JP2009099125A - Image recognition device, image recognition program, and point information collection device and navigation device using them - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image recognition device capable of extracting stop lines at which drivers are obligated to stop, and recognizing them, and capable of increasing the recognition precision. <P>SOLUTION: The present invention includes: a means for acquiring information on a location of one's own vehicle for acquiring information on a location of the own vehicle that shows the present location of the own vehicle; an image information acquisition means for acquiring image information on the periphery of the own vehicle; an image recognition means for performing image recognition processing for an object feature included in the image information, a combination of a stop line provided on the surface of a road and a predetermined character string placed adjacent to the stop line being the object feature; and a point information generation means for generating information on a point at which drives are obligated to stop showing the point at which drivers are obligated to stop, based on the information on the location of the own vehicle, and the result of the image recognition for the object feature when the image recognition for the object feature is successful. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an image recognition device and an image recognition program for performing image recognition processing of a target feature included in image information around a vehicle acquired by an imaging device or the like mounted on the vehicle, and a point information collection device and navigation using the same. Relates to the device.

  2. Description of the Related Art Conventionally, an image recognition device that performs image recognition processing using a stop line provided on a road surface as a target feature from image information around a vehicle acquired by an imaging device or the like mounted on the vehicle is known ( For example, see Patent Document 1). The image recognition apparatus described in Patent Document 1 is a luminance change comparison that compares a luminance of a setting area that detects a luminance of a setting area that is long in the vehicle width direction in front of a traveling road with a time-series luminance change of the setting area. And stop line determination means for determining whether or not a stop line exists based on a time-series luminance change in the setting area. Specifically, a plurality of setting areas are arranged in the vehicle width direction on the lower side of the original image input from the CCD camera, and the average luminance of any of these plurality of setting areas increases from a low state. Then, when it becomes low after that, it is determined that a stop line exists.

JP 2003-85562 A

  Now, the stop line which is the target feature includes a stop line having a so-called temporary stop obligation and a simple stop line. And it is necessary to stop at the stop line with the obligation to stop temporarily. Therefore, it is preferable that the position of the stop line, which is very important for traveling and has a duty to stop, can be recognized with high accuracy in advance.

However, the above technique determines that at least a stop line exists when the average brightness of any of the plurality of setting regions arranged in the vehicle width direction increases from a low state and then decreases again. However, it is not determined whether or not it is a point that must be stopped.
In addition, when there is a paint display other than a stop line having a shape that is long in the vehicle width direction on the road surface, there may be a problem that such other paint display is recognized as a stop line. For example, a center line of a crosswalk or a road intersecting at a T-shaped road is a paint display that is long in the vehicle width direction, and thus there is a high possibility that it is erroneously recognized as a stop line.

  The present invention has been made in view of the above-described problems, and an object of the present invention is to extract and recognize only a stop line that is obligated to pause, and to improve the recognition accuracy. It is an object of the present invention to provide a recognition device, an image recognition program, and a point information collection device and a navigation device using the same.

  In order to achieve the above object, the image recognition apparatus according to the present invention is characterized by a vehicle position information acquisition means for acquiring vehicle position information representing the current position of the vehicle, and image information around the vehicle. The target feature included in the image information is a combination of image information acquisition means, a stop line provided on the road surface, and a predetermined character string provided adjacent to the stop line. An image recognition means for performing the image recognition processing, and when the image recognition of the target feature is successful, based on the vehicle position information and the image recognition result of the target feature, a stop indicating a point where the suspension is obligated And point information generating means for generating duty point information.

  In general, a stop line is provided at a point where there is a duty to pause, and a character string such as “Tare” is painted on the road surface in front of it. On the other hand, a stop line provided at a point where there is no obligation to temporarily stop is a single stop line. Therefore, in the present application, when the above configuration is adopted and image recognition of a combination of a stop line provided on the road surface and a predetermined character string provided adjacent to the stop line is successful, the stop line Since it is determined that the position is a point having a duty to pause, it is possible to extract and recognize only a stop line having a duty to pause. In addition, in order to recognize the combination of the stop line and the predetermined character string, the paint display that is not adjacent to the character string out of the long paint display in the vehicle width direction will be removed from the target feature. The recognition accuracy can be improved compared to the case of image recognition.

  Here, it is preferable that the image recognition unit determines that the target feature has been successfully recognized when the stop line is image-recognized after the predetermined character string is image-recognized.

  According to this configuration, it is possible to perform the image recognition process while moving forward even when the imaging apparatus is provided in either the forward direction or the backward direction with respect to the traveling direction of the host vehicle. In that case, since it is determined that the image recognition of the target feature has succeeded only when the image recognition is performed in the order of the predetermined character string and the stop line, erroneous recognition can be suppressed.

  Here, it is preferable that the image recognition unit determines that the image of the target feature has been successfully recognized when the predetermined character string is image-recognized after the stop line is image-recognized.

  According to this configuration, a combination of a stop line and a predetermined character string for a vehicle traveling in a direction opposite to the traveling direction of the host vehicle can be subjected to image recognition processing as a target feature. Therefore, it becomes possible to efficiently generate the stop duty point information. In that case, since it is determined that the image recognition of the target feature has succeeded only when the image recognition is performed in the order of the stop line and the predetermined character string, it is possible to suppress erroneous recognition.

  Here, as the image recognition processing of the predetermined character string included in the target feature, the image recognition means is a character string of “Tare” represented by any one of hiragana, katakana, kanji, and a combination thereof. It is preferable that the image recognition of at least some of the characters is performed.

  The character paint display, for example, has a more complicated shape than the paint display such as the stop line and the center line. Therefore, when recognizing the image of the character string, only image recognition of a part of the characters is performed. Can be determined sufficiently. Therefore, the above-described effects can be obtained not only when the image recognition process is performed on the entire “to rare” but also when the image recognition process is performed only on a part of the image recognition process. Furthermore, when the number of characters for image recognition is reduced, the processing burden for image recognition can be reduced.

  Here, the image recognition means performs image recognition of features having character characteristics for three characters that do not specify the content of characters as image recognition processing of the predetermined character string included in the target feature. It is preferable.

  According to this configuration, as the predetermined character string included in the target feature, a character string for three characters having character characteristics that does not specify the character content is set. Even when the part is faint and disappearing, the image recognition process can be performed. Moreover, since the content of the character is not specified, the image recognition process can be simplified, and a general-purpose recognition process independent of the character content can be executed. Therefore, it is possible to reduce the processing burden of image recognition.

  Here, it is preferable that the stop duty point information is configured to include position information of the stop line included in the target feature that has been successfully image-recognized and attribute information indicating that there is at least a temporary stop duty.

  According to this configuration, the position information of the newly generated stop line and the attribute information indicating that there is a suspension obligation are output, for example, alerting the driver of the vehicle to suspend, It can be used for various purposes such as vehicle control.

  Here, it is preferable that the position information of the stop line is coordinate information on a map of a representative point of the stop line.

  According to this configuration, it is possible to accurately acquire the position of the point where the suspension is required as absolute coordinate information.

  Further, the information processing apparatus includes map information storage means for storing map information including road network information representing road connection relations by nodes corresponding to intersections and links corresponding to roads connecting the intersections, and the stop duty point information includes image recognition. It is preferable to have attribute information indicating that there is a suspension obligation at the link end in the link where the target feature that has succeeded successfully.

  According to this configuration, it is possible to reduce the processing load for generating the stop duty point information using the information related to the link stored in the map information storage unit. In addition, it is possible to sufficiently provide a small amount of data, for example, for guidance of a stop duty point.

  A characteristic configuration of the spot information collecting apparatus according to the present invention is that the spot information collecting apparatus includes the above-described image recognition apparatus and spot information storage means for storing the stop duty spot information generated by the spot information generating means.

  According to this configuration, since it is possible to store the stop duty point information generated based on the image recognition result in the above image recognition device, a point information collection device capable of collecting and storing information related to the point where the suspension is required is provided. can do.

  Further, another characteristic configuration of the spot information collecting apparatus according to the present invention is the same as the image recognition apparatus described above and the spot information storage means for storing the stop duty spot information generated by the spot information generating means. The estimated position of the point having the duty to pause based on a plurality of the duty points to be stopped for the same point, which is generated by image recognition of the target feature a plurality of times and stored in the point information storage unit An estimated position determination means for determining the statistical position information generating means for generating statistical stop duty point information that uses the estimated position determined by the estimated position determination means as a position of the point where the temporary stop duty exists. And statistical point information storage means for storing the statistical stop obligation point information generated by the statistical point information generation means.

  According to this configuration, based on a plurality of stop duty point information for the same point stored in the point information storage unit by image recognition of the same target feature multiple times, An estimated position is determined. Therefore, even if there is an error in the recognition position of the point with the suspension obligation indicated by each stop duty point information, the error is averaged by determining the estimated position using multiple stop duty point information. It becomes possible to make the estimated position of the point having the obligation to pause as close as possible to the actual position. And statistical stop duty point information can be produced | generated using an estimated position, and this can be memorize | stored. Therefore, highly accurate information can be collected.

  Here, the estimated position determination means determines a representative value of the distribution as the estimated position based on a distribution of positions of the points having the temporary stop obligation represented by the plurality of stop duty point information for the same point. Is preferred.

  According to this structure, the estimated position of the said point can be appropriately determined by the statistical process using the several stop obligation point information about the point with the same temporary stop obligation.

  Here, in the image recognition apparatus that generates the above-described stop duty point information, when image recognition of the target feature is successful, the position of the predetermined character string included in the target feature that has succeeded in the image recognition It is preferable to further include character string information generating means for generating character string information having information and attribute information.

  According to this configuration, it is possible to output position information and attribute information of a newly generated predetermined character string and use it for other purposes.

  Still another characteristic configuration of the point information collecting apparatus according to the present invention is that the character recognition information generated by the character recognition information generation unit and the image recognition device that generates the stop duty point information and the character string information are included. And character string information storage means for storing the information in association with the stop duty point information.

  According to this configuration, it is possible to provide a point information collection device that can collect abundant information about a point for which a suspension is required.

  The characteristic configuration of the navigation device according to the present invention is the guidance means for guiding the point where the suspension is required based on the point information collection device described above and the own vehicle position information and the stop obligation point information. And in the point provided.

  According to this configuration, since the navigation apparatus guides with reference to the stop duty point information collected using the above-described point information collection device, even for points that were not originally stored in the database, It is possible to provide guidance to the effect.

  The characteristic configuration of the image recognition program according to the present invention includes a host vehicle position information acquisition step for acquiring host vehicle position information representing the current position of the host vehicle, an image information acquisition step for acquiring image information around the host vehicle, An image for performing image recognition processing of the target feature included in the image information with a combination of a stop line provided on the road surface and a predetermined character string provided adjacent to the stop line as a target feature A recognition step, and when the image recognition of the target feature is successful, based on the image recognition result of the target feature, a point information generation step of generating stop obligation point information representing a point where there is a suspension obligation; Is to make the computer execute.

  According to this configuration, when the image recognition of a combination of a stop line provided on the road surface and a predetermined character string provided adjacent to the stop line is successful, the position of the stop line is temporarily stopped. Since it is determined that the point is obligatory, it is possible to extract and recognize only the stop line having a temporary stop obligation. In addition, in order to recognize the combination of the stop line and the predetermined character string, the paint display that is not adjacent to the character string out of the long paint display in the vehicle width direction will be removed from the target feature. The recognition accuracy can be improved compared to the case of image recognition.

[First embodiment]
Next, a first embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram showing a schematic configuration of a navigation device 1 to which a point information collection device 2 including an image recognition device 3 according to this embodiment is applied. When the image recognition device 3 succeeds in image recognition of a combination of a stop line provided on the road surface of a road and a predetermined character string provided adjacent to the stop line, the position of the stop line is temporarily changed. The position information and the attribute information of the stop line included in the target feature for which image recognition has been successfully performed are determined as the stop duty point information s. Then, the point information collection device 2 stores the stop duty point information s generated by the image recognition device 3. In addition, the navigation device 1 performs normal processing such as route guidance to the destination, and performs guidance by referring to the information related to the point with the duty of temporary suspension collected by the point information collection device 2.

  Each functional unit of the spot information collection device 2 shown in FIG. 1, specifically, an image information acquisition unit 5, an image recognition unit 6, an own vehicle position information acquisition unit 4, a spot information generation unit 9, an estimated position determination unit 10, The statistical point information generation unit 11 and the navigation calculation unit 12 have a functional unit for performing various processes on input data using a calculation processing device such as a CPU as a core member, or hardware or software (program). Or it is mounted and comprised by both. Each of these functional units is configured to exchange information with each other. Further, the map database DB1 and the feature database DB2 are, for example, a recording medium capable of storing information, such as a hard disk drive, a DVD drive equipped with a DVD-ROM, a CD drive equipped with a CD-ROM, and a driving means thereof The learning database DB3 and the stop obligation point database DB4 are, for example, a rewritable recording medium such as a hard disk drive, a DVD drive equipped with a DVD-RAM, and the like. A device having a driving means is provided as a hardware configuration. Hereinafter, the configuration of each part of the navigation device 1 according to the present embodiment will be described in detail.

1-1. Map database The map database DB1 is a database in which map information M divided into predetermined sections is stored. FIG. 2 is a diagram illustrating an example of the configuration of the map information M stored in the map database DB1. As shown in this figure, the map information M has road information Ra representing a road network by connection relations between a large number of nodes n corresponding to intersections and links k corresponding to roads connecting the intersections. . Each node n has information on the position (coordinates) on the map expressed by latitude and longitude. Each link k is connected via the node n, and has attribute information such as road type, area type, link length, road width, and shape interpolation point for expressing the link shape. . In FIG. 2, only the road information Ra of one section is illustrated, and the road information Ra of other sections is omitted.

1-2. Feature database The feature database DB2 is a database for storing feature information provided on the road surface of the road, that is, feature information F. FIG. 3 is a diagram illustrating an example of the feature information F of the road marking stored in the feature database DB2. In this embodiment, the feature information F stored in the feature database DB2 includes at least a stop line provided on the road surface and a predetermined character string provided adjacent to the stop line. Contains. In addition, the feature database DB2, for example, a pedestrian crossing, a stop line, a speed sign indicating a maximum speed, a zebra zone, a lane line that divides a lane along the road (a variety of sections such as a solid line, a broken line, and a double line) ), And feature information F related to a traveling direction marking (including an arrow marking, for example, a straight arrow, a right turn arrow, etc.) specifying the traveling direction of each lane. In addition to such road markings, the features in which the feature information F is stored can include various features such as traffic lights, signs, overpasses, and tunnels. The feature information F includes position information of each feature and feature attribute information associated therewith as its contents. Here, the position information includes information on the position (coordinates) on the map of the representative point of each feature associated with the link k or the node n constituting the road information Ra, and the direction of each feature. The feature attribute information includes type information indicating the type of each feature, or feature form information such as the shape, size, and color of the feature. Here, the feature type specifically indicates the type of feature having basically the same form such as “pedestrian crossing”, “stop line”, “speed indication (30 km / hour)”, and the like. Information. Although not described in detail, the feature information F can be used for correction of the vehicle position information P, vehicle control, and the like, together with image recognition processing of actual features on the road by the image recognition unit 6. .

1-3. Learning Database The learning database DB3 is a database that stores the stop mandatory point information s generated by the point information generation unit 9 described later. In this learning database DB3, one or more stop obligation point information s is stored for each of a plurality of points having a stop obligation that has been successfully recognized by the image recognition unit 6. The stop obligation point information s stored in the learning database DB3 will be described later. In the present embodiment, this learning database DB3 corresponds to the point information storage means in the present invention.

1-4. Stop Obligation Point Database The stop duty point database DB4 is a database that stores statistical stop duty point information S generated by a statistical point information generation unit 11 described later. In the stop obligation point database DB4, statistical stop obligation point information S generated for each of a plurality of points having a stop obligation that has been successfully recognized by the image recognition unit 6 is stored. The statistical stop duty point information S stored in the stop duty point database DB4 will be described later. In this embodiment, this stop obligation point database DB4 corresponds to the statistical point information storage means in the present invention.

1-5. Own vehicle position information acquisition unit The own vehicle position information acquisition unit 4 functions as own vehicle position information acquisition means for acquiring own vehicle position information P indicating the current position of the own vehicle. Here, the vehicle position information acquisition unit 4 is connected to the GPS receiver 15, the direction sensor 16, and the distance sensor 17. Here, the GPS receiver 15 is a device that receives a GPS signal from a GPS (Global Positioning System) satellite. This GPS signal is normally received every second and output to the vehicle position information acquisition unit 4. The own vehicle position information acquisition unit 4 can analyze signals from GPS satellites received by the GPS receiver 15 and acquire information such as the current position (latitude and longitude), traveling direction, and moving speed of the own vehicle. it can. The direction sensor 16 is a sensor that detects a traveling direction of the host vehicle or a change in the traveling direction. The azimuth sensor 16 includes, for example, a gyro sensor, a geomagnetic sensor, an optical rotation sensor attached to the rotating part of the handle, a rotary resistance volume, and an angle sensor attached to the wheel part. Then, the direction sensor 16 outputs the detection result to the vehicle position information acquisition unit 4. The distance sensor 17 is a sensor that detects a vehicle speed and a moving distance of the host vehicle. The distance sensor 17 integrates, for example, a vehicle speed pulse sensor that outputs a pulse signal every time a drive shaft or wheel of the vehicle rotates by a certain amount, a yaw / G sensor that detects the acceleration of the host vehicle, and the detected acceleration. It is composed of a circuit or the like. Then, the distance sensor 17 outputs information on the vehicle speed and movement distance as the detection result to the vehicle position information acquisition unit 4.

  And the own vehicle position information acquisition part 4 performs the calculation which pinpoints an own vehicle position by a well-known method based on the output from these GPS receiver 15, the direction sensor 16, and the distance sensor 17. FIG. In addition, the vehicle position information acquisition unit 4 acquires map information M around the vehicle position from the map database DB1 and performs known map matching based on the map information M on the road indicated by the map information M. Correction to match is also performed. In this way, the host vehicle position information acquisition unit 4 acquires host vehicle position information P including information on the current position of the host vehicle represented by latitude and longitude and information on the traveling direction of the host vehicle. The vehicle position information P obtained by the vehicle position information acquisition unit 4 is output to the point information generation unit 9.

1-6. Image Information Acquisition Unit The image information acquisition unit 5 functions as an image information acquisition unit that acquires image information G around the host vehicle captured by the imaging device 18. Here, the imaging device 18 is an in-vehicle camera or the like provided with an imaging element, and is provided at a position where at least a road surface of a road around the host vehicle can be imaged. As such an imaging device 18, for example, a back camera that images the road surface behind the host vehicle as shown in FIG. 4 is preferably used. Below, the example which acquires the image information G around the own vehicle using such a back camera is demonstrated. The image information acquisition unit 5 captures image information G captured by the imaging device 18 at predetermined time intervals via a frame memory (not shown) or the like. The time interval for capturing the image information G at this time can be set to, for example, about 10 to 50 ms. Thereby, the image information acquisition unit 5 can continuously acquire the image information G of a plurality of frames captured by the imaging device 18. The image information G acquired here is output to the image recognition unit 6.

1-7. Image recognition unit The image recognition unit 6 uses a combination of a stop line provided on a road surface of a road and a predetermined character string provided adjacent to the stop line as a target feature, and includes a target site included in the image information G. It functions as an image recognition means for performing image recognition processing of an object. The image recognition unit 6 includes a shape recognition unit 7 that performs image recognition processing on target features provided on the road, that is, a stop line and a predetermined character string, a stop line recognized by the shape recognition unit 7, and a predetermined line. And an adjacency determination unit 8 that determines whether or not the character string is adjacent. In the present specification, unless specifically distinguished and used, the image recognition unit 6 means a combination of the shape recognition unit 7 and the adjacent determination unit 8.

1-7-1. Shape Recognizing Unit The shape recognizing unit 7 is a shape recognizing unit that performs image recognition processing on the shape of the target feature included in the image information G using the stop line and the predetermined character string provided on the road surface of the road as the target feature. Function. When recognizing the target feature, the shape recognizing unit 7 first recognizes a predetermined character string and then recognizes the stop line.
Here, the shape recognizing unit 7 performs image recognition processing of a predetermined character string included in the target feature in a character string of “to rare” represented by any one of hiragana, katakana, kanji, and a combination thereof. Image recognition of at least some characters is performed. In the present embodiment, an example will be described in which image recognition is performed on the characters “MA” and “RE” among the character strings “TOREMA”. Here, among the character strings “to rare” included in the target feature, the characters “ma” and “re” are targeted for image recognition because they have character characteristics. This is because “to rare” can be regarded as the object of image recognition. In other words, they have relatively many edges among the various features provided on the road, have complex shapes, and originally have character-like features. This is because it can be considered that the image recognition of the predetermined character string included in the target feature has succeeded. When recognizing the characters “MA” and “RE”, the image information G is subjected to binarization processing, edge detection processing, etc., and contour information of the feature (road marking) included in the image information G To extract. The shape recognizing unit 7 performs pattern matching between the extracted contour information of the feature and the feature quantities of the characters “ma” and “re” as the target feature, thereby performing the character “ma” included in the image information G. "And" re "images are extracted. As a result, when the images of the characters “MA” and “RE” can be extracted from the image information G, the shape recognition unit 7 outputs the result to the adjacency determination unit 8 and then displays the stop line. Perform image recognition on the shape. When recognizing a stop line, as in the case described above, binarization processing and edge detection processing are performed on the image information G, and the contour information of the feature (road marking) contained in the image information G is obtained. Extract. The shape recognizing unit 7 extracts a stop line image included in the image information G by performing pattern matching between the extracted contour information of the feature and the feature amount of the form of the stop line that is the target feature. As will be described later, when an image of the stop line can be extracted from the image information G until the moving distance of the host vehicle exceeds the predetermined distance D, it is provided on the target feature, that is, the road surface of the road. It is determined that the image recognition of the combination of the stop line and a predetermined character string provided adjacent to the stop line is successful. Processing performed by the adjacency determination unit 8 will be described later.

  FIG. 5 is an explanatory diagram for explaining an example of an image recognition process performed by the shape recognition unit 7 among the image recognition units 6 included in the spot information collection device 2 according to the present embodiment. Near the intersection shown in FIG. 5, there are four stop lines f1 to f4. Among these, the stop lines that are obligated to temporarily stop are f2 and f4 in which the character sign “Tare” is present on the road surface in front of it. Now, assuming that the host vehicle is moving in the direction of the arrow in the figure, the back camera mounted on the host vehicle acquires image information G as shown in FIG. That is, as shown in FIG. 6A, the image information G acquired by the back camera should first include the character “Re” in the character string “Tare”. The shape recognition unit 7 performs binarization processing, edge detection processing, and the like on the image information G, and extracts contour information of the feature (road marking) included in the image information G. The shape recognition unit 7 extracts an image of the character “re” included in the image information G by performing pattern matching between the extracted feature outline information and the feature amount of the character “re”. Next, as shown in FIG. 6B, the image information G should include the character “ma” in the character string “rare”. The shape recognition unit 7 performs binarization processing, edge detection processing, and the like on the image information G in the same manner as the image recognition processing of the character “re”, and the features (road markings) included in the image information G ) Contour information is extracted. The shape recognition unit 7 extracts an image of the character “MA” included in the image information G by performing pattern matching between the extracted contour information of the feature and the feature amount of the character “MA”. Note that, when the image recognition of the character “MA” is successful, the shape recognition unit 7 outputs the result to the adjacency determination unit 8. Next, as shown in FIG. 6C, the image information G should include the character “TO” in the character string “TORAARE”. In this embodiment, since the characters “ma” and “re” are the target features in the character string “to rare”, the shape recognition unit 7 does not perform image recognition processing on the characters “to”. Finally, as shown in FIG. 6D, the image information G should contain a stop line. The shape recognition unit 7 performs binarization processing, edge detection processing, and the like on the image information G in the same manner as the image recognition processing of the characters “re” and “ma”, and includes the ground included in the image information G. The contour information of the object (road marking) is extracted. The shape recognition unit 7 extracts a stop line image included in the image information G by performing pattern matching between the extracted feature outline information and the feature amount of the stop line form.

1-7-2. Adjacency determination unit The adjacency determination unit 8 functions as an adjacency determination unit that determines whether or not the stop line image-recognized by the shape recognition unit 7 is adjacent to a predetermined character string. The adjacency determination unit 8 determines whether or not these are adjacent by determining whether or not the distance between the stop line and the character string “to rare” is within a predetermined distance D. Here, the predetermined distance D is set in advance based on a result obtained by checking in advance the distance between the stop line constituting the target feature provided on the actual road surface and the character string “Torare”. For example, a distance that is slightly larger than the maximum distance among the distances between the stop line and the character string “Torare” at a plurality of points that are obliged to be temporarily stopped can be set as the predetermined distance D.
FIG. 7 shows an example for determining whether or not the distance between the stop line and the character string “To rare” is adjacent. In this example, when the shape recognition unit 7 receives an output when the shape recognition unit 7 succeeds in image recognition of the character “ma” from the image information G, the adjacency determination unit 8 starts from the position where the character “ma” is image-recognized. The measurement of the movement distance of the vehicle is started. The adjacency determination unit 8 is connected to the distance sensor 17, and the shape recognition unit 7 continues to execute the stop line image recognition process until the moving distance of the host vehicle exceeds the predetermined distance D. When the shape recognition unit 7 has succeeded in image recognition of the stop line before exceeding the predetermined distance D, the image recognition unit 6 is adjacent to the target feature, that is, the stop line provided on the road surface of the road, and the stop line. It is determined that the image recognition of the combination with the predetermined character string provided is successful. On the other hand, if the moving distance of the host vehicle exceeds the predetermined distance D before the shape recognition unit 7 successfully recognizes the stop line image, the stop line and the predetermined character string are not adjacent to each other. Therefore, it is not determined that the image recognition of the target feature has succeeded.
As described above, when the image recognition of the combination of the stop line provided on the road surface and the predetermined character string provided adjacent to the stop line is successful, the position of the stop line is obliged to pause. By determining that the point is a certain point, it is possible to extract and recognize only the stop line having the obligation of temporary stop. In addition, if the distance between the stop line and the predetermined character string is greater than a certain size, it is not determined that the image of the target feature has been successfully recognized. Therefore, the stop line and some character string have a certain distance. Thus, it is possible to suppress erroneous recognition of a point having a duty of suspension that may occur when image recognition is continued.
When the image recognition unit 6 determines that the image recognition of the target feature has succeeded, the result is output to the point information generation unit 9.

1-8. Point information generating unit When the point recognition generating unit 9 succeeds in image recognition of the target feature in the image recognition processing by the image recognition unit 6, based on the vehicle position information P and the image recognition result of the target feature, It functions as a point information generation means for generating stop obligation point information s representing a point with a temporary stop obligation. The obligatory stop point information s has identification information (point ID) for identifying each point from other points, position information of the point with the obligation to pause, and attribute information indicating that there is at least a pause obligation Yes. In the present embodiment, the position of the stop line included in the target feature that has been successfully image-recognized is set as the position of the point where the temporary stop is required.
The position of the stop line is determined based on the vehicle position information P and the image recognition result of the stop line. Here, the actual position of the stop line included in the image information G is the position of the vehicle and the imaging area calculated in advance based on the mounting position, mounting angle, and angle of view of the imaging device 18 to the host vehicle. By using the positional relationship, the vehicle position information P can be obtained accurately. The position information thus determined is position (coordinate) information on the map of the representative point of the stop line. In the present embodiment, the representative point is set at the center in the length direction and the width direction of each stop line. Further, the attribute information that the stop line has includes at least “temporary stop obligation” information. In addition to this, information such as “stop line”, “character string in front” may be included.
The stop mandatory point information s generated by the point information generation unit 9 is output to the learning database DB3 and stored in the learning database DB3. When the same target feature is image-recognized a plurality of times by the host vehicle traveling a plurality of times on the same road, a plurality of stop obligation point information s for the same point is generated and stored in the learning database DB3. .

1-9. Estimated position determination unit The estimated position determination unit 10 is generated by image recognition of the same target feature a plurality of times, and is stored in the learning database DB3, based on a plurality of stop obligation point information s about the same point, It functions as an estimated position determination means for determining an estimated position at a point where there is a duty to pause. Here, the estimated position determination unit 10 determines the representative value of the distribution of the points having the suspension obligation based on the distribution of the positions of the points having the suspension obligation represented by the plurality of suspension obligation point information s for the same point. It is determined as an estimated position. In this way, by determining the representative value of the distribution as the estimated position of the point having the suspension obligation, the statistical processing using the plurality of suspension point information s about the same suspension point is appropriately performed. It is possible to determine the estimated position of the point having the obligation to pause. In this embodiment, an average value is used as a representative value of the distribution. That is, the estimated position determination unit 10 calculates an average value by calculation based on a plurality of stop duty point information s for points having the same temporary stop duty stored in the learning database DB3, and calculates the average value by the average value. The required position is determined as the estimated position of the point having the obligation to pause. As an example, a determination method for determining the estimated position of the stop line f4 in the example of FIG. 5 will be described. A plurality of learning values as stop obligation point information s for the stop line f4 are stored in the learning database DB3 as schematically shown in FIG. The estimated position determination unit 10 calculates an average value based on these learning values, determines the result as the estimated position of the stop line f4, and outputs it to the statistical point information generation unit 11.

1-10. Statistical point information generation unit The statistical point information generation unit 11 generates statistical stop duty point information S using the estimated position determined by the estimated position determination unit 10 as the position of a point where there is a temporary stop obligation. It functions as point information generation means. The statistical stop duty point information S includes position information of a point with a temporary stop duty and attribute information indicating that there is at least a temporary stop duty. Here, the position of the point having the duty to pause is the estimated position determined by the estimated position determination unit 10, and the attribute information such as “has a duty to temporarily stop” is the same as that included in the point of obligation to stop stop s.
The statistical stop duty point information S generated in the statistical point information generation unit 11 is output to the stop duty point database DB4 and stored in the stop duty point database DB4. FIG. 9 schematically shows how the statistical stop duty point information S is stored in the stop duty point database DB4. In this way, each time the image recognition of the target feature is successful, the stop duty point information s is stored in the learning database DB3, and accordingly, the new statistical stop point information S is stored in the stop duty point database DB4. Thus, the statistical stop duty point information S possessed by the stop duty point database DB4 is sequentially updated. In the present embodiment, by using the average value as the representative value of the distribution, the statistical point information S is generated only by succeeding in the image recognition of the target feature at least once, and the image recognition of the same target feature is performed. The accuracy can be increased as the number of times increases.

1-11. Navigation Calculation Unit The navigation calculation unit 12 operates according to an application program to execute navigation functions such as display of the vehicle position, calculation of a route from the departure point to the destination, route guidance to the destination, and destination search. Arithmetic processing means. For example, the navigation calculation unit 12 acquires the map information M around the host vehicle from the map database DB1 and displays a map image on the display device 19, and on the map image, based on the host vehicle position information. A process for displaying the self-position mark in a superimposed manner is performed. Further, the navigation calculation unit 12 uses one or both of the display device 19 and the audio output device 20 based on the route from the departure point to the destination and the vehicle position information calculated by a known method. Give guidance. At this time, the navigation computing unit 12 refers to the statistical stop duty point information S stored in the stop duty point database DB4 and provides guidance for the point where the stop duty is required. Therefore, in the present embodiment, the navigation calculation unit 12 functions as a guidance unit that provides guidance for a point where there is a temporary stop obligation based on the vehicle position information and the statistical stop duty point information S. In addition, the navigation calculation unit 12 is connected to various known configurations necessary for the navigation device 1 such as a remote controller and a user interface such as a touch panel provided integrally with the display device 19.

1-12. Procedure of spot information collection processing Next, a procedure of spot information collection processing executed in the spot information collection device 2 including the image recognition device 3 applied to the navigation device 1 according to the present embodiment will be described. FIG. 10 is a flowchart showing a procedure of spot information collection processing including image recognition processing according to the present embodiment. The procedure of the spot information collection process described below is executed by hardware or software (program) or a combination of both constituting each functional unit of the spot information collection apparatus 2 described above. When each function part of the point information collection device 2 is configured by a program, the arithmetic processing device included in the point information collection device 2 includes an image recognition program that constitutes each of the above function units and a point information collection program including the image recognition program. It operates as a computer that executes

First, the shape recognition unit 7 performs image recognition processing for the character string “rare” (step # 01). The shape recognizing unit 7 performs pattern matching between the contour information extracted from the features included in the image information G acquired by the imaging device 18 and the feature amount in the form of the character string “rare”, thereby obtaining image information. It is determined whether or not an image of the character string “rare” is included in G (step # 02). When it is determined that the image of the character string “rare” is not included in the image information G (step # 02: No), the shape recognition unit 7 returns to step # 01 again, and the shape recognizing unit 7 determines the character string “rare”. The image recognition process is performed. On the other hand, when it is determined that the image information G includes an image of the character string “rare” (step # 02: Yes), the adjacency determination unit 8 starts distance measurement (step # 03). Next, the shape recognition unit 7 performs image recognition processing for the stop line (step # 04). Whether the image recognition G includes the image of the stop line by performing pattern matching between the contour information extracted from the feature included in the image information G and the feature amount in the form of the stop line. It is determined whether or not (step # 05). When it is determined that the image information G does not include a stop line image (step # 05: No), the adjacency determination unit 8 determines whether or not the measured distance is equal to or less than a predetermined distance D set in advance. Is determined (step # 06). If it is determined that the measured distance is greater than the predetermined distance D set in advance (step # 06: No), the process ends. On the other hand, when it is determined that the measured distance is equal to or less than the predetermined distance D set in advance (step # 06: Yes), the process returns to step # 04 again, and the shape recognition unit 7 displays the image about the stop line. Perform recognition processing. If it is determined that the image information G includes a stop line image (step # 05: Yes), the target feature, that is, the stop line provided on the road surface, and the stop line are provided adjacent to the target line. It is determined that the image recognition of the combination with the received character string “rare” has succeeded. Next, the point information generation unit 9 generates stop obligation point information s representing a point where there is a temporary stop obligation based on the vehicle position information P and the image recognition result of the target feature (step # 07).
According to such image recognition processing, when image recognition of a combination of a stop line provided on the road surface of a road and a predetermined character string provided adjacent to the stop line is successful, the stop line Since it is determined that the position is a point having a duty to pause, it is possible to extract and recognize only a stop line having a duty to pause. In addition, since the combination of the stop line and the predetermined character string is recognized as an image, the recognition accuracy can be improved as compared with the case where the stop line is recognized alone.

The generated stop duty point information s is output to the learning database DB3 and stored in the learning database DB3 (step # 08). Next, the estimated position determination unit 10 determines the estimated position of the point where there is a suspension obligation based on one or more stop obligation point information s for the same point stored in the learning database DB3 (step #). 09). Next, the statistical point information generating unit 11 generates the statistical stop obligation point information S including position information that uses the estimated position determined by the estimated position determination unit as the position of the point where there is a temporary stop obligation (step #). 10). Finally, the statistical stop duty point information S generated in the statistical point information generation unit 11 is output to the stop duty point database DB4 and stored in the stop duty point database DB4 (step # 11).
According to such point information collection processing, even if there is an error in the recognition position of the point having the suspension obligation indicated by the respective suspension obligation point information s, one or more of the same point each By the statistical processing using the stop duty point information s, it is possible to determine and store the estimated position of the point where the stop duty is appropriately placed. Therefore, highly accurate information can be collected.

[Second Embodiment]
Next, a second embodiment of the present invention will be described.
FIG. 11 is a block diagram illustrating a schematic configuration of the navigation device 1 to which the spot information collection device 2 including the image recognition device 3 according to the present embodiment is applied. Basically, it is almost the same as in the case of the first embodiment, except that a predetermined character string is not specified for the target feature whose image is recognized by the image recognition unit 6. This is different from the first embodiment. Further, the navigation device 1 according to the present embodiment includes a character string information generation unit 13 that generates character string information l for a predetermined character string included in the target feature, and a plurality of character strings for the same predetermined character string. And a statistical character string information generation unit 14 that generates statistical character string information L based on the information l. In addition, a learning database for storing them and a stop obligation point database are provided. Hereinafter, the navigation device 1 according to the present embodiment will be described focusing on differences from the first embodiment. Note that the processing contents of the image recognition program and the spot information collection program according to this embodiment are partially different from those of the first embodiment, but the processing order is the same as that of the first embodiment, so here Description is omitted.

2-1. Learning Database The learning database DB3 is a database that stores the stop duty point information s generated by the point information generation unit 9 and the character string information l generated by the character string information generation unit 13. In this learning database DB3, one or two or more stop obligation point information s for each of a plurality of points having a stop obligation that has been successfully recognized by the image recognition unit 6 and a predetermined character string adjacent thereto One or more character string information 1 is stored. The stop obligation point information s and the character string information l are stored in association with each other. The stop obligation point information s and the character string information l stored in the learning database DB3 will be described later. In the present embodiment, the learning database DB3 is configured to serve as the point information storage means and the character string information storage means in the present invention.

2-2. Stop Obligation Point Database The stop duty point database DB4 stores the statistical stop duty point information S generated by the statistical point information generation unit 11 and the statistical character string information L generated by the statistical character string information generation unit 14. Database. In this stop duty point database DB4, statistical stop duty point information S generated for each of a plurality of points having a temporary stop duty that has been successfully recognized by the image recognition unit 6, and a predetermined character string adjacent thereto Statistical character string information L for is stored. The statistical stop duty point information S and the statistical character string information L are stored in association with each other. The statistical stop duty point information S and the statistical character string information L stored in the stop duty point database DB4 will be described later. In the present embodiment, the stop obligation point database DB4 is configured to serve also as the statistical point information storage means and the statistical character string information storage means in the present invention.

2-3. Image recognition unit The image recognition unit 6 uses a combination of a stop line provided on a road surface of a road and a predetermined character string provided adjacent to the stop line as a target feature, and includes a target site included in the image information G. It functions as an image recognition means for performing image recognition processing of an object. The image recognizing unit 6 includes a shape recognizing unit 7 that performs image recognizing processing on the shape of the target feature provided on the road, and the stop line recognized by the shape recognizing unit 7 and a predetermined character string are adjacent to each other. And an adjacency determination unit 8 that determines whether or not

2-3-1. Shape Recognizing Unit The shape recognizing unit 7 is a shape recognizing unit that performs image recognition processing on the shape of the target feature included in the image information G using the stop line and the predetermined character string provided on the road surface of the road as the target feature. Function. When recognizing the target feature, the shape recognizing unit 7 first recognizes a predetermined character string and then recognizes the stop line.
In the present embodiment, the shape recognizing unit 7 performs image recognition processing of a predetermined character string included in the target feature as a feature (hereinafter referred to as a book) having character characteristics for three characters that do not specify the character content. Image recognition of “unspecified character string” in the specification is performed. Here, the feature having the character feature is a feature having a relatively large number of edges and a complicated shape. When recognizing features having such characteristics, first, binarization processing, edge detection processing, and the like are performed on the image information G, and the features (road markings) included in the image information G are detected. Extract contour information. The shape recognition unit 7 extracts an image of the target feature included in the image information G by performing pattern matching between the extracted contour information of the feature and the feature amount of the target feature. In the present embodiment, since an unspecified character string is set as the predetermined character string included in the target feature, pattern matching is performed between the extracted contour information of the feature and the feature amount of the general character form. Is called. As a result, when the image of the unspecified character string can be extracted from the image information G, the shape recognition unit 7 outputs the result to the adjacency determination unit 8 and then the image of the stop line shape. Recognize. Then, when the stop line image can be extracted from the image information G until the moving distance of the host vehicle exceeds the predetermined distance D, it is determined that the image recognition of the target feature has succeeded. As described above, in the present embodiment, the image recognition unit 6 performs the feature including character characteristics for three characters that do not specify the character content as the image recognition processing of the predetermined character string included in the target feature. Therefore, the image recognition process can be performed even when a part of the character string “Tarema” is fading and disappearing. Moreover, since the content of the character is not specified, the image recognition process can be simplified, and a general-purpose recognition process independent of the character content can be executed. Therefore, it is possible to reduce the processing burden of image recognition.

2-3-2. Adjacency determination unit The adjacency determination unit 8 functions as an adjacency determination unit that determines whether or not the stop line image-recognized by the shape recognition unit 7 is adjacent to a predetermined character string. The adjacency determination unit 8 determines whether or not these are adjacent by determining whether or not the distance between the stop line and the unspecified character string is within a predetermined distance D.
When the shape recognition unit 7 receives an output when the image of the unspecified character string is extracted from the image information G, the adjacency determination unit 8 starts measuring the distance from the position where the unspecified character string is image-recognized. To do. The adjacency determination unit 8 is connected to the distance sensor 17, and the shape recognition unit 7 continues to execute the stop line image recognition process until the moving distance of the host vehicle exceeds the predetermined distance D. When the shape recognition unit 7 has succeeded in image recognition of the stop line before exceeding the predetermined distance D, the image recognition unit 6 is adjacent to the target feature, that is, the stop line provided on the road surface of the road, and the stop line. It is determined that the image recognition of the combination with the predetermined character string provided is successful, and the result is output to the point information generation unit 9.

2-4. Point information generating unit When the point recognition generating unit 9 succeeds in image recognition of the target feature in the image recognition processing by the image recognition unit 6, based on the vehicle position information P and the image recognition result of the target feature, It functions as a point information generation means for generating stop obligation point information s representing a point with a temporary stop obligation. The stop obligation point information s includes identification information (point ID) for identifying each point from other points, position information, and attribute information indicating that there is at least a temporary stop obligation.
The position of the stop line is determined based on the vehicle position information P and the image recognition result of the stop line. Here, the actual position of the stop line included in the image information G is the position of the vehicle and the imaging area calculated in advance based on the mounting position, mounting angle, and angle of view of the imaging device 18 to the host vehicle. By using the positional relationship, the vehicle position information P can be obtained accurately. The position information thus determined is position (coordinate) information on the map of representative points associated with the link k or the node n constituting the road information Ra. In the present embodiment, the representative point is set at the center in the length direction and the width direction of each stop line. Further, the attribute information that the stop line has includes at least “temporary stop obligation” information. In addition to this, for example, information such as “stop line” and “character string in front” may be included.
The stop mandatory point information s generated by the point information generation unit 9 is output to the learning database DB3 and stored in the learning database DB3.

2-5. Character string information generation unit When the image recognition of the target feature is successful, the character string information generation unit 13 is a character having position information and attribute information of a predetermined character string included in the target feature that has been successfully recognized. It functions as character string information generating means for generating the column information l. The character string information 1 includes identification information (character string ID) for identifying each character string from other character strings in addition to the position information and the attribute information.
The position of the representative point of the character string is the position of the vehicle position calculated in advance based on the mounting position, mounting angle, angle of view, and the like of the imaging device 18 to the host vehicle, as is the position of the stop line. By using the relationship, the vehicle position information P can be obtained accurately. The representative point is set at the center in the length direction and width direction of the entire character string. As the attribute information that the character string has, at least “character string” information is included. In addition, for example, information such as “Tarema”, “There is a duty point for temporary suspension” may be included. good.

  The character string information 1 generated by the character string information generating unit 13 is output to the learning database DB3 and stored in the learning database DB3. At this time, the character string information 1 is stored in association with the stop duty point information s generated by the point information generation unit 9 and stored in the learning database DB3. FIG. 12 schematically shows a state in which the character string information 1 and the stop duty point information s having the same association information are stored in the learning database DB3. As described above, in this embodiment, the character string information l is generated in addition to the stop duty point information s, and these are stored in association with each other. A possible point information collecting device can be provided.

2-6. Statistical character string information generation unit The statistical character string information generation unit 14 is provided adjacent to a stop line at a point where there is a duty to pause by performing the same processing as the processing performed by the statistical point information generation unit 11. It functions as a statistical character string information generating means for generating statistical character string information L for a given character string. The statistical character string information L includes position information of a predetermined character string included in the target feature that has been successfully image-recognized and attribute information indicating that there is at least some character string. Here, the position of the predetermined character string is the estimated position determined by the estimated position determining means, and the attribute information such as “character string exists” is the same as the character string information l has.

  The statistical character string information L generated in the statistical character string information generation unit 14 is output to the stop duty point database DB4 and stored in the stop duty point database DB4. At this time, the statistical character string information L is stored in association with the statistical stop duty point information S generated by the statistical point information generation unit 11 and stored in the stop duty point database DB4. In this way, each time the image recognition of the target feature is successfully performed, the stop mandatory point information s and the character string information l are stored in the learning database DB3, and accordingly new statistical stop mandatory point information S and statistical characters are stored. By storing the column information L in the stop duty point database DB4, the statistical stop duty point information S and the statistical character string information L included in the stop duty point database DB4 are sequentially updated. In the present embodiment, the statistical stop duty point information S and the statistical character string information L are treated as the feature information F. Therefore, like the feature information F of other road markings stored in the feature database DB2, these pieces of information can be used for correction of the vehicle position information P, vehicle control, and the like.

[Third embodiment]
Next, a third embodiment of the present invention will be described.
FIG. 13 is a block diagram illustrating a schematic configuration of the navigation device 1 to which the point information collection device 2 including the image recognition device 3 according to the present embodiment is applied. Basically, it is almost the same as in the case of the first embodiment, but the order of the target features to be image-recognized by the image recognition unit 6 is not necessarily in the order of a predetermined character string and stop line. The second embodiment is different from the first embodiment in that it corresponds to the case of the order of the stop line and the predetermined character string. Moreover, the navigation apparatus 1 which concerns on this embodiment differs from the 1st embodiment in the specific content of the stop duty point information s which the point information generation part 9 produces | generates. In addition, the navigation apparatus 1 which concerns on this embodiment is not provided with the estimated position determination part 10, the statistical point information generation part 11, and learning database DB3 in connection with this. Hereinafter, the navigation device 1 according to the present embodiment will be described focusing on differences from the first embodiment.

3-1. Stop Obligation Point Database The stop duty point database DB4 is a database that stores the stop duty point information s generated by the point information generation unit 9. This stop duty point database DB4 stores stop duty point information s generated for each of a plurality of points with a temporary stop duty that have been successfully recognized by the image recognition unit 6. The stop duty point information s stored in the stop duty point database DB4 will be described later. In the present embodiment, this stop obligation point database DB4 corresponds to the point information storage means in the present invention.

3-2. Image recognition unit The image recognition unit 6 uses a combination of a stop line provided on a road surface of a road and a predetermined character string provided adjacent to the stop line as a target feature, and includes a target site included in the image information G. It functions as an image recognition means for performing image recognition processing of an object. The image recognition unit 6 includes a shape recognition unit 7 that performs image recognition processing on target features provided on the road, that is, a stop line and a predetermined character string, a stop line recognized by the shape recognition unit 7, and a predetermined line. And an adjacency determination unit 8 that determines whether or not the character string is adjacent. In the present specification, unless specifically distinguished and used, the image recognition unit 6 means a combination of the shape recognition unit 7 and the adjacent determination unit 8.

3-2-1. Shape Recognizing Unit The shape recognizing unit 7 is a shape recognizing unit that performs image recognition processing on the shape of the target feature included in the image information G using the stop line and the predetermined character string provided on the road surface of the road as the target feature. Function. When recognizing the target feature, the shape recognition unit 7 first recognizes one of the stop line and the predetermined character string, and then recognizes the other of the stop line and the predetermined character string. That is, the shape recognizing unit 7 according to the present embodiment can first recognize a predetermined character string as an image, and then recognize a stop line as an image. Conversely, the shape recognizer 7 first recognizes a stop line as an image, and then recognizes a predetermined line. The character string can be recognized as an image. Therefore, in the present embodiment, the shape recognition unit 7 is configured to be able to recognize not only backward characters but also forward characters when performing image recognition of a predetermined character string. Here, the forward direction character represents a character in a normal direction, and the reverse direction character represents a character in an upside down direction. Since the mode for performing image recognition in the order of the character string and the stop line has already been described in detail in the first embodiment, the emphasis is mainly on the mode for performing image recognition in the order of the stop line and the character string. explain.

  First, the shape recognition unit 7 performs image recognition on the shape of the stop line. When the stop line image is recognized, binarization processing, edge detection processing, and the like are performed on the image information G, and contour information of the feature (road marking) included in the image information G is extracted. The shape recognizing unit 7 extracts a stop line image included in the image information G by performing pattern matching between the extracted contour information of the feature and the feature amount of the form of the stop line that is the target feature. As a result, when the image of the stop line can be extracted from the image information G, the shape recognizing unit 7 outputs the result to the adjacent determining unit 8, and then the predetermined character included in the target feature Recognize row images. Here, as in the first embodiment, an example will be described in which image recognition is performed for the characters “MA” and “RE” in the character string “TOREMA”. In the image recognition of the characters “MA” and “RE”, binarization processing and edge detection processing are performed on the image information G in the same manner as described above, and the features (roads) included in the image information G are processed. The outline information of the sign is extracted. The shape recognizing unit 7 performs pattern matching between the extracted contour information of the feature and the feature quantities of the characters “ma” and “re” as the target feature, thereby performing the character “ma” included in the image information G. "And" re "images are extracted. Then, when images of the characters “MA” and “RE” can be extracted from the image information G until the moving distance of the host vehicle exceeds the predetermined distance D, it is provided on the target feature, that is, the road surface of the road. It is determined that the image recognition of the combination of the stop line and the predetermined character string provided adjacent to the stop line has been successful.

FIG. 14 is an explanatory diagram for explaining an example of image recognition processing performed by the shape recognition unit 7 among the image recognition units 6 included in the spot information collection device 2 according to the present embodiment. Four stop lines f1 to f4 exist in the vicinity of the intersection shown in FIG. Among these, the stop lines that are obligated to temporarily stop are f2 and f4 in which the character sign “Tare” is present on the road surface in front of it. Here, the road where f2 and f4 exist has a relatively narrow road width. Therefore, considering the case where the vehicle travels in the direction of the arrow in the figure and turns right at the intersection from a wide road to a narrow road, the back camera mounted on the own vehicle is as shown in FIG. Image information G is acquired. That is, the image information G acquired by the back camera should first include a stop line as shown in FIG. The shape recognition unit 7 performs binarization processing, edge detection processing, and the like on the image information G, and extracts contour information of the feature (road marking) included in the image information G. The shape recognition unit 7 extracts a stop line image included in the image information G by performing pattern matching between the extracted feature outline information and the feature amount of the stop line form. Note that when the image recognition of the stop line is successful, the shape recognition unit 7 outputs the result to the adjacency determination unit 8. Next, as shown in FIG. 15 (b), the image information G should include the character “TO” in the character string “TOREARE”. In this embodiment, since the characters “ma” and “re” are the target features in the character string “to rare”, the shape recognition unit 7 does not perform image recognition processing on the characters “to”. Next, as shown in FIG. 15C, the image information G should include the character “ma” in the character string “rare”. The shape recognition unit 7 performs binarization processing, edge detection processing, and the like on the image information G in the same manner as the stop line image recognition processing, and the features (road markings) included in the image information G are displayed. Extract contour information. The shape recognition unit 7 extracts an image of the character “MA” included in the image information G by performing pattern matching between the extracted contour information of the feature and the feature amount of the character “MA”. Finally, as shown in FIG. 15D, the image information G should include the character “re” in the character string “to rare”. The shape recognition unit 7 performs binarization processing, edge detection processing, and the like on the image information G in the same manner as the image recognition processing of the stop line and the character “MA”, and the features included in the image information G The contour information of (road marking) is extracted. The shape recognition unit 7 extracts an image of the character “re” included in the image information G by performing pattern matching between the extracted contour information of the feature and the feature amount of the stop line form. Then, when images of the characters “MA” and “RE” can be extracted from the image information G until the moving distance of the host vehicle exceeds the predetermined distance D, it is provided on the target feature, that is, the road surface of the road. It is determined that the image recognition of the combination of the stop line and the predetermined character string provided adjacent to the stop line has been successful.
In this example, the case where image recognition of a target feature is performed on a road having a relatively narrow road width has been described as an example. However, image information can be obtained by setting a wide angle of view of the back camera as the imaging device 18 or the like. If the image information G acquired by the acquisition unit 5 includes a stop line and a predetermined character string in the oncoming lane, it is sufficiently possible to perform image recognition of the target feature even on a wide road. .

  In the present embodiment, the shape recognition unit 7 also determines whether the predetermined character string that has been successfully image-recognized is a forward character or a backward character. This determination result is output to the spot information generation unit 9.

3-2-2. Adjacency determination unit The adjacency determination unit 8 functions as an adjacency determination unit that determines whether or not the stop line image-recognized by the shape recognition unit 7 is adjacent to a predetermined character string. The adjacency determination unit 8 determines whether or not these are adjacent by determining whether or not the distance between the stop line and the character string “to rare” is within a predetermined distance D. Here, the predetermined distance D is set in advance based on a result obtained by checking in advance the distance between the stop line constituting the target feature provided on the actual road surface and the character string “Torare”. For example, a distance that is slightly larger than the maximum distance among the distances between the stop line and the character string “Torare” at a plurality of points that are obliged to be temporarily stopped can be set as the predetermined distance D.
In the present embodiment, when the shape recognition unit 7 receives an output when the shape recognition unit 7 succeeds in the image recognition of the stop line from the image information G, the adjacency determination unit 8 starts from the position where the stop line is image-recognized. Start measuring the distance traveled by the vehicle. The adjacency determination unit 8 is connected to the distance sensor 17, and the shape recognition unit 7 continues to perform image recognition processing of the characters “MA” and “RE” until the moving distance of the host vehicle exceeds the predetermined distance D. . When the shape recognition unit 7 has succeeded in image recognition of the characters “MA” and “RE” before the predetermined distance D is exceeded, the image recognition unit 6 detects the target feature, that is, the stop line provided on the road surface of the road. Then, it is determined that the image recognition of the combination with a predetermined character string provided adjacent to the stop line is successful. On the other hand, if the movement distance of the host vehicle exceeds the predetermined distance D before the shape recognition unit 7 succeeds in image recognition of the characters “MA” and “RE”, a stop line and a predetermined character string are displayed. Since it is not adjacent, it is not determined that the image recognition of the target feature has succeeded.

As described above, when the image recognition of the combination of the stop line provided on the road surface and the predetermined character string provided adjacent to the stop line is successful, the position of the stop line is obliged to pause. By determining that the point is a certain point, it is possible to extract and recognize only the stop line having the obligation of temporary stop. In addition, if the distance between the stop line and the predetermined character string is greater than a certain size, it is not determined that the image of the target feature has been successfully recognized. Therefore, the stop line and some character string have a certain distance. Thus, it is possible to suppress erroneous recognition of a point having a duty of suspension that may occur when image recognition is continued. Further, in the present embodiment, the shape recognition unit 7 first recognizes a character string and then recognizes a stop line, but conversely, first, recognizes a stop line and recognizes the character string. Even when the image is recognized, since it is determined that the image recognition of the target feature has been successful, the stop line for the vehicle traveling in the direction opposite to the traveling direction of the host vehicle and the predetermined line A combination with a character string can also be subjected to image recognition processing as a target feature. Therefore, it is possible to efficiently generate the stop duty point information s.
When the image recognition unit 6 determines that the image recognition of the target feature has succeeded, the result is output to the point information generation unit 9.

3-3. Point information generating unit When the point recognition generating unit 9 succeeds in image recognition of the target feature in the image recognition processing by the image recognition unit 6, based on the vehicle position information P and the image recognition result of the target feature, It functions as a point information generation means for generating stop obligation point information s representing a point with a temporary stop obligation. In the present embodiment, the stop mandatory point information s is either one of links in the identification information (point ID) for identifying each point from other points and the link k where the target feature that has succeeded in image recognition exists. It has attribute information indicating that there is a suspension obligation at the end.
As described above, the map database DB1 is a database in which the map information M divided for each predetermined section is stored. The map information M includes information on a number of nodes n corresponding to intersections and links k corresponding to roads connecting the intersections. In the present embodiment, this map database DB1 functions as “map information storage means” in the present invention. Road information Ra representing a road network is configured by the connection relationship between the node n and the link k.

  The information regarding the stop duty position that the stop duty point information s has is determined based on the vehicle position information P, the map information M, and the image recognition result of the target feature. That is, when it is determined that the image recognition unit 6 has succeeded in image recognition of the target feature, the point information generation unit 9 refers to the vehicle position information P and the map information M, and the link where the target feature exists. k is specified. Specifically, the link k corresponding to the road on which the host vehicle is traveling at the time when the image recognition unit 6 determines that the image recognition of the target feature has succeeded is specified as the link k where the target feature exists. . And the point information production | generation part 9 produces | generates the attribute information showing that there exists a temporary stop obligation at the link end in the link k in which the said target feature exists as stop stop point information s. The link k has two link ends at both ends thereof, but the point information generation unit 9 generates stop obligation point information s indicating that one of these link ends has a temporary stop obligation.

  Here, with reference to FIG. 16 showing the connection relationship between the node n and the link k corresponding to the example shown in FIG. 14, generation of the stop duty point information s in the present embodiment will be described. The broken line portions are nodes n and links k corresponding to intersections and roads not shown in FIG. In FIG. 14, when the own vehicle proceeds in the direction of the arrow in the drawing and turns right at the intersection, the target feature including the stop line f4 is image-recognized. The link k corresponding to the road on which the target feature including the stop line f4 exists is the link k4 in FIG. In addition, in this embodiment, the point information generation unit 9 receives an input from the shape recognition unit 7 and has a duty to pause based on information on the direction of characters when a predetermined character string is image-recognized. Determine the end of the link. Specifically, in the case of the example shown in FIGS. 14 and 15, the shape recognition unit 7 recognizes an image of a predetermined character string as a forward direction character, so that the point information generation unit 9 corresponds to the traveling direction of the host vehicle. It is determined that there is a temporary stop obligation at the link end on the rear side (node n1 side), and such stop duty point information s is generated. On the other hand, in the case of the example shown in FIGS. 5 and 6 in the first embodiment, the shape recognition unit 7 recognizes an image of a predetermined character string as a backward character, so It is determined that there is a temporary stop obligation at the link end on the front side with respect to the traveling direction of the vehicle, and such stop duty point information s is generated.

  The stop duty point information s generated in this way has attribute information indicating that there is a duty to pause at the link end in the link k where the target feature having succeeded in image recognition exists, although not particularly illustrated. The same association information is stored in association with the information about the corresponding link k in the map information M of the map database DB1. Thus, the processing load for generation | occurrence | production of the stop duty point information s can be reduced by producing | generating the stop duty point information s using the information regarding the link k memorize | stored in map database DB1. In addition, the stop obligation point information s is stored in association with the information of the link k included in the map database DB1, so that each link k has information on whether or not there is a temporary stop obligation at the link end as its attribute information. Therefore, it is possible to provide enough for the guidance of the stop duty point by the navigation calculation unit 12 with a small amount of data. As a result, it is possible to reduce the processing load on the navigation calculation unit 12.

3-4. Procedure of spot information collection processing Next, a procedure of spot information collection processing executed in the spot information collection device 2 including the image recognition device 3 applied to the navigation device 1 according to the present embodiment will be described. FIG. 17 is a flowchart illustrating a procedure of spot information collection processing including image recognition processing according to the present embodiment. The procedure of the spot information collection process described below is executed by hardware or software (program) or a combination of both constituting each functional unit of the spot information collection apparatus 2 described above. When each function part of the point information collection device 2 is configured by a program, the arithmetic processing device included in the point information collection device 2 includes an image recognition program that constitutes each of the above function units and a point information collection program including the image recognition program. It operates as a computer that executes

  First, the shape recognition unit 7 performs image recognition processing for either one of the stop line and the character string “rare” (step # 21). The shape recognition unit 7 performs pattern matching between the contour information extracted from the feature included in the image information G acquired by the imaging device 18 and the feature amount in the form of the stop line or the character string “rare”. Then, it is determined whether or not any one of the stop line and the character string “rare” is included in the image information G (step # 22). When it is determined that neither the stop line nor the character string “rare” is included in the image information G (step # 02: No), the process returns to step # 21 again, and the shape recognition unit 7 stops. Image recognition processing is performed for either one of the line and the character string “rare”. On the other hand, when it is determined that the image information G includes any one of the stop line and the character string “rare” (step # 22: Yes), the adjacent determination unit 8 starts distance measurement. (Step # 23). Next, the shape recognition unit 7 performs image recognition processing on the other of the stop line and the character string “rare” (step # 24). The shape recognition unit 7 performs pattern matching between the contour information extracted from the features included in the image information G and the feature quantity of the form of the stop line, so that the stop line and the character string “rare” are included in the image information G. It is determined whether or not the other image is included (step # 25). When it is determined that the other image is not included in the image information G (step # 25: No), the adjacency determination unit 8 determines whether or not the measured distance is equal to or less than a predetermined distance D set in advance. Is determined (step # 26). If it is determined that the measured distance is greater than the preset predetermined distance D (step # 26: No), the process ends. On the other hand, when it is determined that the measured distance is equal to or less than the predetermined distance D set in advance (step # 26: Yes), the shape recognition unit 7 returns to step # 24 again, and the shape recognition unit 7 determines the stop line and the character string. The image recognition process for the other of the “rare” is performed. When it is determined that the image information G includes the other image of the stop line and the character string “rare” (step # 25: Yes), the target feature, that is, the stop provided on the road surface of the road It is determined that the image recognition of the combination of the line and the character string “rare” provided adjacent to the stop line is successful. Next, the point information generation unit 9 determines whether the character string “rare” is in the forward direction or the reverse direction based on the result of image recognition in step # 22 or step # 25 (step # 27). ). Then, based on the vehicle position information P and the image recognition result of the target feature, stop mandatory point information s representing a point where there is a temporary stop obligation is generated (step # 28). At that time, based on the determination result of the direction of the character string “rare”, the link end on the side having the obligation to pause is determined. The generated stop duty point information s is output to the stop duty point database DB4 and stored in the stop duty point database DB4 (step # 29).

  According to such image recognition processing, when image recognition of a combination of a stop line provided on the road surface of a road and a predetermined character string provided adjacent to the stop line is successful, the stop line Since it is determined that the position is a point having a duty to pause, it is possible to extract and recognize only a stop line having a duty to pause. In addition, since the combination of the stop line and the predetermined character string is recognized as an image, the recognition accuracy can be improved as compared with the case where the stop line is recognized alone. Furthermore, as described above, the combination of a stop line for a vehicle traveling in the direction opposite to the traveling direction of the host vehicle and a predetermined character string can also be subjected to image recognition processing as a target feature. The point information s can be generated efficiently.

[Other Embodiments]
(1) In each of the above embodiments, when the stop line is image-recognized after image recognition of a predetermined character string, it is determined that the target feature has been successfully image-recognized, and the stop line is image-recognized. An example has been described in which it is determined that image recognition of a target feature has been successful when a predetermined character string is image-recognized. However, in the image recognition apparatus 3 according to the present invention, the order of image recognition of the characters constituting the stop line and the predetermined character string can be arbitrarily set. That is, when the target feature is a stop line and the characters “MA” and “RE”, the image information G acquired at once includes all of the stop line and the characters “MA” and “RE”. For example, even if image recognition is performed in the order of the character “MA”, the stop line, and the character “RE”, it can be determined that the image recognition of the target feature has been successful.

(2) In the first embodiment described above, an example in which image recognition of characters “MA” and “RE” is performed as image recognition processing of a predetermined character string included in the target feature has been described. However, the predetermined character string recognized by the image recognition apparatus 3 according to the present invention is any one of “Tarema”, “Stop”, “Tomare”, “Stopmare”, or at least one of these character strings. What is necessary is just to perform the image recognition of the character of a part.

(3) In the second embodiment described above, as image recognition processing of a predetermined character string included in the target feature, image recognition of the feature having character characteristics for three characters that do not specify the character content An example of performing is described. However, the image recognition apparatus 3 according to the present invention may be configured to perform image recognition of a feature having a character characteristic of less than 3 characters or more than 4 characters.

(4) In each of the above embodiments, the image information acquisition unit 5, the image recognition unit 6, the vehicle position information acquisition unit 4, the point information generation unit 9, the learning database DB3, the estimated position determination unit 10, and the statistical point information The point information collection device 2 that stores the more accurate estimated position of the target point by statistical processing using the plurality of stop point information s for the same point by providing the generation unit 11 and the stop point point database DB4 The example of was described. However, the point information collection device 2 according to the present invention omits the estimated position determination unit 10, the statistical point information generation unit 11, and the learning database DB3, and requires the stop mandatory point information s obtained at the time of the first image recognition. May be directly recorded in the stop obligation point database DB4. In this way, the entire apparatus can be simplified depending on the accuracy of specifying the position of the target feature. In that case, the stop obligation point database DB4 functions as the point information storage means in the present invention.

(5) In each of the embodiments described above, the example in which the image information G around the host vehicle is acquired using the back camera that captures the road surface behind the host vehicle has been described. However, the image recognition apparatus 3 according to the present invention may be configured to acquire image information G around the host vehicle using a front camera that images the road surface ahead of the host vehicle. In general, the front camera is often provided so as to acquire a relatively wide range of image information G in front of the host vehicle. In this case, the image recognition unit 6 includes a stop line and a predetermined character string. Only by recognizing both of them from one image information G, the image recognition of the combination of the stop line provided on the road surface of the road and a predetermined character string provided adjacent to the stop line was successful. Can be determined. Therefore, the adjacency determination unit 8 that determines whether or not the stop line and the predetermined character string are adjacent to each other can be omitted.

(6) In each of the above-described embodiments, the adjoining determination unit 8 uses a stop line at a plurality of points where there is a pause obligation as the predetermined distance D for determining that the stop line and the predetermined character string are adjacent to each other. An example in which a distance slightly larger than the maximum distance among the distances between the character string and the character string “to rare” is set. However, if it is a distance that can be determined that the paint display of the character string (for example, “towing”, “bus priority”, etc.) at a point where there is no obligation to pause and the stop line that appears after that is not relevant, Such a distance may be set.

(7) In each of the embodiments described above, the position where the adjacency determination unit 8 recognizes an image of the predetermined character string as a processing method for determining whether the stop line and the predetermined character string are adjacent to each other. An example is shown in which the distance traveled by the host vehicle is measured using the distance sensor 17 and compared with a predetermined distance D set in advance. However, for example, the stop line and the position (coordinates) of a predetermined character string are obtained by analyzing GPS signals at the time of image recognition, and the distance between these coordinates is compared with a predetermined distance D set in advance. You may comprise so that determination may be performed.

(8) In each of the above-described embodiments, the estimated position determination unit 10 uses an average value as a representative value of the distribution of the positions of the points having the temporary stop obligations represented by the plurality of stop duty point information s for the same points. An example is shown. However, a mode value or a median value may be used as the representative value. For example, when the mode value is used, the accuracy of the position information included in the statistical stop duty point information S is high. Therefore, according to the point information collection device 2 according to the present invention, a highly reliable stop duty point database. DB4 can be created.

(9) In the second embodiment, the position information included in the character string information 1 is calculated in advance based on the mounting position, mounting angle, angle of view, etc. of the imaging device 18 to the host vehicle. The example which calculated | required based on the own vehicle position information P using the positional relationship of a vehicle position and an imaging region was shown. However, based on the stop line position information included in the stop stop point information s, a position that is separated by a certain distance in a certain direction with the position of the stop line as a reference is determined as position information representing the position of the predetermined character string. Also good. In this case, it is possible to fully collect point information while reducing the burden of calculation processing of the point information generation unit 9.

(10) In the second embodiment, the example in which both the stop duty point information s and the character string information 1 are stored in the learning database DB3 is shown. However, these may be stored in different databases. Similarly, the statistical stop obligation point information S and the statistical character string information L may be stored in separate databases.

(11) In the third embodiment, the point information generation unit 9 receives an input from the shape recognition unit 7 and temporarily receives information on the direction of characters when a predetermined character string is image-recognized. The example of determining the link end on the side having the obligation to stop has been described. However, for example, the link end on the side having the obligation to pause may be determined based on the positional relationship between the predetermined character string recognized by the shape recognition unit 7 and the stop line. For example, the point information generation unit 9 can be configured to generate stop obligation point information s indicating that there is a temporary stop obligation at the link end closer to the stop line than a predetermined character string.

(12) The image recognition device 3, the spot information collection device 2, and the navigation device 1 can be configured by appropriately combining the functional units in the above embodiments. For example, in the image recognition device 3 and the like according to the first embodiment, the shape recognition unit 7 can be configured to perform image recognition in the order of stop lines and character strings as in the third embodiment. Also good.

(13) In each of the above embodiments, the example in which all the configurations of the image recognition device 3, the spot information collection device 2, and the navigation device 1 are loaded on the vehicle has been described. However, the image recognition device 3, the point information collection device 2, and the navigation device 1 according to the present invention may be configured such that some of them are provided outside the vehicle via a communication network. For example, a part or all of each functional unit and each database may be provided in an external server, and image recognition processing, point information collection processing, and navigation calculation processing may be performed via a communication network.

  INDUSTRIAL APPLICABILITY The present invention can be suitably used for an image recognition device that performs image recognition processing of a target feature included in image information around a vehicle acquired by an imaging device or the like mounted on the vehicle. Further, it can be suitably used for a spot information collecting device and a navigation device using the image recognition device.

The block diagram which shows schematic structure of the navigation apparatus which concerns on 1st embodiment of this invention. The figure which shows the example of a structure of the map information memorize | stored in the map database The figure which shows the example of the feature information of the road marking memorize | stored in the feature database The figure which shows an example of the arrangement configuration of the imaging device to the own vehicle Explanatory drawing of the target feature by the image recognition means of the present invention The figure which shows the image information of the surroundings of the own vehicle which an image information acquisition part acquires when passing a temporary stop obligation point The figure which shows an example of the method of measuring the distance between a stop line and a predetermined character string The schematic diagram which shows the stop duty point information stored in the learning database of the navigation apparatus which concerns on 1st embodiment of this invention. The schematic diagram which shows the statistical stop duty point information stored in the stop duty point database of the navigation apparatus which concerns on 1st embodiment of this invention. The flowchart which shows the process order of the point information collection program containing the image recognition program of the navigation apparatus which concerns on 1st embodiment of this invention. The block diagram which shows schematic structure of the navigation apparatus which concerns on 2nd embodiment of this invention. The schematic diagram which shows the stop duty point information and character string information which were stored in the learning database of the navigation apparatus which concerns on 2nd embodiment of this invention. The block diagram which shows schematic structure of the navigation apparatus which concerns on 3rd embodiment of this invention. Explanatory drawing of the target feature by the image recognition means of the present invention The figure which shows the image information of the surroundings of the own vehicle which an image information acquisition part acquires when passing a temporary stop obligation point Explanatory drawing for demonstrating the production | generation of the stop duty point information which concerns on 3rd embodiment of this invention. The flowchart which shows the processing order of the point information collection program containing the image recognition program of the navigation apparatus which concerns on 3rd embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Navigation apparatus 2 Point information collection apparatus 3 Image recognition apparatus 4 Own vehicle position information acquisition part 5 Image information acquisition part 6 Image recognition part 7 Shape recognition part 8 Adjacent determination part 9 Point information generation part 10 Estimated position determination part 11 Statistical point Information generator 12 Navigation calculator 13 Character string information generator 14 Statistical character string information generator DB1 Map database DB3 Learning database DB4 Stop obligation point database s Stop obligation point information S Statistical stop obligation point information l Character string information L Statistical character string information G Image information P Own vehicle position information M Map information n Node k Link

Claims (16)

Own vehicle position information acquisition means for acquiring own vehicle position information representing the current position of the own vehicle;
Image information acquisition means for acquiring image information around the host vehicle;
An image for performing image recognition processing of the target feature included in the image information with a combination of a stop line provided on the road surface and a predetermined character string provided adjacent to the stop line as a target feature Recognition means;
A point information generating means for generating stop duty point information representing a point where there is a duty to temporarily stop based on the vehicle position information and the image recognition result of the target feature when image recognition of the target feature is successful; ,
An image recognition apparatus comprising:   The image recognition device according to claim 1, wherein the image recognition unit determines that the image recognition of the target feature has succeeded when the stop line is image recognized after image recognition of the predetermined character string.   3. The image recognition according to claim 1, wherein the image recognition unit determines that the image recognition of the target feature is successful when the predetermined character string is image recognized after image recognition of the stop line. apparatus.   The image recognition means, as an image recognition process of the predetermined character string included in the target feature, in a character string of “to rare” represented by any one of hiragana, katakana, kanji, and combinations thereof The image recognition apparatus according to claim 1, wherein image recognition of at least some characters is performed.   The image recognition means performs image recognition of a feature having character characteristics for three characters that do not specify character content as image recognition processing of the predetermined character string included in the target feature. 4. The image recognition device according to claim 1.   The said stop duty point information has the attribute information which shows the positional information on the said stop line contained in the said target feature which succeeded in image recognition, and having at least a temporary stop duty. The image recognition apparatus described.   The image recognition apparatus according to claim 6, wherein the position information of the stop line is coordinate information on a map of a representative point of the stop line. Comprising map information storage means for storing map information including road network information representing connection relations of roads by nodes corresponding to intersections and links corresponding to roads connecting the intersections;
The said stop duty point information has attribute information showing that there exists a temporary stop duty in any one link end in the link in which the said target feature which succeeded in image recognition exists. The image recognition apparatus described in 1.   A point information collection device comprising: the image recognition device according to claim 1; and a point information storage unit that stores the stop duty point information generated by the point information generation unit. An image recognition device according to any one of claims 1 to 7,
Point information storage means for storing the stop duty point information generated by the point information generation means;
Estimating a point having the duty to stop temporarily based on a plurality of points of duty to stop for the same point, which are generated by image recognition of the same target feature multiple times and stored in the point information storage means Estimated position determining means for determining a position;
Statistical point information generating means for generating statistical stop duty point information that uses the estimated position determined by the estimated position determination means as a position of the point where the temporary stop duty is present;
And a statistical point information storage unit that stores the statistical stop obligation point information generated by the statistical point information generation unit.   The said estimated position determination means determines the representative value of the said distribution as said estimated position based on the distribution of the position of the point with the said temporary stop obligation which the said several said stop mandatory point information about the same point represents. The point information collection device described in 1.   Character string information generating means for generating character string information having position information and attribute information of the predetermined character string included in the target feature that has been successfully recognized when the image recognition of the target feature is successful The image recognition apparatus according to any one of claims 1 to 7, further comprising:   A point information collection device comprising: the image recognition device according to claim 12; and character string information storage means for storing the character string information generated by the character string information generation means in association with the stop duty point information. .   A navigation apparatus comprising: the spot information collecting apparatus according to claim 9 or 13; and guidance means for guiding a spot having the duty to stop temporarily based on the own vehicle position information and the stop duty spot information. .   12. A spot information collection device according to claim 10 or 11, and a guidance means for guiding a spot having a duty to stop temporarily based on the own vehicle position information and the statistical stop duty spot information. Navigation device. Own vehicle position information acquisition step for acquiring own vehicle position information indicating the current position of the own vehicle;
An image information acquisition step of acquiring image information around the host vehicle;
An image for performing image recognition processing of the target feature included in the image information with a combination of a stop line provided on the road surface and a predetermined character string provided adjacent to the stop line as a target feature A recognition step;
A point information generating step for generating, when the image recognition of the target feature is successful, based on the image recognition result of the target feature, stop obligation point information indicating a point where there is a duty to temporarily stop;
Recognition program that causes a computer to execute