the matching process for the characteristic part of the said real image, and the characteristic part specified from the said landscape information for every lane in which the said mobile body can exist. The moving body position measuring apparatus according to any one of the above. A white line recognition unit for recognizing a white line drawn on the road surface from the captured real image;
The white line recognition unit recognizes a white line drawn on a road surface from the actual image, detects an offset distance between the moving body and the white line, and corrects the position of the moving body. Item 6. The moving body position measuring apparatus according to any one of Items 1 to 5. A real image feature extraction step of capturing a real image by capturing the front of the moving body and extracting a feature of the real image;
A position detecting step for detecting a position and an azimuth angle of the moving body;
Obtaining map information based on the position and azimuth of the moving body, and a possible position estimating step for estimating a possible position of the moving body;
A landscape feature specifying step for acquiring landscape information associated with the position and azimuth of the mobile body from the position and azimuth of the mobile body, and specifying a feature part included in the landscape information;
A feature matching processing step for performing matching processing between the feature of the real image and the feature of the landscape information;
A moving body position measuring method comprising: a moving body position estimating step for estimating a position of the moving body based on the result of the feature matching process. The actual image feature extraction step includes:
A white line recognition step for recognizing a white line drawn on the road surface from the captured real image;
An offset distance estimation step for estimating an offset distance between the moving object and the white line from the recognized white line;
The moving body position measuring method according to claim 7, further comprising: a moving body position correcting step for correcting the position of the moving body based on the offset distance. The landscape feature specifying step specifies, for each lane, a feature included in the landscape information based on a position and an azimuth angle of the mobile body in a lane where the mobile body can exist,
The feature matching processing step performs feature matching processing for each lane on the feature portion of the real image and the feature portion included in the landscape information specified for each lane. 8. The moving body position measuring method according to 8.

Images