JP4775658B2 - Feature recognition device, vehicle position recognition device, navigation device, feature recognition method - Google Patents

Description

  The present invention acquires image feature acquisition means that sequentially captures at least image information obtained by imaging at least a road surface by an image pickup device mounted on a vehicle, and acquires road feature information related to features around the image information pickup position from map information. A road feature information acquisition unit, an image information recognition unit that performs recognition processing of the image information, and recognizes an image of a recognition object corresponding to the feature included in the image information, and the road feature information acquisition unit The feature recognition apparatus for recognizing the feature based on the acquired road feature information and the arrangement in the image information of the image of the recognition object recognized by the image information recognition means. The present invention relates to a feature recognition method used in this feature recognition device, and further to a vehicle position recognition device and a navigation device equipped with the feature recognition device.

  2. Description of the Related Art In recent years, for example, in a navigation apparatus or the like, a position specifying method using a radio wave signal from a GPS (Global Positioning System) is widely used to specify the position of a traveling vehicle. However, the accuracy of specifying the position of the vehicle by GPS includes an error of about several meters, and it is difficult to specify a detailed position with higher accuracy. In view of this, various techniques have been proposed so far in order to compensate for the poor accuracy of position determination by GPS.

  As such a technique, in Patent Document 1, a recognition unit recognizes a lane marking based on the brightness of an image or the like for image information captured by an imaging device mounted on the vehicle, whereby the vehicle is traveling. A technique for determining whether a road is a general road or a highway is disclosed.

  In this technique, for example, a portion in a window having a certain width in a captured image where the luminance exceeds a certain reference is recognized as a lane marking image. The lane lines recognized in this way are output to the determination unit as feature extraction data such as their lengths, the lengths of the broken portions between the lane lines, and the repetition pitches of the lane lines. Then, the determination unit can determine whether the traveling road is a general road or a highway based on the lane marking installation standards standardized for the general road and the highway. As a result, according to this technique, the speed of the vehicle can be controlled according to whether the road on which the vehicle is traveling is a general road or a highway.

  However, in the above technique, the position of the vehicle on the road that is being traveled, such as which lane the vehicle is traveling on when the vehicle is traveling on a road having a plurality of lanes in the traveling direction, etc. In addition, it is impossible to specify a detailed position such as a position in a running road.

Therefore, in Japanese Patent Application Laid-Open No. 2004-228867, the applicant executes image recognition based on feature information (referred to as road feature information) at the imaging position when executing image recognition of an image obtained from the imaging device. In addition, it is proposed that the recognition target object is recognized individually with high accuracy and the detailed position of the own vehicle is specified based on the relationship between the position information of the recognized feature and the position of the imaging device.
In this patent document 2, in specifying the position along the longitudinal direction of the road, features such as manholes, stop lines, signs, etc. that are not provided along the road are used. In specifying the position along the width direction of the road, a feature such as a lane marking (solid line, broken line) provided along the road is used. In this example, the feature is recognized based on the relationship between the road feature information and the position in the image information of the image of the recognition target object.

  For example, as shown in FIG. 10 of this specification, there are lane lines P1a, P2a, P2b, and P1b from the sidewalk I1 to the separation body I2, and between the lane lines. In the case where there is a travel direction traffic section P3, these lane lines and travel direction traffic sections are individually identified and recognized, and the vehicle position is specified from the positional relationship. Therefore, this position specifying method is based on the premise that each feature is recognized well. In other words, for example, when trying to specify the travel line in which the vehicle is traveling or the position in the road width direction, at least a solid line and a broken line that are division lines need to be accurately identified and recognized.

  Identification and recognition between the solid line and the broken line, which are lane markings, is relatively easy when a captured image as shown in FIG. 7 of Patent Document 2 is obtained, but as shown in FIG. 9A of the present specification. When a captured image is obtained, it may be difficult to recognize a broken line. In this figure, in recognizing the image GP2b of the lane marking, it may be difficult to identify whether the lane marking P2b is a broken line or a solid line because the broken portion is located on the back side of the image. In view of this, regarding a plurality of images (a plurality of frame images) sequentially captured by the imaging device, the same recognition target object is sequentially recognized, and a recognition target object (in this case, a dividing line in this case) recognized between the plurality of frames. ), A recognition method for recognizing the lane marking in which the fracture portion is recognized as a broken line and recognizing the lane marking in which the fracture portion is not recognized as a solid line is also adopted.

Japanese Utility Model Publication No. 5-23298 (page 6-8, Fig. 1-3) Japanese Patent Laying-Open No. 2006-208223 (FIG. 4)

  However, as described above, even when the configuration in which the solid line and the broken line are distinguished over a plurality of frames is adopted, the imaging is performed when the vehicle is stopped or when traveling at a very low speed. Since there is no substantial change in the image captured by the device, as a result, it will continue to image a part without a broken part, or it will continuously image a broken part, and the broken line is identified as a solid line. It has been found that there is a case in which there is no broken line even though the broken line exists.

  Also, when the vehicle stops for some reason, for example, on a travel direction traffic category that is a kind of paint display drawn on the road, the image recognition side recognizes that the vehicle has stopped on the lane line, and It has been found that there is a case where it is erroneously recognized that the so-called straddling is performed.

  The present invention has been made in view of the above problems, and an object of the present invention is to recognize a feature on the basis of imaging information of the feature and road feature information regarding the feature stored in the map database. An object of the recognition apparatus is to obtain a feature recognition apparatus that does not cause a problem of erroneously identifying a solid line and a broken line as lane markings.

To achieve the above purpose,
Image information acquisition means for sequentially capturing image information obtained by imaging at least a road surface by an imaging device mounted on a vehicle;
Road feature information acquisition means for acquiring road feature information related to features around the imaging position of the image information from map information;
Image information recognition means for performing recognition processing of the image information and recognizing an image of a recognition object corresponding to the feature included in the image information , and acquired by the road feature information acquisition means A first feature configuration of the feature recognition apparatus that recognizes the feature based on road feature information and an arrangement in the image information of the image of the recognition object recognized by the image information recognition unit is:
The image information recognizing means recognizes an image of the recognition object included in the single frame for a single frame when performing image recognition using the paint display provided on the road surface of the road as the recognition object. The first recognition process is performed, and the recognition target object recognized as a solid line or a broken lane line by the first recognition process is subjected to a plurality of consecutive frames and is substantially the same position between the plurality of frames. When the lane markings taken in the above are recognized as the same lane markings, and when the rupture portion is recognized as a broken line with respect to the lane markings recognized as the same, the lane marking is recognized as a broken line and the rupture portion is not recognized. It is configured to execute a second recognition process that recognizes as a solid line,
A velocity information acquiring means for acquiring the traveling speed information of the vehicle,
When the traveling speed information obtained is lower than a predetermined value, in that and a recognition correction control means for stopping the recognition of the recognition target object by the image information recognizing unit.

Even in this feature recognition device, image recognition for recognizing the image of the recognition target corresponding to the feature in the image information obtained from the imaging device is performed, and the recognition target recognized as a result of the image recognition The road feature information including information about the feature is compared and contrasted to recognize the feature, but the traveling speed of the vehicle on which the feature recognition device is mounted is separately acquired by the speed information acquisition means, When the traveling speed is smaller than the predetermined value by the recognition correction processing control means, the recognition processing by the image information recognition processing means is not performed.
That is, when the traveling speed is smaller than the predetermined value, the image capturing device continuously captures the same image (when stopped), or captures the same image continuously (when traveling at an extremely low speed). However, such image information is not subject to recognition processing. In other words, the image information that is subject to image recognition is an image that sequentially transitions as the vehicle travels. As a result, image information obtained by imaging the recognition target object from different imaging positions is obtained as image information, and image information in which features of the feature that is the recognition target object appear well is obtained, or between a plurality of frames. When identifying and recognizing from crossing image information, the feature of the recognition target can be satisfactorily recognized and the feature can be recognized accurately.

In the feature recognition apparatus having the first characteristic configuration, as described in claim 7 , the following image recognition method is used.
That is, an image information acquisition step for sequentially capturing image information obtained by imaging at least a road surface by an imaging device mounted on a vehicle;
Road feature information acquisition step for acquiring road feature information related to features around the imaging position of the image information from map information;
Recognizing the image information and recognizing an image of a recognition object corresponding to the feature included in the image information; and
A feature that recognizes the feature based on the road feature information acquired in the road feature information acquisition step and the arrangement in the image information of the image of the recognition object recognized in the image information recognition step. To run the recognition method,
In the image information recognition step, when performing the image recognition using the paint display provided on the road surface of the road as the recognition object, the image recognition of the recognition object included in the single frame is performed with respect to a single frame. After performing the first recognition process to be performed, with respect to the recognition target object recognized as a solid line or a broken lane line by the first recognition process, it is substantially the same across the plurality of frames for a plurality of consecutive frames. When a lane line photographed at a position is recognized as the same lane line, and when a broken portion is recognized with respect to the lane line recognized as the same, the lane line is recognized as a broken line and no broken portion is recognized Is configured to execute a second recognition process for recognizing a solid line,
Execute a speed information acquisition step of acquiring vehicle travel speed information;
When the acquired travel speed information is lower than a predetermined value, the recognition of the recognition object by the image information recognition step is stopped.
In this case, after the recognition of the recognition object is stopped, when the vehicle starts to travel and the traveling speed becomes higher than a predetermined value, the recognition is naturally restarted. In this way, features can be accurately recognized according to the running state of the vehicle.

Now, in the feature recognition device having the first characteristic configuration, as described in claim 2, the approximate position of the vehicle is obtained based on at least the positioning system position information obtained from the global positioning system. Comprising location information acquisition means,
It is preferable that the position of the vehicle obtained by the position information acquisition unit at the time when the image information is captured is set as the imaging position of the image information. This configuration is the second characteristic configuration of the present application.

  Although the position information from the global positioning system is coarse position information, it is sufficient information to identify features that may be included in the image information. Therefore, when specifying the position reference and the imaging position for acquiring the road feature information acquired by the road feature acquisition means, this information is used to obtain the image information of the recognition target object included in the image information and the road feature information. The imaged feature (recognition target) can be recognized quickly and accurately.

Now, vehicle position recognition apparatus having a feature recognition apparatus that has been described thus far, as described in claim 3, in addition to the feature recognizer,
Performing image recognition of the measurement points for the recognition object included in the image information for the measurement points predetermined for the recognition object,
Vehicle position recognition correction means for updating and correcting the vehicle position information based on the image recognition result of the recognized measurement point and the position information of the measurement point obtained in advance;
It can be set as the structure provided with.
This own vehicle position recognizing device is provided with own vehicle position recognition correcting means in addition to each means constituting the feature recognizing device.
On the other hand, measurement points are determined in advance for the recognition object (feature). For example, in the traveling method passage classification that is an arrow, the tip of the arrow is determined as a measurement point, and if it is a stop line, the outline on the front side in the traveling direction is determined as the measurement point. Then, from the image recognition result of the measurement point, the imaging position (substantially the own vehicle position) is determined from the measurement point, and the own vehicle position is obtained based on the position information of the measurement point obtained in advance. Update and correct information. In this way, the detailed vehicle position can be specified accurately.

As such a measurement point, as described in claim 4 , according to the contour shape of each feature, on the contour line substantially orthogonal to the traveling direction of the host vehicle, or when the contour shape has a corner portion Is preferably set on the corner.
By making the measurement point on the contour line approximately perpendicular to the traveling direction, it is easy to find the change in luminance in the image information generated on the contour line, and to easily identify the measurement point. By being orthogonal, the position of the vehicle along the longitudinal direction of the road can be easily identified.
When the contour shape has a corner, it is set on the corner, so that a change in luminance in the image information generated on the corner can be found well and the measurement point can be easily specified. . Furthermore, the position of the vehicle in the longitudinal direction and the width direction of the road can be easily identified.

The vehicle control apparatus provided with the own vehicle position recognizing device described so far includes, in addition to the own vehicle position recognizing device, a map information storing means for storing map information, as described in claim 5 .
A vehicle control device that performs travel control of the host vehicle based on the map information of the traveling direction of the host vehicle acquired from the map information storage unit and the host vehicle position information after the update correction by the host vehicle position recognition correction unit; can do.
This vehicle control device is provided with a map information storage means in addition to each means constituting the vehicle position recognition device. In the vehicle control, an accurate own vehicle position is acquired from the own vehicle position recognition device side, and the own vehicle is travel-controlled based on the map information obtained from the map information storage means. As a result, accurate and reliable vehicle control can be realized.

Navigation system having a vehicle position recognition apparatus described so far, as described in claim 6, in addition to the vehicle position recognition apparatus, comprising a map information storage means for storing map information,
Based on the map information of the traveling direction of the own vehicle acquired from the map information storage unit and the own vehicle position information after the update correction by the own vehicle position recognition correction unit, the navigation apparatus performs route guidance of the own vehicle. be able to.
This navigation device is provided with a map information storage means in addition to each means constituting the vehicle position recognition device. In navigation, an accurate vehicle position is acquired from the vehicle position recognition device side, and the vehicle is navigated based on map information obtained from the map information storage means. As a result, accurate and reliable navigation can be realized.

Embodiments of the present invention will be described below with reference to the drawings.
If it is in this application, self-vehicle position recognition device 100 concerning this application is explained about a first embodiment and a second embodiment. This own vehicle position recognizing device 100 is also a feature recognizing device referred to in the present application.
In the first embodiment, the own vehicle position recognition device 100 specifies the detailed position of the own vehicle in the road longitudinal direction and the road width direction on the road, and updates and corrects the own vehicle position to a highly accurate detailed position. It is comprised so that the driving | running | working impossible area around the own vehicle may also be recognized.
On the other hand, in the second embodiment, the own vehicle position recognition device 100 recognizes the travel lane (own lane) on the road being traveled, specifies the detailed position in the road width direction, and determines the own vehicle position (specifically Is configured to update and correct the own lane position) to a detailed position with high accuracy.
In the present application, in such specification of the detailed position and recognition of the untravelable area, recognition of the lane markings painted on the road (particularly, recognition recognition of solid lines and broken lines) should be performed accurately. Is important. Therefore, in the present application, the vehicle position recognition device 100 is constructed so as not to perform recognition processing and subsequent processing under specific conditions in which erroneous recognition between a solid line and a broken line is likely to occur.

  The vehicle position recognition device 100 according to the present application obtains the approximate position of the vehicle based on information from the GPS receiver 4 or the like when specifying and updating the vehicle position, and regarding the image information G captured at the same time, Recognize the imaged feature using the road feature information C, which is information about the feature that should be present at the approximate position, and specify the detailed position based on the recognized feature. The vehicle position is updated and corrected to the determined vehicle position. When executing such processing, as described above, if an error occurs in the recognition of the lane marking, the reliability of specifying the vehicle position is lowered. Therefore, in the vehicle position recognition apparatus 100 according to the present application, the vehicle recognizes the image recognition, the vehicle position identification, the update correction of the vehicle position associated with these processes, and the recognition of the non-travelable area. It is not executed when the vehicle is stopped or traveling at a very low speed. This execution stop is common in the first and second embodiments. Therefore, based on FIGS. 1 to 3, first, the control function part, which is a common configuration, will be described.

FIG. 1 is a functional block diagram for explaining the outline of the hardware configuration of the common control function part. As shown in FIG. 1, the host vehicle position recognition apparatus 100 includes a travel speed deriving unit 101, a recognition correction processing control unit 102, and a host vehicle position recognition correction unit 103. Based on the derived traveling speed information V, the recognition correction processing control unit 102 is configured to control whether or not the processing in the vehicle position recognition correction unit 103 can be executed.
A detailed configuration of the vehicle position recognition correction unit 103 is shown in FIG. 4 for the first embodiment and FIG. 15 for the second embodiment. Details of the vehicle position recognition correction unit 103 will be described later based on these drawings and flowcharts of the parts.

  As illustrated in FIG. 2, the traveling speed deriving unit 101 derives the traveling speed using rotation information from a rotation sensor 101 a that detects the rotation of a drive transmission shaft or the like connected to the traveling wheel W. In the traveling speed deriving unit 101, a traveling stop state where the vehicle is stopped and a low traveling state where the traveling speed is traveling at an extremely low speed such as less than 1 km / hr are also detected. In this case, the rotation sensor 101a and the traveling speed deriving unit 101 correspond to the “speed information acquisition unit” in the present invention.

The recognition correction processing control unit 102 has a configuration in which a functional unit for performing various processes on input data is implemented in hardware and / or software using an arithmetic processing unit such as a CPU as a core member. In addition, based on the travel speed information V derived by the travel speed deriving unit 101, a determination is made as to whether or not to execute the process in the vehicle position recognition correction unit 103. FIG. 3 shows a flowchart in the recognition correction processing control unit 102.
When the vehicle position recognition device 100 is in operation, the traveling speed deriving unit 101 sequentially receives the rotation information from the rotation sensor 101a and derives the traveling speed information V. The derived traveling speed information V is acquired by the recognition correction processing control unit 102 (step # 1).
In the recognition correction processing control unit 102, the acquired traveling speed information V is compared with preset traveling reference information (step # 2). This traveling reference information is traveling speed information for determining whether the host vehicle is traveling or stopped, and is set to 1 km / hr, for example.

In the recognition correction processing control unit 102, when the traveling speed information V is larger than the traveling reference information (step # 2: yes), the processing in the own vehicle position recognition correction unit 103 is permitted. On the other hand, when the traveling speed information V is smaller than the traveling reference information (step # 2: no), the host vehicle is stopped or traveling at an extremely low speed, so that the processing by the host vehicle position recognition correction unit 103 is not performed. , Repeated acquisition of travel speed information. That is, the processing in the vehicle position recognition correction unit 103 is not performed.
As described above, in the subject vehicle position recognition device 100 of the present application, the subject vehicle position recognition correction process is not executed when the subject vehicle is stopped or traveling at an extremely low speed. It is configured. Therefore, the recognition correction processing control unit 102 corresponds to “recognition correction processing control means” in the present invention. Further, the vehicle position recognition correction unit 103 corresponds to the “vehicle position recognition correction unit” in the present invention.

  Hereinafter, with respect to each of the first and second embodiments, the configuration and operation of the vehicle position recognition correction unit 103 will be described in order.

[First embodiment]
As shown in FIG. 4, the vehicle position recognition correction unit 103 in this embodiment is based on both the recognition result of the image information G captured by the imaging device 2 and the road feature information C acquired from the map information. The detailed position in the road 11 on which the vehicle M is traveling, that is, the detailed position in the longitudinal direction and the width direction of the own vehicle is specified, and the own vehicle position recognized by the apparatus 100 is specified. Update correction to position. Furthermore, the travel impossible area I is also recognized.

  As shown in FIG. 4, the vehicle position recognition correction unit 103 mainly includes an image information acquisition unit 3 that captures image information G from the imaging device 2 mounted on the vehicle M (see FIG. 2), and a GPS. (Global positioning system) An approximate position specifying calculation unit 7 that performs an operation for specifying an approximate image pickup position by the image pickup device 2 based on outputs from the receiver 4, the direction sensor 5, and the distance sensor 6, and a map information database 8, a road feature information acquisition calculation unit 9 that performs a calculation for acquiring road feature information C related to features around the approximate imaging position by the imaging device 2 from the map information stored in FIG. 8, and the acquired road feature information C The image information G is recognized using the image information G, and the image information recognition calculation unit 10 that performs calculation for recognizing the image of the recognition target object corresponding to the feature included in the image information G, road feature information Based on the road feature information C acquired by the acquisition calculation unit 9 and the arrangement in the image information G of the image of the recognition object recognized by the image information recognition calculation unit 10, the inside of the road 11 on which the vehicle M is traveling A vehicle position specifying calculation unit 17 that specifies a detailed position of the vehicle and corrects the recognized position of the vehicle.

  Here, the approximate position specifying calculation unit 7, the GPS receiver 4, the direction sensor 5, the distance sensor 6, and the map information database 8 are mounted on a vehicle and use the configuration of a navigation device for performing route guidance of the vehicle. can do. In this case, the approximate position specifying calculation unit 7, the GPS receiver 4, the azimuth sensor 5, the distance sensor 6 and the like correspond to the “position information acquisition unit” in the present invention.

  The imaging device 2 includes an imaging element such as a CCD sensor or a CMOS sensor, and a lens that constitutes an optical system for guiding light to the imaging element. The imaging device 2 is arranged in front of or behind the vehicle M, for example, at positions indicated by Q1 to Q3 in FIG. 2, and at least a road surface of the road 11 on which the vehicle M travels is photographed. It is provided so that the surroundings can also be photographed. As such an imaging device 2, an in-vehicle camera or the like that has been provided so far in order to capture images of the front and rear of the vehicle M is preferably used.

  The image information acquisition unit 3 includes an interface circuit 12 for connecting to the imaging device 2, an image preprocessing circuit 13 that performs preprocessing on the image information G from the imaging device 2, and preprocessed image information G. And an image memory 14 for storage. The interface circuit 12 includes an analog / digital converter or the like, takes in analog image information G imaged by the imaging device 2 at a predetermined time interval, converts it into a digital signal, and outputs it to the image preprocessing circuit 13. The time interval for capturing the image information G by the interface circuit 12 can be set to about 10 to 50 ms, for example. Thereby, the image information acquisition part 3 can acquire the image information of the road 11 in which the vehicle M is running substantially continuously. Here, the image preprocessing circuit 13 performs processing for facilitating image recognition by the image information recognition calculation unit 10 such as binarization processing and edge detection processing as preprocessing for the image information G. Then, the image information G after such preprocessing is stored in the image memory 14.

In addition to the image information G sent to the image preprocessing circuit 13, the interface circuit 12 outputs the image information G directly to the image memory 14. Therefore, the image memory 14 stores both the image information G2 after the preprocessing by the image preprocessing circuit 13 and the image information G1 as it is not subjected to the preprocessing.
In the present embodiment, the image information acquisition unit 3 constitutes “image information acquisition means” in the present invention.

  In this embodiment, the approximate position specifying calculation unit 7 is connected to the GPS receiver 4, the direction sensor 5, and the distance sensor 6. Here, the GPS receiver 4 is a device that receives a signal from a GPS satellite, and can obtain various information such as a position (latitude and longitude) and a moving speed of the GPS receiver 4. The azimuth sensor 5 includes a geomagnetic sensor, a gyro sensor, an optical rotation sensor attached to the rotating part of the steering wheel, a rotary resistance volume, an angle sensor attached to the wheel part, etc., and detects the traveling direction of the vehicle M. can do. The distance sensor 6 includes a combination of a vehicle speed sensor that detects the number of rotations of the wheel, a yaw / G sensor that detects the acceleration of the vehicle M, and a circuit that integrates the detected acceleration twice. The distance can be detected.

  Then, the approximate position specifying calculation unit 7 performs an operation for obtaining the current approximate position of the vehicle M based on outputs from the GPS receiver 4, the direction sensor 5, and the distance sensor 6. The position of the vehicle M calculated in this way becomes the position of the imaging device 2.

The accuracy of the approximate position of the vehicle that can be obtained by the approximate position specifying calculation unit 7 is greatly influenced by the accuracy of the GPS receiver. For this reason, at present, an error of about several meters is included. Therefore, the approximate position specifying calculation unit 7 cannot specify the detailed position of the vehicle M in the road longitudinal direction and the road width direction.
As a result, as shown in the present application, the detailed position of the vehicle M is identified using the information stored in the map information database 8 regarding the image information G and the feature, and the vehicle position recognition is performed based on the identification result. It is necessary to update and correct the own vehicle position recognized by the apparatus 100.

The approximate position specifying calculation unit 7 is also connected to the interface circuit 12 of the image information acquisition unit 3. The interface circuit 12 outputs a signal to the approximate position specifying calculation unit 7 at the timing of imaging by the imaging device 2. Therefore, the approximate position specifying calculation unit 7 can specify the approximate image pickup position of the image information G by calculating the position of the image pickup device 2 at the timing when the signal input from the interface circuit 12 is received. . Thus, the approximate imaging position of the image information G obtained by the approximate position specifying calculation unit 7 is expressed by latitude and longitude information, and is output to the road feature information acquisition calculation unit 9.
The approximate position specifying calculation unit 7 is configured such that a function unit for performing various processes on input data is implemented by hardware and / or software using an arithmetic processing unit such as a CPU as a core member. It has.

The road feature information acquisition calculation unit 9 is connected to the approximate position specification calculation unit 7 and the map information database 8.
FIG. 5 is an explanatory diagram showing the contents of the map information stored in the map information database 8. As shown in this figure, the map information database 8 according to this embodiment stores a road network layer L1, a road shape layer L2, and a feature layer L3 as map information.
The road network layer L1 is a layer indicating connection information between the roads 11. Specifically, it has information on a large number of nodes N having position information on a map expressed by latitude and longitude, and information on a large number of links L that constitute the road 11 by connecting the two nodes N. Configured. Each link L includes information such as the type of road 11 (type of highway, toll road, national road, prefectural road, etc.) and link length as link information.
The road shape layer L2 is stored in association with the road network layer L1 and indicates the shape of the road 11. Specifically, information on a number of road shape complementary points S that are located between two nodes N (on the link L) and have position information on the map expressed in latitude and longitude, and each road shape complementary point Information of the road width W in S.

  The feature layer L3 is a layer that is stored in association with the road network layer L1 and the road shape layer L2, and indicates information on various features provided on the road 11 and in the vicinity of the road 11. As the feature information stored in the feature layer L3, at least information on the position, shape, color, and the like related to the feature that can be a recognition target in the vehicle position recognition correction unit 103 is stored. Specifically, in the present embodiment, the paint display P provided on the road surface of the road 11, the non-travelable area I adjacent to the road 11, various signs 15 and traffic lights 16 provided on the road 11, etc. The various features are configured to have the location information on the map with reference to the road shape complementary point S or the node N, and the feature information such as the shape and color.

  Here, the paint display P includes, for example, lane markings that divide the lane (including information on the types of lane markings such as solid lines, broken lines, and double lines), zebra zones, and traveling directions that specify the traveling direction of each lane. Includes separate traffic classification display, stop line, pedestrian crossing, pedestrian crossing notice (with pedestrian crossing), turn prohibition, speed display, etc. The travel impossible area I includes, for example, a shoulder, a sidewalk, a separation zone, etc. adjacent to the road 11.

  The map information database 8 is an apparatus having a recording medium capable of storing information and its driving means, such as a hard disk drive, a DVD drive with a DVD-ROM, a CD drive with a CD-ROM, and the like. Is provided as a hardware configuration.

  Then, the road feature information acquisition calculation unit 9 calculates from the map information stored in the map information database 8 based on the latitude and longitude information indicating the imaging position of the image information G specified by the approximate position specification calculation unit 7. Then, a calculation is performed to acquire road feature information C relating to features around the imaging position of the image information G. Here, the road feature information acquisition calculation unit 9 obtains feature information such as the position, shape, color, and the like of features included in an area including at least the imaging area of the imaging device 2 around the imaging position of the image information G. A calculation process for extracting the road feature information C from the feature layer L3 of the map information database 8 is performed.

This road feature information acquisition calculation unit 9 has a functional unit for performing various processes on input data as hardware or software, or both, with an arithmetic processing unit such as a CPU as a core member. It has a configuration.
In the present embodiment, the road feature information acquisition calculation unit 9 constitutes “road feature information acquisition means” in the present invention.

The image information recognition calculation unit 10 performs a recognition process of the image information G and performs a calculation for recognizing an image of a recognition target included in the image information G. In the present embodiment, the image information recognition calculation unit 10 is connected to the image memory 14 and the road feature information acquisition calculation unit 9 of the image information acquisition unit 3, and the road feature acquired by the road feature information acquisition calculation unit 9. Recognition processing of the image information G is performed using the information C.
In addition, a recognition object for which image recognition is performed in the image information recognition calculation unit 10 is stored as feature information in the above-described paint display P, the travel impossible region I, and the feature layer L3 such as various signs 15 and traffic lights 16. Corresponding to the feature being made.

The image information recognition calculation unit 10 has a functional unit for performing various processes on input data as hardware or software, or both, with an arithmetic processing unit such as a CPU as a core member. It has.
In the present embodiment, the image information recognition calculation unit 10 constitutes “image information recognition means” in the present invention.

As a specific method for recognizing the image information G using the road feature information C in the image information recognition calculation unit 10, in the vehicle position recognition device 1, when one frame included in the image information G is targeted, These two methods are used alone or in combination.
One image recognition processing method extracts an image candidate of a recognition object from the image information G, compares it with the road feature information C, and recognizes an image candidate having high consistency with the road feature information C as an image of the recognition object. It is a processing method to do.
In another image recognition processing method, an area where an image of a recognition object exists in the image information G is estimated based on the road feature information C, and it is estimated that an image of the recognition object exists based on the estimation result. In this processing method, the recognition algorithm is adjusted so that the criterion for determining whether or not the object is a recognition object is lower than in other areas, and the image of the recognition object is recognized from the image information G.
These two image recognition processing methods are both image recognition processing methods applicable to image information in a single frame, and correspond to the “first recognition processing” of the present invention.

Further, regarding the recognition and recognition between the solid line and the broken line, which are lane markings, the same image is recognized between a plurality of consecutive frames captured in a unit time with respect to the image of the recognition object recognized as a solid line or a broken line by the above method. In the recognition process, if a broken part is recognized in the recognition process, the recognition line is recognized as a broken line. If the part is not recognized, it is recognized as a solid line.
This image recognition processing method targets image information obtained over a plurality of frames, and corresponds to the “second recognition processing” of the present invention.

  In the present embodiment, as will be described later with reference to a flowchart, the image information recognition calculation unit 10 combines these recognition processing methods, and paint display P provided on the road surface of the road 11 and adjacent to the road 11. The process which recognizes the driving | running | working impossible area I to perform is performed. Therefore, the image information recognition calculation unit 10 includes a paint display recognition calculation unit 10a, a road feature information matching unit 10b, an area estimation unit 10c, and an untravelable area recognition unit 10d.

The vehicle position specifying calculation unit 17 is based on the road feature information C acquired by the road feature information acquisition calculation unit 9 and the arrangement in the image information G of the image of the recognition target recognized by the image information recognition calculation unit 10. Thus, a calculation for specifying a detailed position in the road 11 on which the vehicle M is traveling is performed. For this reason, in the present embodiment, a calculation process for specifying the detailed positions of both the road longitudinal direction of the vehicle M and the road width direction of the vehicle M is performed.
In the present embodiment, as will be described later with reference to a flowchart, the vehicle position specifying calculation unit 17 includes the image information G in the image of one or more recognition objects recognized by the image information recognition calculation unit 10. By comparing the arrangement and the position information of the object corresponding to one or more recognition objects included in the road feature information C, the imaging position of the image information G is specified in detail, and the road length of the vehicle M A calculation process for specifying the direction and the detailed position of the vehicle M in the road width direction is performed. Then, the detailed position of the vehicle M obtained by such calculation processing is updated and corrected as a new position of the own vehicle. Therefore, the vehicle position specification calculation unit 17 includes an arrangement information extraction unit 17a, a comparison calculation unit 17b, an imaging position specification unit 17c, and a host vehicle position recognition correction unit 17d.

  The vehicle position specifying calculation unit 17 has a configuration in which a function unit for performing various processes on input data is implemented by hardware and / or software using an arithmetic processing unit such as a CPU as a core member. It has.

  Next, based on the recognition result of the image information imaged by the imaging device 2 and the road feature information C acquired from the map information, the specification of the non-travelable area I and the details of the road 11 on which the vehicle M is traveling are detailed. A specific example of the process of specifying the position and updating and correcting the vehicle position will be described in detail with reference to the flowcharts shown in FIGS.

  As shown in FIG. 6, the own vehicle position recognition correction unit 103 first performs a process of capturing image information G captured by the imaging device 2 (step # 61). Specifically, the image information G captured by the imaging device 2 including an in-vehicle camera is transmitted to the image preprocessing circuit 13 and the image memory 14 via the interface circuit 12. At this time, the interface circuit 12 outputs a signal to the approximate position specifying arithmetic unit 7 at the timing of capturing the image information G from the imaging device 2, that is, at almost the same timing as the timing of imaging by the imaging device 2. Do. This signal is a signal for informing the approximate position specifying calculation unit 7 of the imaging timing.

  In response to the input of the image information G, the image preprocessing circuit 13 performs preprocessing on the image information G (step # 62). As preprocessing performed here, for example, various processes for facilitating image recognition by the image information recognition calculation unit 10 such as binarization processing and edge detection processing are performed. FIG. 9A is an example of the image information G (G1) imaged by the imaging device 2, and FIG. 9B is a preprocess for the image information G shown in FIG. It is an example of back image information G (G2). In the example shown in FIG. 9B, an image showing the outline of an object captured as image information G by the edge detection process is extracted. Then, the image information G2 after the preprocessing in step # 62 and the image information G1 sent directly from the interface circuit 12 are stored in the image memory 14 (step # 63).

  In parallel with the processes of steps # 62 and # 63, the approximate position specifying calculation unit 7 performs a process for obtaining an approximate imaging position of the image information G (step # 64). Specifically, when a signal indicating the timing of capturing the image information G is output from the interface circuit 12, it is used as the timing of imaging by the imaging device 2 based on the GPS receiver 4, the azimuth sensor 5, and the distance sensor 6. To calculate the approximate current position of the vehicle M. Then, the approximate current position of the vehicle M obtained in this way is set as the approximate imaging position of the image information G. The information of the approximate imaging position obtained here is sent to the road feature information acquisition calculation unit 9 as latitude and longitude information.

  Next, the road feature information acquisition calculation unit 9 performs a process of acquiring the road feature information C related to the features around the imaging position of the image information G from the map information stored in the map information database 8 (step # 65). . At this time, the road feature information acquisition calculation unit 9 selects a road feature within a certain range R around the approximate imaging position obtained in step # 64 from a wide range of map information stored in the map information database 8. Information C is extracted and acquired. Here, it is preferable that the fixed range R around the approximate imaging position is set so as to include at least a region imaged as the image information G by the imaging device 2.

  FIG. 10 shows an example of the road feature information C acquired by the road feature information acquisition calculation unit 9 in a graphic form. In this example, the features included in the road feature information C include, as the paint display P, between the two lane lines P1a and P1b indicating the outer edge of the roadway of the three-lane road 11 and between the three lanes. There are two broken lane markings P2a and P2b and a traveling direction passage segment P3 provided in three lanes. Further, as the non-travelable area I, there are a sidewalk I1 adjacent to the right side of the road 11 and a central separation zone I2 adjacent to the left side of the road 11. FIG. 10 is merely an example, and various features are included in the road feature information C depending on the imaging position of the image information G.

The content of the road feature information C is composed of position information, shape information, and color information of these various features. Here, the position information of each feature is represented by position information based on a road shape complementing point S (including a node N (see FIG. 5) where a node N such as an intersection is present). That is, for example, it is provided along roads 11 such as solid lane markings P1a and P1b and broken lane markings P2a and P2b of the paint display P, or a sidewalk I1 and a median I2 in the non-running region I. The existing feature is represented only by the distance (offset amount) from the road shape complement point S (or node N). On the other hand, for example, with respect to features such as a travel direction passage section P3 that is not provided along the road 11 and features such as stop lines and signs, the positional information is based on the distance and direction from the specific road shape complement point S (or node N). Is represented.
The shape information of each feature includes information such as what size and shape (silhouette) each has in the vertical, horizontal, and height directions based on the position obtained from the position information. Have. It is preferable that the shape information is simplified and expressed so as to be easily compared with the image information G.
The color information of each feature is preferably information in which color information is stored for each region in the shape information in the case of a non-uniform color feature such as a road sign.

Next, the image information recognition calculation unit 10 performs a process of recognizing the image information G, and performs a process of recognizing the image of the recognition object included in the image information G (step # 66). In the present embodiment, among the images of the features included in the image information G, the paint display P and the untravelable area I (particularly the sidewalk I1) are used as recognition objects. Then, as the image recognition processing of the recognition target object, after the image recognition of the paint display P that is relatively easy to recognize, the recognition algorithm is adjusted based on the recognition, and it is difficult to travel more difficult than the paint display P. Processing for performing image recognition of the region I is performed. A specific example of such image information G recognition processing is shown in the flowchart of FIG.
Note that the image recognition of the untravelable area I is more difficult than the image recognition of the paint display P. The paint display P has a large difference in luminance and color from the road surface of the road 11, and the image recognition is relatively easy. On the other hand, in the non-travelable region I such as a shoulder, a sidewalk, and a separation zone, it is difficult to specify an outline even by edge detection or the like because a difference in luminance and color between the road 11 and its surroundings is small. Because there are many.

  In the recognition processing of the image information G, as shown in FIG. 7, first, in the paint display recognition calculation unit 10 a of the image information recognition calculation unit 10, an image having the possibility of the paint display P included in the image information G. Processing for extracting candidates is performed (step # 71). Specifically, as shown in FIG. 9B, a template or the like that represents the characteristics of the paint display P such as lane markings is specified in advance from the image information G2 after the preprocessing such as the edge detection processing. An image that is highly consistent with the feature data that has been extracted is extracted and used as an image candidate for the paint display P. In the example shown in FIG. 10, FIG. 9 (b), the right solid line image GP1a, the right broken line image GP2a, the left broken line image GP2b, the sidewalk curb on the outside The image GI1a, the solid line image GP1b on the left side, and the traveling direction image GP3 are extracted as image candidates for the paint display P.

Thereafter, in the road feature information matching unit 10b of the image information recognition calculation unit 10, the image candidates of the paint display P extracted in step # 71 and the paint display P in the road feature information C acquired in step # 65. The information is compared (step # 72). Then, as a result of this comparison, image candidates having high consistency with respect to the positional relationship, shape, and information such as color and brightness are extracted, and the extracted image candidates are recognized as images of the paint display P (step # 73). ). Based on the road feature information C related to the paint display P in FIG. The positional relationship with the section P3 and the shape, color, brightness, and the like of the partition lines P1a, P1b, P2a, P2b and the traveling direction passage section P3 are known. Therefore, by extracting the image candidates of the paint display P in the image information G having high consistency with the road feature information C, it is possible to extract only the image candidates that are highly likely to be the paint display P. In the case of the example shown in FIG. 10, the image GI1a of the curb on the sidewalk outside the image GP1a of the left solid line is excluded by the process of step # 73. The extracted image candidate is recognized as an image of the paint display P. Note that information such as the color and brightness of the image candidate of the paint display P can be acquired from the image information G stored in the image memory 14 and not subjected to preprocessing.
FIG. 11A shows only the image of the paint display P extracted from the image information G by the process of step # 73.

Next, the image information G and the road feature information C are collated with reference to the recognized image of the paint display P (step # 74). In other words, by associating the position of the recognized image of the paint display P in the image information G with the position of the paint display P included in the road feature information C so as to match, the various features included in the road feature information C Can be associated with the image included in the image information G. In this association, when identifying the solid lines GP1a and GP1b and the broken lines GP2a and GP2b, the images of a plurality of frames of the image information G acquired sequentially are recognized to be the same over a plurality of consecutive frames. When individual recognition objects GP1a, GP1b, GP2a, GP2b are sequentially extracted and tracked, and in the recognition process, a broken portion is recognized with respect to the partition lines GP1a, GP1b, GP2a, GP2b that are recognized to be the same between consecutive frames. The partition lines are recognized as broken lines GP2a and GP2b, and are recognized as solid lines GP1a and GP1b when a broken portion is not recognized.
This identification method will be described with reference to FIGS. FIG. 9A shows the image information G that is currently recognized as described above, and FIG. 13 shows the image information G several frames before as shown in FIG. 9A. . There is no change in the driving lane between the two. Comparing the two figures, the solid lines GP1a and GP1b are recognized as white lines continuing at substantially the same position, whereas the broken lines GP2a and GP2b are substantially the same by changing the positions of the broken portions GS. The white line in a broken state is photographed at the position. Therefore, when the vehicle is in a running state, when the luminance in the image crossing direction is tracked for each lane line over a plurality of frames at a specific height position of the image, no broken portion appears in the solid line, and the broken line appears. The rupture part GS appears sequentially. Thus, the solid lines GP1a and GP1b and the broken lines GP2a and GP2b can be more accurately identified by such identification processing.

  The positional relationship in the width direction of the road 11 can be accurately associated by using the features such as the lane markings GP1a and GP2a provided along the road 11, and the road 11 is not provided along the road 11. The positional relationship in the longitudinal direction along the road 11 can be accurately associated by using the traveling direction passage section P3 and features such as stop lines and signs (not shown) as references.

  Thereafter, in the region estimation unit 10c of the image information recognition calculation unit 10, there is an image of the untravelable region I in the image information G based on the result of collation between the road feature information C and the image information G in step # 74. A process of estimating the area is performed (step # 75). That is, based on the collation result between the image information G and the road feature information C in the above step # 74, it is possible to guess the arrangement of the images of the various features including the paint display P and the untravelable area I in the image information G. It is. Therefore, a calculation is performed to estimate the region in the image information G corresponding to the position and shape of the non-travelable region I included in the road feature information C from the collation result in step # 74. Then, the area calculated as a result of the process of step # 75 is estimated as the area where the image of the untravelable area I exists.

  In the present embodiment, as shown in FIG. 11 (b), the image range captured as the image information G is divided into lane markings P1a, P1b, P2a in the paint display P recognized by the process of step # 73. Based on P2b, the area is simply divided into areas A1 to A4 to which each of the division lines P1a, P1b, P2a, and P2b belongs and areas A5 to A8 divided by these areas A1 to A4. Then, by determining whether each of the areas A5 to A8 includes the travel impossible area I based on the collation result in step # 74, a process for estimating the area where the image of the travel impossible area I exists. I do. Here, as shown in FIG. 11B, based on the road feature information C, it is assumed that there is a non-travelable area I (sidewalk I1) outside the solid line P1a on both sides of the road 11. Since it can be determined, it can be inferred that the image of the untravelable area I exists in the area A5 outside the area A1 to which the solid line P1a of the road 11 belongs (step # 75).

Next, based on the estimation result of Step # 75, the recognition algorithm in the non-running region recognition unit 10d of the image information recognition calculation unit 10 is adjusted (Step # 76), and the non-running included in the image information G is not possible. The image in the area I is recognized (step # 77).
In the present embodiment, in the area A5 in which it is estimated that the image of the untravelable area I exists in step # 75, the criterion for determining whether or not the non-travelable area I is another area (here, the area A6). The recognition algorithm is adjusted to be lower than ~ A8). That is, as described above, it is difficult to specify the contour of the non-travelable area I such as the sidewalk I1 and the road shoulder by edge detection or the like because the brightness and color difference between the road 11 and the surrounding area is small. In many cases, image recognition is generally more difficult than paint display P. Therefore, by adjusting the recognition algorithm for the area A5 in which the image of the untravelable area I is estimated to exist so that it can be recognized as the unrunnable area I more than the other areas, The recognition rate can be increased.
In order to adjust the recognition algorithm so that the criterion for determining whether or not the vehicle is in the non-travelable region I is lower than that in the other regions, the region A5 in which the image of the non-driveable region I is estimated to exist exists. In addition to a method for lowering the criterion for other regions, a method for increasing the criterion for other regions relative to the region A5, a method for lowering the criterion for the region A5 relative to other regions, and other For example, there is a method of raising the determination criterion of the region with respect to the region A5. The specific adjustment method of this recognition algorithm is a method according to the recognition method of the travel impossible area I.

  For example, in the present embodiment, as the recognition algorithm for the image of the untravelable area I, edge detection processing is performed on the image information G, the number of edge points at each position in the width direction of the road 11 is detected, and the number of edge points is detected. This is an algorithm for recognizing that a place where is equal to or greater than a predetermined threshold is the travel impossible region I. As the threshold value at this time, as shown in FIG. 12, a first threshold value t1 set to a low value and a second threshold value t2 set to a high value are used. That is, the first threshold value t1 is used in the area A5 where it is estimated that the image of the untravelable area I exists, and the second threshold value t2 is used in the other areas A6 to A8. The recognition algorithm is adjusted so that the criterion for determining whether or not it is the non-travelable region I is lower than the other regions A6 to A8 in the region A5 where it is estimated that the image of the possible region I exists.

FIG. 12 is a graph showing the result of detecting the number of edge points at each position in the width direction of the road 11 for the image information G shown in FIG. As shown in this figure, the areas A1 to A4 have a large number of edge points because there are lane markings P1a, P1b, P2a, and P2b, but these areas A1 to A4 are image recognition of the untravelable area I. It is not subject to. Since the areas A6, A7, and A8 are only the asphalt road surface except the position where the traveling direction traveling section P3 exists, the number of edge points is reduced as a whole.
On the other hand, the area A5 has a certain number of edge points. In conclusion, the area A5 has a large number of edge points due to the presence of the non-travelable area I such as the sidewalk I1. However, it is difficult to determine whether or not it is the non-running region I from only the number of edge points.

  Therefore, based on the estimation result of step # 75, the threshold value for determining the non-travelable region I is set to a low value in the region A5 where the image of the non-travelable region I is estimated to exist. In the other areas A6 to A8, the threshold value for determining that the vehicle is in the non-travelable area I is set to a second threshold value t2 set to a high value. Thereby, based on the estimation result of step # 75, the detection omission of the non-runnable area I is reduced in the area A5 where the image of the non-runnable area I is estimated to exist, and in the other areas A6 to A8 Thus, erroneous detection of the travel impossible region I can be prevented. Therefore, the recognition rate of the travel impossible area I can be increased. The values of the first threshold value t1 and the second threshold value t2 may be determined appropriately experimentally and statistically. Further, the values of the first threshold value t1 and the second threshold value t2 change based on other information extracted from the image information G, signals from other sensors mounted on the vehicle M, and the like. A variable value is also one preferred embodiment.

  As described above, the image information recognition calculation unit 10 performs the recognition processing of the image information G, whereby the paint display P as the recognition target object included in the image information G and the image of the travel impossible region I are recognized. The In the example of the image information G shown in FIG. 11, as shown in FIG. 11 (b), the images GP1a, GP1b, GP2a, GP2b of the lane markings P1a, P1b, P2a, P2b, the image Gp3 of the traveling direction passage section P3, the section The image GI1 of the sidewalk I1 on the right side of the image GP1a of the line P1a is recognized. In this recognition processing, the ability to accurately recognize solid lines and broken lines, which are lane markings, greatly contributes to improving the reliability of the feature recognition apparatus 100 according to the present application. Is low, the configuration in which the image recognition process is not performed is adopted, so that the reliability of the apparatus is greatly improved.

  Next, as shown in FIG. 6, the vehicle position specifying calculation unit 17 arranges the road feature information C acquired at step # 65 and the image of the recognition target image recognized at step # 66 in the image information G. Based on the above, a detailed position in the road 11 on which the vehicle M is traveling is specified and a process for updating and correcting the own vehicle position is performed (step # 67). In the present embodiment, the arrangement in the image information G of the image of the recognition object recognized in step # 66 and the object corresponding to the recognition object included in the road feature information C acquired in step # 65. By comparing the position information with the imaging position of the image information G in detail, a calculation process is performed to specify the detailed position of the vehicle M in the road longitudinal direction and the vehicle M in the road width direction.

  A specific example of the process for specifying the detailed position of the vehicle M is shown in the flowchart of FIG. In this process, first, the arrangement information extraction unit 17a of the vehicle position specifying calculation unit 17 extracts arrangement information in the image information G of each recognition object recognized in step # 66 (step # 81). Here, the arrangement information in the image information G of the recognition target includes information on the position of each recognition target in the image information G and information on the shape and color of each recognition target corresponding thereto. . In the example of the image information G shown in FIG. 9, as shown in FIGS. 9A and 14, as recognition objects, the images GP1a, GP1b, GP2a, GP2b of the lane markings P1a, P1b, P2a, P2b, the traveling direction Since the image GP3 of the traffic section P3 and the image GI1 of the sidewalk I1 are recognized, the arrangement information in the image information G for these recognition objects is extracted in step # 81.

  Next, in the comparison calculation unit 17b of the vehicle position specifying calculation unit 17, the arrangement information in the image information G of each recognition object extracted in step # 81 and the road feature information C acquired in step # 65. Comparison processing is performed (step # 82). Here, the arrangement information in the image information G of each recognition object is compared with the arrangement of each feature included in the road feature information C, and the image of the recognition object in the image information G with high consistency and the road A process for associating the feature information in the feature information C with each other is performed. Thereby, the image of each recognition object in the image information G is an image of which feature in the road feature information C and at which position on the road 11 represented by the road feature information C. Can be specified.

  Thereafter, the imaging position specifying unit 17c of the vehicle position specifying calculation unit 17 performs processing for specifying the imaging position of the image information G in detail (step # 83). FIG. 14 is a diagram schematically showing the procedure of this processing. That is, the process of specifying the imaging position in detail is based on the comparison result of step # 82, and the imaging position of the image information G that matches the layout of the image of each recognition object in the image information G is determined based on the road feature information. Corresponding to the position in C, it is performed by specifying in detail in the road longitudinal direction and road width direction of the road 11.

Therefore, a specific description will be given of a calculation processing method for specifying the imaging position of the image information G based on the recognition result of the image information G.
First, when specifying the imaging position in the longitudinal direction of the road, unlike lane markings and sidewalks, images of recognition objects such as travel direction traffic sections, stop lines, signs, traffic lights, etc. that are not provided along the road 11 are used. By using as a reference, it is possible to specify in detail the imaging position in the direction along the road 11, that is, in the longitudinal direction of the road. That is, when the arrangement of the images of the recognition objects that are not provided along the road 11 in the image information G is analyzed, it can be seen that there is an image GP3 of the traveling direction passage section in the example shown in FIG. Here, since the imaging device 2 is fixed to the vehicle M so as to be in a predetermined imaging direction at a predetermined height, the arrangement in the image information G of the image GP3 of the traveling direction passage section, particularly in the height direction Based on the arrangement, the distance D from the imaging position to the travel direction passage section P3 can be calculated. At this time, in the example shown in FIG. 14, the distance D is calculated by paying attention to the measurement point FC set at the tip of the traveling direction passage section P3, which is an arrow. The absolute position of the attention point FC is stored in the map information database 8 as latitude / longitude information. Thereby, the imaging position of the image information G can be specified in detail in the road longitudinal direction. In the example shown in FIG. 14, the imaging position of this image information G is specified as the B1 position.

  When specifying the position in the road width direction, when analyzing the arrangement of the images of each recognition object in the image information G, there is a broken line GP2a on the left side from the center of the image information G, and a solid line on the right side. It can be seen that there is a marking line image GP1a. Further, there is a sidewalk image GI1 on the left side of the solid lane line image GP1a. Further, there is a traveling direction between the broken line lane line image GP2a on the left side and the solid line lane line image GP1a on the right side. There is a traffic classification image GP3. Since these images of the respective recognition objects are associated with the information on the various features included in the road feature information C by the processing of step # 82, the arrangement of the images of the respective recognition objects in the image information G is arranged. In the road feature information C shown in FIG. 14, the imaging position in the road width direction is the imaging position in the right lane of the three-lane road 11 (the lane in which the B1 position exists). Can be identified. Further, based on the arrangement in the image information G of the solid line image GP1a or the broken line GP2a, particularly in the width direction, a more detailed position such as the left side or the right side in the left lane is also specified. It is possible.

For example, when the imaging position of the image information G is in the central lane of the three-lane road 11 as in the B2 position in FIG. 14, the broken line P2a and P2b on both sides from the center of the image information G The image will be recognized. Further, for example, when the imaging position of the image information G is in the right lane of the three-lane road 11 as in the B3 position in FIG. 14, the image of the broken line P2b on the right side from the center of the image information G However, the image of the solid dividing line P1b is recognized on the left side.
In the present application, it is important to recognize and distinguish between the solid lane markings P1a and P1b and the dashed lane markings P2a and P2b. However, in the state where the vehicle is stopped or traveling at a very low speed, it is recognized. Since the configuration in which the processing itself is not performed is adopted, the solid line and the broken line can be identified and recognized with high reliability.

  With the above arithmetic processing, the imaging position of the image information G can be specified in detail in both the road longitudinal direction and the road width direction. And since the imaging device 2 is mounted in the vehicle M, this specified imaging position is specified as the detailed position of the vehicle M (step # 84). Then, the position of the vehicle M specified in this way is used as a highly accurate position for detailed correction of the own vehicle position (step # 85).

  The series of processing steps of steps # 61 to # 67 described above are repeatedly performed at predetermined time intervals. Thereby, the detailed position specification of the vehicle M that is traveling can always be performed in real time.

  The detailed vehicle position specifying result by the vehicle position specifying calculation unit 17 is output to, for example, a travel control device or a navigation device of the vehicle M (not shown), and travel control of the vehicle M such as a lane keep or the vehicle speed. Or used for display of a detailed vehicle position in a navigation device.

[Second Embodiment]
Next, a second embodiment will be described based on the drawings. FIG. 15 is a block diagram showing an outline of the hardware configuration of the vehicle position recognition correction unit 103 according to this embodiment. Also in this example, when the vehicle is stopped or when traveling at a very low speed, the processing in the vehicle position recognition correction unit 103 is not performed.
The own vehicle position recognition correction unit 103 according to the present embodiment includes a plurality of different road feature information C regarding the features around the imaging position of the image information G from the map information in the road feature information acquisition calculation unit 9 for each lane of the road 11. Lane-specific road feature information C ′ based on the position of each lane, and for each lane-specific road feature information C ′, contrast with the arrangement of the recognition object corresponding to the feature in the image information G The point of specifying the lane position of the vehicle M (a kind of position of the host vehicle) is different from the first embodiment.
In addition, the vehicle position recognition correction unit 103 according to the present embodiment acquires information from the vehicle M regarding the route of the vehicle M and information regarding the past travel route of the vehicle M to estimate the lane position of the vehicle M. It differs from said 1st embodiment also by the point which is provided with the estimation calculating part 18, and specifies the lane position of the vehicle M using the estimation result by this.

  As shown in FIG. 15, the own vehicle position recognition correction unit 103 according to the present embodiment further includes a vehicle position estimation calculation unit 18 in addition to the configuration described in the first embodiment as a main configuration. . The vehicle position estimation calculation unit 18 includes a vehicle information acquisition unit 19 that acquires information from the vehicle M regarding the route of the vehicle M, and a past route storage unit that acquires and stores information related to past travel routes of the vehicle M. 20 is connected, and based on the information from the vehicle information acquisition unit 19 and the past route storage unit 20, a calculation process for estimating the position of the lane of the vehicle M is performed. And the estimation result by this vehicle position estimation calculating part 18 is output to the road characteristic information acquisition calculating part 9, and is used for the process for acquisition of road characteristic information C 'according to lane.

In this embodiment, the vehicle information acquisition unit 19 is connected to the driving operation detection unit 21, the GPS receiver 4, the direction sensor 5, and the distance sensor 6. Here, the GPS receiver 4, the azimuth sensor 5, and the distance sensor 6 are shared with those already connected to the approximate position specifying calculation unit 7, and these outputs are also transmitted to the vehicle information acquisition unit 19. It is configured to output. Thereby, the vehicle information acquisition part 19 can acquire information, such as a moving direction of the vehicle M, a moving distance, and steering wheel operation.
Further, the driving operation detection unit 21 is a driving operation by the driver, for example, operation of a direction indicator or steering wheel operation (excluded because they are duplicated when provided as a configuration of the direction sensor 5), accelerator operation, brake operation, and the like. The detection result is also output to the vehicle information acquisition unit 19.

And the vehicle information acquisition part 19 produces | generates the information regarding the course of the vehicle M combining the vehicle information acquired from each part of these vehicles, and outputs the information to the vehicle position estimation calculating part 18 and the past route memory | storage part 20 To do. Specifically, the information regarding the course of the vehicle M is information such as whether or not the course of the vehicle M has been changed and its angle.
The vehicle information acquisition unit 19 has a functional unit for performing various processes on input data using a calculation processing device such as a CPU for performing such calculation processing as a core member. Or it has the structure mounted by both.

The past route storage unit 20 stores information related to the route of the vehicle M output from the vehicle information acquisition unit 19 in association with information such as the travel distance and travel time of the vehicle M as information related to the past travel route of the vehicle M. Process. Information relating to the past travel route of the vehicle M stored in the past route storage unit 20 is output to the vehicle position estimation calculation unit 18 in accordance with a command signal from the vehicle position estimation calculation unit 18.
The past path storage unit 20 includes a calculation processing device for performing such calculation processing and a memory for storing the calculation result as a core member, and a functional unit for performing various processing on input data. It has a configuration implemented in hardware and / or software.

  In addition, the vehicle position recognition correction unit 103 according to the present embodiment includes a road feature information acquisition calculation unit 9 including a lane information acquisition unit 9a and a road-specific road feature information acquisition calculation unit 9b, and a vehicle position The specific calculation unit 17 is also different from the first embodiment in that it includes an imaging lane identification unit 17e instead of the arrangement information extraction unit 17a and the imaging position identification unit 17c. These differences are due to the fact that the vehicle position recognition processing by the vehicle position recognition correction unit 103 according to the present embodiment is different from that of the first embodiment. Therefore, processing performed by each of these units will be described in detail with reference to flowcharts below.

FIG. 16 is a flowchart showing a specific example of the process of specifying the lane position during travel of the vehicle M by the vehicle position recognition correction unit 103 according to the present embodiment.
As shown in this figure, the vehicle position recognition / correction unit 103 according to the present embodiment first captures image information G imaged by the imaging device 2 (step # 101), and the image preprocessing circuit 13 uses the image information G. Is pre-processed (step # 102). Then, the image information G2 after the preprocessing in step # 102 and the image information G1 sent directly from the interface circuit 12 are stored in the image memory 14 (step # 103). In parallel with the processing in steps # 102 and # 103, the approximate position specifying calculation unit 7 performs processing for obtaining the approximate imaging position of the image information G (step # 104). Note that the processing of these steps # 101 to # 104 is the same as the processing of steps # 61 to # 64 in FIG. 6 for the first embodiment, and detailed description thereof will be omitted.

  Next, the vehicle position estimation calculation unit 18 performs a process of estimating the lane in which the vehicle M is traveling (step # 105). The process of estimating the lane is performed based on information from the vehicle information acquisition unit 19 and the past route storage unit 20. That is, the vehicle information acquisition unit 19 outputs information related to the course of the vehicle M to the vehicle position estimation calculation unit 18 based on information from sensors and the like of each unit of the vehicle M. The past route storage unit 20 associates the information about the route of the vehicle M output from the vehicle information acquisition unit 19 with information such as the travel distance and the travel time of the vehicle M as information about the past travel route of the vehicle M. I remember it. Therefore, the vehicle position estimation calculation unit 18 determines from the vehicle information acquisition unit 19 and the past route storage unit 20 the history of the number of course changes made by the vehicle M in the past, the angle of each course change, and the presence / absence of the current course change. The information of the situation etc. can be obtained. Further, the vehicle position estimation calculation unit 18 can determine whether or not each route change is a lane change from an angle of a route change performed by the vehicle M, an operation of a direction indicator, or the like. Based on these pieces of information, the vehicle position estimation calculation unit 18 estimates a running lane according to a predetermined determination algorithm.

This determination algorithm can be performed as follows, for example. That is, for example, it is estimated that the lane when the vehicle M starts traveling is the leftmost lane. Moreover, when the vehicle M performs the course change corresponding to the lane change n times in the right direction from the state, it can be estimated that the vehicle M is in the nth lane from the left (n is a natural number). Furthermore, when the vehicle M has made a course change corresponding to the lane change m times in the left direction, it can be estimated that the vehicle M is in the (n−m) th lane from the left (m is a natural number). . At this time, if (nm) becomes zero or a negative value, the estimated lane is incorrect, so that the lane at that time is corrected to be the leftmost lane. . Also in this case, the identification recognition of the solid lane line and the broken lane line with the lane line recognition plays an important role.
The above determination algorithm is merely an example, and various other algorithms can be used as the determination algorithm in the vehicle position estimation calculation unit 18.

  Thereafter, based on the estimation result in step # 105, the road feature information acquisition calculation unit 9 performs processing for acquiring road-specific road feature information C ′ for the lane in which the vehicle M is estimated to be traveling (step # 106). . Here, the lane information acquisition unit 9a first acquires lane information including the number of lanes of the road 11 around the approximate imaging position specified in step # 104 from the map information database 8, and then acquires the acquired lane. Based on the information, the lane-specific road feature information C ′ for the lane estimated as the estimation result in step # 105 is acquired by the lane-specific road feature information acquisition calculation unit 9b. In step # 108, which will be described later, the acquired road-specific road feature information C ′ is compared with the image information G, so the order of acquiring the road-specific road feature information C ′ is determined based on the estimation result in step # 105. By doing so, the contrast order of the road characteristic information C ′ for each lane is also determined.

  The lane-specific road feature information C ′ specifies the road feature information C related to the features around the approximate imaging position obtained in step # 104 from a wide range of map information stored in the map information database 8. It is the information extracted on the basis of the lane. FIG. 17 is a diagram showing an example of the road-specific road characteristic information C ′. As shown in this figure, in this example, the road feature information C ′ for each lane is the road feature information C around the approximate imaging position specified in step # 104, which is 3 for the left lane, the center lane, and the right lane. For each lane, each lane is used as a reference and is extracted in a range including information on the lane and information on features existing in a predetermined range on both sides thereof. 17A shows road characteristic information C′1 by lane based on the left lane, FIG. 17B shows road characteristic information C′2 by lane based on the center lane, and FIG. 17C shows the right lane. The lane-specific road feature information C′3 as a reference is shown. The arrangement of the entire features on the road 11 shown in FIG. 17 is the same as the example shown in FIG.

  Next, the image information recognition calculation unit 10 performs a process of recognizing the image information G, and performs a process of recognizing the image of the recognition object corresponding to the feature included in the image information G (step # 107). . Since this process is the same process as step # 66 of FIG. 6 for the first embodiment, detailed description thereof is omitted.

  Thereafter, in the contrast calculation unit 17b of the vehicle position specifying calculation unit 17, the image information G including the image of the recognition object recognized in step # 107 and the road-specific road characteristic information C ′ acquired in step # 106 are obtained. Comparison processing is performed (step # 108). In the present embodiment, the lane-specific road feature information C ′ is converted into an information format that can be compared with the image information G, and the converted lane-specific road feature information C ′ is compared with the image information G. Therefore, it is determined whether or not the consistency is high. Here, as shown in the figure, as the conversion processing of the road feature information C ′ by lane, as shown in the figure, the road feature information C ′ by lane is set to approximately the center of the lane based on the road feature information C ′ by lane. It is assumed that the processing is converted into data in which each feature included in the road characteristic information C ′ for each lane is arranged so that the arrangement corresponds to the image information that is considered to be imaged when the imaging position is set. In the example shown in FIG. 18, (a) is data obtained by converting road-specific road feature information C′1 based on the left lane shown in FIG. 17 (a), and (b) is shown in FIG. 17 (b). This is data obtained by converting road-specific road characteristic information C′2 based on the central lane, and (c) is data obtained by converting road-specific road characteristic information C′3 based on the right lane shown in FIG. It is.

  By performing such conversion processing, the arrangement of the image of the recognition object in the image information G can be easily compared with the arrangement of the various features corresponding to the recognition object included in the lane-specific road feature information C ′. can do. More specifically, the comparison processing is performed by the recognition of the position of the image of each recognition object in the image information G, the shape and color of the image of each recognition object, and the recognition included in the road-specific road feature information C ′. The position of each feature corresponding to the object and the information such as its shape and color are compared to determine whether or not the consistency is high. For example, when the image information G is as shown in the example shown in FIG. 9, the road-specific road feature information C′1 based on the left lane shown in FIG. 17A is the lane marking P1a on both sides of the lane. , P2a, the traveling direction passage section P3, and the position, shape, color, and the like of the sidewalk I1 are matched with the image information G.

  If the result of the comparison in step # 108 in the comparison calculation unit 17b is high in consistency (step # 109: yes), the imaging lane identification unit 17b in the vehicle position identification calculation unit 17 uses the road-specific road feature. The lane based on the information C ′ is specified as the lane in which the vehicle M is traveling (step # 110).

  On the other hand, if the result of the comparison in step # 108 in the comparison calculation unit 17b is that the consistency is low (step # 109: no), the road-specific road characteristic information C ′ of the adjacent lane is acquired from the map information database 8. Processing is performed (step # 106). Here, even if the road characteristic information C ′ for each lane of the adjacent lane is acquired even if the estimation result in step # 105 is off, there is a possibility that the vehicle M is traveling in a lane close to that. Because it is expensive. At this time, as a result of the judgment in step # 109, for example, when the road characteristic information C ′ classified by lane first compared in step # 108 is based on the center lane of the three lanes, as a result of the determination, If there are lanes that are adjacent to a low-lane lane that is not yet subject to comparison, for example, the left-hand lane should be compared first, and then determined according to a predetermined algorithm. Become.

Then, with respect to the new lane-specific road feature information C ′ acquired at step # 110, the processing after step # 108 is performed again, and it is determined that the consistency is high at step # 109, and the lane in which the vehicle M is traveling is determined. Steps # 108 to # 110 are repeated until the identification process is performed or the contrast processing of step # 108 is performed for the lane-specific road feature information C ′ of all lanes of the road 11 on which the vehicle M is traveling. . Also in this case, the identification recognition of the solid lane line and the broken lane line with the lane line recognition plays an important role. Although not shown in the flowchart of FIG. 16, as a result of performing the contrast processing of step # 108 on the road feature information C ′ for each lane of all the lanes of the road 11 on which the vehicle M is traveling, the lane having high consistency is obtained. If the separate road feature information C ′ does not exist, it is determined that the lane position is unknown, and the processing of steps # 101 to # 111 is executed for the next image information G.
With the above processing, the lane in which the vehicle M is traveling can be specified.
Then, the own lane update correction unit 17f updates and corrects the own lane to the newly recognized detailed lane (step # 112).
Also in this embodiment, the own line and the broken line, which are lane markings, are accurately recognized, and whether or not the lane change across the lane markings has been accurately recognized, and a highly reliable apparatus can be obtained.

[Other Embodiments]
(1) In the above embodiment, the configuration as a feature recognition device that executes recognition of features and the configuration as a vehicle position recognition device have been described. However, these devices use a map database that stores map information, In order to provide as “map information storage means”, when the vehicle control information can be generated using the vehicle position information obtained from the vehicle position recognition device and the map information, the device is a vehicle control device. be able to.

(2) In the above embodiment, the configuration as a feature recognition device that performs recognition of a feature and the configuration as a vehicle position recognition device have been described, but these devices use a map database that stores map information, In order to provide as "map information storage means", when the navigation information can be generated using the vehicle position information obtained from the vehicle position recognition device and the map information, the device may be a navigation device. it can.

(3) In the above embodiment, when the detailed vehicle position is specified, the case where the arrow tip of the traveling section according to the traveling direction is adopted as the measurement point is shown, but in addition to using the arrow tip, the base end of the arrow Or the root of an arrow. That is, the corner of the contour shape of an arbitrary feature can be selected. Furthermore, as described above, as features, stop lines, pedestrian crossings, pedestrian crossing notices, turn prohibitions, speed indications, etc. can be used, but these are on contour lines that are substantially orthogonal to the direction of road travel. Measurement points can be set in

(4) In the vehicle position recognition device 1 according to the second embodiment described above, the case where the lane position where the vehicle M is traveling is specified as the detailed position of the vehicle M has been described, but the present invention is not limited to this. It is not something. For example, as road feature information C for each position based on a plurality of positions different in the width direction of the road 11, road feature information C is obtained for each position further subdivided than for each lane, thereby providing a more detailed road. It is also a preferred embodiment to have a configuration capable of specifying the position in the width direction.

(5) In the vehicle position recognition device 1 according to the second embodiment, the configuration for specifying the lane position in the road width direction as the detailed position of the vehicle M has been described. In the configuration, as in the first embodiment, the direction along the road 11, that is, the road, is based on the image of the recognition object such as a stop line, a sign, or a traffic light that is not provided along the road 11. A configuration in which the longitudinal imaging position can be specified in detail is also one preferred embodiment.

  In a feature recognition device that recognizes a feature based on feature image pickup information and road feature information stored in the map database, for example, a solid line as a division line and a broken line are erroneously identified. It was possible to obtain a feature recognition device that would not cause any serious problems.

The block diagram which shows the outline of the hardware constitutions of the recognition correction control function which concerns on 1st, 2nd embodiment The figure which shows the example of arrangement | positioning of the imaging device and rotation sensor in the own vehicle position recognition correction apparatus which concern on 1st, 2nd embodiment. The flowchart which shows the specific example of the recognition correction control process which concerns on 1st, 2nd embodiment. The block diagram which shows the outline of the hardware constitutions of the own vehicle position recognition correction | amendment part which concerns on 1st embodiment. Explanatory drawing which shows the content of the map information stored in the map information database in the vehicle position recognition apparatus which concerns on 1st embodiment. The flowchart which shows the specific example of the recognition processing of the vehicle position in the vehicle position recognition apparatus which concerns on 1st embodiment. The flowchart which shows the detail of a process of step # 66 of FIG. The flowchart which shows the detail of a process of step # 67 of FIG. (A): An example of image information picked up by the image pickup apparatus, (b): An example of image information after pre-processing the image information shown in (a) The figure which showed in figure the example of the road characteristic information acquired by the road characteristic information acquisition calculating part which concerns on 1st embodiment (A): a diagram showing only a paint display image extracted from the image information by the process of step # 73; (b): a state in which regions are partitioned by lane markings based on the image shown in (a). Illustration A graph showing the result of detecting the number of edge points at each position in the width direction of the road for the image information shown in FIG. The figure which shows an example of the image information imaged with the imaging device several frames before Fig.9 (a) Explanatory drawing for demonstrating an example of the method of specifying a detailed imaging position in the vehicle position specific calculation part which concerns on 1st embodiment. The block diagram which shows the outline of the hardware constitutions of the own vehicle position recognition correction | amendment part which concerns on 2nd embodiment. The flowchart which shows the specific example of the recognition process of the vehicle position which concerns on 2nd embodiment. The figure which showed in figure the example of the road characteristic information according to lane acquired by the road characteristic information acquisition calculating part which concerns on 2nd embodiment The figure which shows the data which converted the road feature information according to lane shown in FIG. 17 into the information format which can be compared with the image information G

Explanation of symbols

2: Imaging device 3: Image information acquisition unit (image information acquisition means)
8: Map information database (map information storage means)
9: Road feature information acquisition calculation unit (road feature information acquisition means)
10: Image information recognition calculation unit (image information recognition means)
17: Vehicle position specifying calculation unit (vehicle position specifying means)
18: Vehicle position estimation calculation unit 101a: Rotation sensor (speed information acquisition means)
101: Traveling speed deriving unit (speed information acquiring means)
102: Recognition correction processing control unit (recognition correction processing control means)
103: Own vehicle position recognition correction unit (own vehicle position recognition correction means)
G: Image information C: Road feature information

Claims (7)

Image information acquisition means for sequentially capturing image information obtained by imaging at least a road surface by an imaging device mounted on a vehicle;
Road feature information acquisition means for acquiring road feature information related to features around the imaging position of the image information from map information;
Image information recognition means for performing recognition processing of the image information, and recognizing an image of a recognition object corresponding to the feature included in the image information ,
The feature for recognizing the feature based on the road feature information acquired by the road feature information acquisition means and the arrangement in the image information of the image of the recognition object recognized by the image information recognition means. A recognition device,
The image information recognizing means recognizes an image of the recognition object included in the single frame for a single frame when performing image recognition using the paint display provided on the road surface of the road as the recognition object. The first recognition process is performed, and the recognition target object recognized as a solid line or a broken lane line by the first recognition process is subjected to a plurality of consecutive frames and is substantially the same position between the plurality of frames. When the lane markings taken in the above are recognized as the same lane markings, and when the rupture portion is recognized as a broken line with respect to the lane markings recognized as the same, It is configured to execute a second recognition process that recognizes as a solid line,
A velocity information acquiring means for acquiring the traveling speed information of the vehicle,
When the traveling speed information obtained is lower than a predetermined value, the feature recognition apparatus and a recognition correction control means for stopping the recognition of the recognition target object by the image information recognizing unit. At least a position information acquisition means for determining the approximate position of the vehicle based on positioning system position information obtained from the global positioning system,
The feature recognition apparatus according to claim 1, wherein a position of the vehicle obtained by the position information acquisition unit when the image information is captured is set as an imaging position of the image information. The feature recognition apparatus according to claim 1 or 2 ,
Performing image recognition of the measurement points for the recognition object included in the image information for the measurement points predetermined for the recognition object,
Vehicle position recognition correction means for updating and correcting the vehicle position information based on the image recognition result of the recognized measurement point and the position information of the measurement point obtained in advance;
A vehicle position recognition device comprising: The measuring points, depending on the contour of each feature, contour substantially perpendicular to the traveling direction of the vehicle, or claim 3 which is set on the corner portion if the contour has a corner The vehicle position recognition device described in 1. The host vehicle position recognition device according to claim 3 or 4 and map information storage means storing map information,
A vehicle control device that performs travel control of the host vehicle based on map information of the traveling direction of the host vehicle acquired from the map information storage unit and host vehicle position information after update correction by the host vehicle position recognition correction unit. The vehicle position recognition device according to claim 3 or 4, and map information storage means storing map information,
A navigation device that provides route guidance for the host vehicle based on the map information of the traveling direction of the host vehicle acquired from the map information storage unit and the host vehicle position information after the update correction by the host vehicle position recognition correction unit. An image information acquisition step for sequentially capturing image information obtained by imaging at least the road surface of the road by an imaging device mounted on the vehicle;
Road feature information acquisition step for acquiring road feature information related to features around the imaging position of the image information from map information;
Recognizing the image information and recognizing an image of a recognition object corresponding to the feature included in the image information; and
A feature that recognizes the feature based on the road feature information acquired in the road feature information acquisition step and the arrangement in the image information of the image of the recognition object recognized in the image information recognition step. To run the recognition method,
In the image information recognition step, when performing the image recognition using the paint display provided on the road surface of the road as the recognition object, the image recognition of the recognition object included in the single frame is performed with respect to a single frame. After performing the first recognition process to be performed, with respect to the recognition target object recognized as a solid line or a broken lane line by the first recognition process, it is substantially the same across the plurality of frames for a plurality of consecutive frames. When a lane line photographed at a position is recognized as the same lane line, and when a broken portion is recognized with respect to the lane line recognized as the same, the lane line is recognized as a broken line and no broken portion is recognized Is configured to execute a second recognition process for recognizing a solid line,
Execute a speed information acquisition step of acquiring vehicle travel speed information;
A feature recognition method for stopping recognition of the recognition object by the image information recognition step when the acquired traveling speed information is lower than a predetermined value.

Images