JP2006330908A - Position recording device and position recording method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a position recording device and method for performing the accurate position recording of an object to be recognized by carrying out recording processing under the consideration of the detection error of the object to be recognized. <P>SOLUTION: The image recognition of a sign is carried out based on image information picked up by a camera 2 (S12), and the position of the image-recognized sign is detected (S20), and the error of the detected position information of the sign is calculated (S22), and the position information of the sign and the position detection error information are recorded (S24), and the measurement errors of the sign position of map information and the position information of the sign detected this time are compared (S34), and the position information whose error is smaller is recorded as map data (S36). <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a position recording apparatus and a position recording method for recognizing an object to be recognized and recording position information of the object to be recognized.

Conventionally, as a device for recognizing a recognition object and recording position information of the recognition object, map information and a plurality of registered figures as recognition object candidates are stored as described in JP-A-2002-243469. In addition, there is known an apparatus that recognizes the presence and content of a registered figure in an image photographed by a camera when the vehicle travels, and updates map information based on the recognition result. This device is intended to promptly update map information in response to alterations such as road signs.
JP 2002-243469 A

  In such an apparatus, the map information is updated based on new information data. Therefore, when the information contains a large amount of detection error when the new information is acquired, the map information is updated. As a result, the updated information may be inaccurate. For example, when a road sign is image-recognized and the position of the road sign is detected by a sensor or the like, if the detection error is large, the position information of the road sign is shifted.

  Accordingly, an object of the present invention is to provide a position recording apparatus and a position recording method that can perform accurate position recording of a recognition target object by performing a recording process in consideration of a detection error of the recognition target object.

  That is, the position recording apparatus according to the present invention includes an imaging unit for imaging a recognition object, an image recognition unit that performs image recognition of the recognition object based on image information captured by the imaging unit, Position detecting means for detecting the position of the recognition object image-recognized by the image recognizing means; error calculating means for calculating an error of the position information detected by the position detecting means; position information of the recognition object and its And recording means for recording position detection error information.

  In the position recording apparatus according to the present invention, when the recording unit acquires a plurality of pieces of position information for the same recognition target object, the position information having the smallest position detection error is obtained from the position information. It is preferable to record it as position information.

  According to these inventions, the recognition object is photographed to recognize the image, and the position information of the recognition object is stored together with the position detection error information. For this reason, when the position information of the recognition object is acquired again, the position information already recorded is compared with the detection error, and the smallest error is recorded as the position information of the recognition object. , Accurate position recording can be performed.

  In the position recording apparatus according to the present invention, when the recording unit acquires a plurality of pieces of position information for the same recognition target object, the position information having the smallest position detection error is obtained from the position information. It is preferable to record in map data as position information.

  According to the present invention, when a plurality of pieces of position information are acquired for the same recognition target object, the smallest position detection error among the position information is recorded in the map data as the position information of the recognition target object. Thus, it is possible to update the map data by selecting position information with a small detection error. For this reason, appropriate map data can be updated and accurate map data can be constructed.

  In the position recording apparatus according to the present invention, it is preferable that the recognition object is a road sign.

  Further, the position recording method according to the present invention includes an image recognition step for performing image recognition of the recognition target based on image information captured by an imaging unit, and a position of the recognition target recognized by the image recognition step. A position detecting step for detecting the error, an error calculating step for calculating an error in the position information detected in the position detecting step, and a recording step for recording the position information of the recognition object and the position detection error information. Configured.

  Further, in the position recording method according to the present invention, when the recording step acquires a plurality of pieces of position information for the same recognition target object, the position information having the smallest position detection error is selected from the position information. It is preferable to record it as position information.

  According to these inventions, the recognition object is photographed to recognize the image, and the position information of the recognition object is stored together with the position detection error information. For this reason, when the position information of the recognition object is acquired again, the position information already recorded is compared with the detection error, and the smallest error is recorded as the position information of the recognition object. , Accurate position recording can be performed.

  Further, in the position recording method according to the present invention, when the recording step acquires a plurality of pieces of position information for the same recognition target object, the position information having the smallest position detection error is selected from the position information. It is preferable to record in map data as position information.

  According to the present invention, when a plurality of pieces of position information are acquired for the same recognition target object, the smallest position detection error among the position information is recorded in the map data as the position information of the recognition target object. Thus, it is possible to update the map data by selecting position information with a small detection error. For this reason, appropriate map data can be updated and accurate map data can be constructed.

  In the position recording method according to the present invention, it is preferable that the recognition object is a road sign.

  According to the present invention, by using the map information indicating the arrangement position of the recognition object, the image recognition process can be performed quickly and the recognition accuracy can be improved.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the description of the drawings, the same elements are denoted by the same reference numerals, and redundant description is omitted.

  FIG. 1 is a schematic configuration diagram of a position recording apparatus according to an embodiment of the present invention.

As shown in FIG. 1, a position recording apparatus 1 according to the present embodiment is an apparatus that recognizes a recognition object with respect to a captured image and records position information of the recognition object, and is mounted on a vehicle, for example. .
This embodiment demonstrates the case where a road sign is recognized as a recognition target object. The position recording device 1 includes a camera 2, a control processing unit 3, and a database 4.

  The camera 2 is an image pickup means for picking up an object to be recognized, and for example, a camera provided with an image sensor such as a CCD or C-MOS is used. The camera 2 is installed so as to image the front of the vehicle, for example. As this camera 2, it is preferable to use a stereo camera capable of stereo viewing. In this case, the three-dimensional position of the object in the image can be detected based on the images captured by the two cameras. Further, it can be used as a relative position detecting means for detecting a relative distance from the camera 2 to the recognition target imaged.

  Note that the camera 2 may be configured by a single camera. In this case, it is preferable to provide a measuring instrument that detects the relative position of the recognition object from the camera 2, such as a radar device.

  The control processing unit 3 is connected to the camera 2, inputs an image captured by the camera 2, performs image processing on the captured image, recognizes a recognition target in the captured image, and positions the recognition target. Information is recorded. The control processing unit 3 includes, for example, an image capturing unit 31, a sign recognition unit 32, a position deriving unit 33, a vehicle position / orientation / orientation detecting unit 34, an error deriving unit 35, a measurement point information recording unit 36, and a measurement point information holding unit 37. The measuring point categorizing unit 38, the measuring point determining unit 39, the information comparing unit 40, and the map information record updating unit 41 are functionally provided.

  The image capturing unit 31 captures an image captured by the camera 2. The sign recognition unit 32 recognizes a sign in the captured image by pattern matching or the like. The position deriving unit 33 derives the position of the sign that has been image-recognized based on the position, orientation, and attitude of the vehicle and the distance between the vehicle and the sign. The vehicle position / orientation / orientation detector 34 detects the position, orientation, and attitude of the vehicle. As means for detecting the position of the vehicle, for example, GPS (Global Positioning System) is used. The direction of the vehicle can be detected based on a change in the position of the vehicle, for example. The posture of the vehicle can be detected using, for example, a roll angle sensor, a yaw rate sensor, or the like.

  The error deriving unit 35 derives an error in the absolute position of the sign. For example, the error of the measurement distance to the sign by the camera 2 and the error of the measurement position of the vehicle by the GSP are calculated, and the maximum error of the absolute position of the sign is calculated from these errors. The measurement point information recording unit 36 records information such as the position of a sign (measurement point). When a plurality of pieces of sign information are acquired through a series of image recognition, each piece of information is recorded. The measurement point information holding unit 37 holds sign information.

  The measurement point categorizing unit 38 categorizes the information of the same label based on the measurement positions of the plurality of label information recorded as measurement points. The measurement point determination unit 39 selects a representative point from a plurality of categorized marker information. The information comparison unit 40 compares the sign information of the representative points with the sign information already recorded on the map data. The map information recording / updating unit 41 updates one of the sign information of the representative points compared and the sign information already recorded on the map data as map information.

  Next, the operation of the position recording apparatus and the position recording method according to this embodiment will be described.

  FIG. 2 is a flowchart showing a basic operation and a position recording method of the position recording apparatus according to this embodiment. The control process of FIG. 2 is executed by, for example, the control processing unit 3.

  First, as shown in S10 of FIG. 2, camera image acquisition processing is performed. This process is a process in which an image signal output from the camera 2 is taken into the control processing unit 3 and a captured image is read. The captured image is an image obtained by capturing the front of the vehicle and is stored as a digital image. When the image signal output from the camera 2 is not digitized, the control processing unit 3 may perform analog-digital conversion (A / D conversion).

  And it transfers to S12 and a label | marker recognition process is performed. The sign recognition process is a process for recognizing a sign in an image captured by the camera 2. For example, the sign recognition process is performed using a pattern matching technique. Templates of various signs used for pattern matching are set in advance, and those that show a correlation value with a template or more in a captured image are determined to be recognition objects. As the pattern matching, for example, difference summation, normalized correlation, neural network, or the like is used.

  And it transfers to S14 and a vehicle attitude | position information acquisition process is performed. This process is a process for acquiring the posture information of the vehicle on which the camera 2 is mounted. For example, posture information such as a roll angle, a pitch angle, a yaw angle, and a vehicle height of the vehicle is acquired by a sensor mounted on the vehicle.

  Then, the process proceeds to S16, and a relative position deriving process for the sign is performed. The sign relative position deriving process is a process for deriving the relative position between the image-recognized sign and the vehicle. For example, the relative position between the camera 2 and the sign is first derived, and then the relative position with respect to the vehicle is derived. Specifically, as shown in FIG. 3, when a stereo camera is used as the camera 2, the base line length between the imaging positions of the right camera and the left camera is b, and the focal length of the right camera and the left camera is f. If the imaging position coordinate value of the sign in the left camera is Pl (xl, yl) and the imaging position coordinate value of the sign in the right camera is Pr (xr, yr), the position coordinate value P of the sign in the camera coordinate system of the camera 2 (Xps, Yps, Zps) is expressed by the following equations (1) to (3).

Xps = b · (xl + xr) / 2 · (xl−xr) (1)
Yps = b · f / (xl−xr) (2)
Zps = b · yl / (xl−xr) (3)
As shown in FIG. 4, if the sensor origin of the camera 2 is translated by ax, ay, az on the Xc axis, Yc axis, and Zc axis of the vehicle coordinate system with respect to the vehicle origin, the vehicle coordinates of the sign The position coordinate value P (Xpc, Ypc, Zpc) in the system is expressed by the following equations (4) to (6).

Xpc = Xps + ax (4)
Ypc = Yps + ay (5)
Zpc = Zps + az (6)
Further, as shown in FIG. 4, when the Xc axis of the vehicle coordinate system is inclined by the roll angle φ with respect to the horizontal plane and the Yc axis of the vehicle coordinate system is inclined by the pitch angle ρ with respect to the horizontal plane, The position coordinate value P (Xpf, Ypf, Zpf) in the corrected vehicle coordinate system is expressed by the following equations (7) to (9).

Xpf = Xpc · cosφ + Ypc · sinφ · cosρ + Zpc · sinφ · sinρ (7)
Ypf = −Ypc · sinρ + Zpc · cosρ (8)
Zpf = Xpc · sinφ + Ypc · cosφ · cosρ + Zpc · cosφ · sinρ (9)
The relative position of the sign relative to the vehicle can be specified by the position coordinate value P (Xpf, Ypf, Zpf) of the vehicle coordinate system whose posture is corrected.

  And it transfers to S18 of FIG. 2 and a vehicle position direction information acquisition process is performed. This process is a process for acquiring information on the position and orientation of the vehicle on which the camera 2 is mounted. For example, the latitude and longitude of the vehicle origin in the vehicle coordinate system and the corrected vehicle coordinate system, the altitude, and the traveling direction (azimuth angle) of the vehicle are acquired based on the GPS detection information.

  Then, the process proceeds to S20, where the absolute position derivation process of the sign is performed. This process is a process for deriving the absolute position of the sign. For example, the absolute position of the sign is derived based on the position coordinate value P of the corrected vehicle coordinate system derived in S16 and the latitude, longitude, altitude, and azimuth angle of the vehicle acquired in S18.

  Specifically, as shown in FIG. 5, the position (ψp, λp) of the sign in the latitude / longitude coordinate system is derived based on the position (ψom, λom) of the vehicle. Further, as shown in FIG. 6, the vehicle position (Xcg, Ycg, Zcg) and the sign position (Xpg, Ypg, Zpg) in the world plane rectangular coordinate system are derived based on the position of the vehicle (ψom, λom). . The world plane rectangular coordinate system is a plane coordinate system in which a north point is an X axis and an east direction is a Y axis with a certain point as a reference point.

  Then, the process proceeds to S22, and measurement error derivation processing is performed. This process is a process for calculating an error in the absolute position of the sign. For example, the error of the measurement distance to the sign by the camera 2 and the error of the measurement position of the vehicle by the GSP are calculated, and the maximum error of the absolute position of the sign is calculated from these errors.

  For example, the maximum error of the measurement distance to the sign by the camera 2 is calculated according to the distance to the sign based on the angle of view of the camera 2, the number of pixels, the baseline length between the cameras, and the like. Further, the maximum error in the vehicle measurement position by GSP is estimated from the GPS radio wave reception state. Then, as shown in FIG. 7, the maximum error (Ec + Eg) of the absolute position of the sign is calculated based on the maximum error Ec of the measurement distance to the sign by the camera 2 and the maximum error Eg of the vehicle measurement position by the GSP.

  The maximum error of the absolute position of the sign is not only the error of the measurement distance to the sign by the camera 2 and the error of the measurement position of the vehicle by GSP, but also the measurement of the sensor that detects the posture such as the roll angle and yaw angle of the vehicle. The calculation may be performed in consideration of errors. In this case, a more accurate maximum error of the absolute position of the sign can be obtained.

  Then, the process proceeds to S24, and the absolute position (measurement result) of the sign and the maximum error of the position information are recorded and held. At that time, the type information of the sign is simultaneously recorded together with the position information of the sign. Next, the process proceeds to S26, where the same label categorizing process is performed. This process is a process of categorizing the position information as information of the same sign when a plurality of pieces of position information are recorded for the same sign.

  For example, as shown in FIG. 8, when a traveling vehicle continuously measures the position of the same sign, a plurality of different position data p1 to pn are recorded for the same sign due to a measurement error. In this case, the group of position data p1 to pn is categorized as position data of the same label under a predetermined condition. For example, the error circles of each position data overlap each other, the types of recognition objects are the same, the vehicle traveling direction at the time of position measurement is within a set value, and they are not simultaneously present in the same image. It is a condition for categorization.

  Here, the error circle is a circle having the position of the marker data as the center and the magnitude of the error as a radius or a diameter. As shown in FIG. 9, when the position data p1 to p3 are recorded, the error circle c1 of the position data p1, the error circle c2 of the position data p2, and the error circle c3 of the position data p3 overlap each other. This is a condition for categorizing the position data p1 to p3.

  Then, the process proceeds to S28 in FIG. 2, and a representative point is selected from the categorized position data group (measurement point group). For example, the point with the smallest maximum error in the measurement position is selected as the representative point. In this case, the position data group having the highest data content reliability can be selected as the representative point. In addition, as a selection method, there is a case where position data whose measurement distance to the sign is closest to a predetermined distance is selected as a representative point. At this time, any one of 10 to 20 m is set as the predetermined distance. As a selection method, an average position of position data may be selected as a representative point.

  And it transfers to S30 and the search process of a map database is performed. This process is a process of searching for data having the same label as the image-recognized label in the data already recorded in the database 4. For example, the error circle overlaps with the error circle of the representative point sign selected in S28, the object type is the same type as the representative point sign, and the measurement vehicle traveling direction is the representative point measurement vehicle traveling direction. It is searched whether recorded data that is within a predetermined angle exists.

  Then, the process proceeds to S32, and it is determined in S30 whether or not a sign corresponding to the representative point exists in the map information (map database). When it is determined in S30 that the sign corresponding to the representative point does not exist in the map information, the information of the representative point is recorded in the map data as the sign data (S38). This recording process includes, for example, the type of sign, the center position (latitude and longitude position) of the sign, the altitude of the sign, the height of the sign from the road or the ground, the azimuth angle of the sign display direction, the size and position of the sign This is done by recording each maximum error in time.

  On the other hand, when it is determined in S30 that the sign corresponding to the representative point exists in the map information, the measurement error of the sign of the map information is compared with the measurement error of the sign of the representative point. For example, the maximum error of the measurement position in the sign of the map information is compared with the maximum error of the measurement position of the sign (representative point) recognized this time.

  Then, the process proceeds to S36, and the one with the smallest maximum error among the measurement position of the sign in the map information and the measurement position of the sign recognized this time is stored in the map data as the position data of the sign. Then, the process proceeds to S40, and it is determined whether or not a termination condition for the control process is satisfied. When it is determined that the end condition is not satisfied, the process returns to S10. On the other hand, when it is determined that the termination condition is satisfied, the control process is terminated.

  As described above, according to the position recording apparatus and the position recording method according to the present embodiment, a sign that is a recognition target is photographed to recognize an image, and the position information of the sign is stored together with the position detection error information. For this reason, when the position information of the sign is obtained again, the position information that has already been recorded is compared with the detection error, and the smallest error is recorded as the position information of the sign. Can record.

  For example, when the latest information is recorded as the position information of the sign, if the latest measured position information includes a large error, the incorrect position information is updated. On the other hand, in the position recording apparatus and the position recording method according to the present embodiment, in such a case, old information with a small error is recorded as the position information of the sign and is not updated. Thus, accurate position information can be recorded.

  Further, according to the position recording apparatus and the position recording method according to the present embodiment, when a plurality of pieces of position information are acquired for the same sign, the position information with the smallest position detection error is obtained from the position information. By recording in the map data as information, the map data can be updated by selecting position information with a small detection error. For this reason, appropriate map data can be updated and accurate map data can be constructed.

  In the present embodiment, the case of recognizing a road sign on a road has been described. However, the position recording apparatus and the position recording method according to the present invention are not limited to the above, and traffic lights, buildings, public The present invention may be applied when recognizing images other than signs such as facilities.

1 is a schematic configuration diagram of a position recording apparatus according to an embodiment of the present invention. It is a flowchart which shows operation | movement of the position recording apparatus and position recording method which concern on embodiment of this invention. It is explanatory drawing of the label | marker relative distance calculation in the position recording device and position recording method which concern on embodiment of this invention. It is explanatory drawing of the vehicle attitude | position calculation in the position recording device and position recording method which concern on embodiment of this invention. It is explanatory drawing of the label | marker absolute position calculation in the position recording device and position recording method which concern on embodiment of this invention. It is explanatory drawing of the label | marker absolute position calculation in the position recording device and position recording method which concern on embodiment of this invention. It is explanatory drawing of the error calculation in the position recording apparatus and position recording method which concern on embodiment of this invention. It is explanatory drawing of the marker categorization in the position recording apparatus and position recording method which concern on embodiment of this invention. It is explanatory drawing of the marker categorization in the position recording apparatus and position recording method which concern on embodiment of this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Position recording device, 2 ... Camera (imaging means), 3 ... Control processing part, 4 ... Database (recording means).

Claims (8)

An imaging means for imaging a recognition object;
Image recognition means for performing image recognition of the recognition object based on image information imaged by the imaging means;
Position detection means for detecting the position of the recognition object recognized by the image recognition means;
Error calculating means for calculating an error of the position information detected by the position detecting means;
Recording means for recording the position information of the recognition object and its position detection error information;
A position recording apparatus comprising:   In the case where a plurality of pieces of position information are acquired for the same recognition object, the recording unit records the position information having the smallest position detection error as position information of the recognition object. The position recording apparatus according to claim 1.   When the recording means acquires a plurality of pieces of position information for the same recognition object, the recording means records the smallest position detection error among the position information as position information of the recognition object in the map data. The position recording apparatus according to claim 1, characterized in that:   The position recording apparatus according to claim 1, wherein the recognition object is a road sign. An image recognition step of performing image recognition of the recognition object based on image information imaged by the imaging means;
A position detection step of detecting the position of the recognition object image-recognized in the image recognition step;
An error calculating step of calculating an error of the position information detected in the position detecting step;
A recording step of recording position information of the recognition object and its position detection error information;
A position recording method comprising:   In the recording step, when a plurality of pieces of position information are acquired for the same recognition target object, the position information having the smallest position detection error is recorded as the position information of the recognition target object. The position recording method according to claim 5.   In the recording step, when a plurality of pieces of position information are acquired for the same recognition object, the smallest position detection error among the position information is recorded in the map data as the position information of the recognition object. The position recording method according to claim 5, wherein:   The position recording method according to claim 5, wherein the recognition object is a road sign.

Images