JP2009229180A - Navigation device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a navigation device which enables a user to easily designate a building, improves accuracy in identifying the building, and reduces a burden when the user acquires information. <P>SOLUTION: The navigation device includes a positioning section 2, a map building database 1, a model construction section 6, an image acquiring section 3, an attention direction receiving section 4, a shape recognition section 5, a building identifying section 7, a navigation information acquiring section 8, and a displaying section 9. The model construction section 6 constructs a solid model of a current location periphery. The image acquiring section 3 acquires photographed images. The attention direction receiving section 4 determines the direction to the building designated in a displaying device which displays photographed images. The shape recognition section 5 extracts shape character information of surrounding buildings. The building identifying section 7 identifies the designated building using the solid model, the direction, and the shape character information. The navigation information acquiring section 8 acquires navigation information of the building from the map building database 1, and the displaying section 9 outputs navigation information with at least one of the voice and image. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a vehicle-mounted navigation device, and more particularly to a navigation device for obtaining information related to buildings around a vehicle.

  In a conventional navigation device, a user inputs appropriate search conditions to identify a building and acquire the information. In addition, in a conventional car navigation system, a live-action video from an in-vehicle camera is displayed on the screen of the navigation device, a building is specified by touching a building that is visible on the screen with a finger, etc. The position information of the object is obtained using the coordinate information of the obtained feature point and the corresponding point and the position information of the camera obtained from GPS (Global Positioning System). The building information was listed and displayed sequentially, and the user selected it. (For example, refer to Patent Document 1).

JP 2004-138556 A (page 4, FIG. 1)

  In conventional navigation devices, it is necessary to input appropriate search conditions, or because the accuracy of building identification is not sufficient, the user has to select from a plurality of buildings. There was a point.

  The present invention has been made to solve the above-described problems. The user can easily specify a building, improve the accuracy of specifying the building, and reduce the burden when the user obtains information. The object is to obtain a navigation device.

  A navigation device according to the present invention includes a positioning unit, a map building database, a model building unit, an image acquisition unit, a direction-of-interest receiving unit, a shape recognition unit, a building identification unit, and a guidance information acquisition unit. And a display unit. The model construction unit constructs a three-dimensional model around the current position from the current position information from the positioning unit, the map from the map building database, and the building information. The image acquisition unit captures the periphery of the current position and acquires a real image. The attention direction receiving unit receives position information on the screen of the designated building on the display device that displays the photographed image from the image acquisition unit on the screen, and obtains the direction to the designated building with respect to the current position. The shape recognizing unit extracts the shape feature information of the surrounding building from the photographed image from the image acquiring unit. The building identification unit uses the 3D model from the model building unit, the direction from the target direction receiving unit, and the geometric feature information from the shape recognition unit to determine which building in the 3D model is on the display screen. Identify whether it is a thing. The guide information acquisition unit acquires the guide information of the building specified by the building specifying unit from the map building database, and the display unit outputs the guide information by at least one of audio or video.

  Since the navigation device according to the present invention includes the image acquisition unit and the attention direction receiving unit, the building can be easily specified on the display screen. In addition, since it has a positioning unit, a map building database, a model building unit, a shape recognition unit, and a building specifying unit, a 3D model around the current position is built based on the map building database, and this is actually captured Since the captured real image is collated, the accuracy of identifying the building is improved. As a result, building information can be easily acquired, and the burden on the user can be reduced.

Embodiment 1 FIG.
FIG. 1 is a system configuration diagram of a navigation apparatus according to Embodiment 1 for carrying out the present invention. In the figure, the map building database 1 stores a map and three-dimensional shape information of the building and guidance information. The three-dimensional shape information is information for constructing a three-dimensional model of a building in association with map information, and includes, for example, the position of the building, the site shape, and the number of floors. Instead of site shape and floor number, you may have a more detailed solid model. Guidance information is information for presenting a user with guidance regarding a building, such as a name and detailed information. The information listed here may be acquired by communicating with an external network. The building here is not limited to an artifact, but may be a park, a mountain, a river, a coast, or the like.

  The positioning unit 2 obtains the current position and posture of the vehicle. The position and orientation are calculated using measured values of a sensor group such as GPS, a vehicle speed sensor, and a gyro sensor, map matching, and the like, as is done in a conventional car navigation device.

  The image acquisition unit 3 captures an image of the vicinity of the current position and acquires a live-action image of the periphery. Here, as the photographed image, for example, an image in front of the own vehicle is acquired by an in-vehicle camera.

  The attention direction receiving unit 4 obtains the direction to the building designated by the user on the display screen and the attention direction angle. Here, the attention direction angle is expressed as an angle formed by the direction to the designated building and the direction in which the vehicle is facing, based on the vehicle position. Specifically, a peripheral captured image (actual image) acquired by the image acquisition unit 3 is displayed on a display device such as a display, and position information of a point of interest of a building that the user has noted and specified on the display screen is displayed. Receive. Coordinate conversion is performed using the position of the point of interest on the received real image, the position and orientation at which the camera is installed, and camera parameters, and the angle between the direction of the point of interest and the direction in which the vehicle is facing is calculated. As a method of coordinate conversion, for example, “Basics and Applications of 3D CG” ISBN 7819-1080-7, P.I. 22-24, P.I. The inverse transformation of the perspective transformation described in 59-66 is used. The position of the point of interest on the image is specified by touching a display having a touch panel function or operating a cursor on the display with a remote controller, for example.

  The shape recognizing unit 5 extracts the shape feature information of the surrounding building using the photographed image acquired by the image acquiring unit 3. Here, the surrounding building is a building that is actually seen from the vehicle. As the shape feature information of the surrounding building, for example, feature points, edges, planes, etc. of the photographed image are used. Examples of a method for extracting such shape feature information from an image include “Introduction to Image Processing” ISBN 4-7898-1834-9, P.I. The method described in 50-62 is used.

  The model construction unit 6 first sets a peripheral region using the own vehicle position / posture acquired from the positioning unit 2. Here, the peripheral area is represented by a coordinate system in which the position of the vehicle is the origin and the direction of the vehicle posture is the depth direction (= direction in which the vehicle is facing). The range of the peripheral region may be set in advance so as to take a certain range, for example, 50 [m] in the left-right direction of the own vehicle and 500 [m] in the forward direction. The range of the surrounding area may be dynamically determined in consideration of the type of road that is running, the number of buildings in the vicinity, and the like. In that case, for example, if the vehicle is traveling on an expressway, the distance ahead may be set to 10 [km], or the range may be expanded until the number of surrounding tourist facilities reaches 20. Next, the three-dimensional shape information of the building located in the peripheral region is acquired from the map building database 1, and a three-dimensional model is constructed using this information.

  The building specifying unit 7 receives the direction to the designated building and the direction angle of interest received by the attention direction receiving unit 4, the shape feature information of the surrounding building recognized by the shape recognition unit 5, and the model construction unit 6. The target building is identified using the three-dimensional model of the constructed peripheral area.

  The guide information acquisition unit 8 acquires guide information about the noted building specified by the building specifying unit 7 from the map building database 1. As guide information to acquire, there exists a name of a building, for example. If the building is for sightseeing, the construction time, historical background, sightseeing information, etc. may be acquired. If the building is for commercial use, business hours, sale information, etc. may be acquired. If the building is a station or an airport, a timetable or vacancy information may be displayed.

  The display unit 9 outputs the guidance information acquired by the guidance information acquisition unit 8 to the speaker by voice or displays it on the display as a video. At this time, building guidance information may be superimposed on the actual image acquired by the image acquisition unit 3.

  FIG. 2 is an example of a hardware configuration for realizing the first embodiment of the present invention. The map building database 1 is realized by storing map data and building data in the storage device 102 in advance. Alternatively, the map building database 1 acquires map data and building data from an external storage medium or network via the external device interface 104 and temporarily stores them in the storage device 102 via the microcomputer interface 103. It will be realized.

  The positioning unit 2 uses the microcomputer interface 103 to mediate the speed received by the vehicle interface 105 from the vehicle, the estimated vehicle position calculated by the GPS receiver 106 receiving radio waves from the GPS satellite, and the angular velocity of the vehicle detected by the gyro sensor 107. Using the map data stored in the storage device 102 in the map building database 1 and the above-described speed, estimated vehicle position, and angular velocity, the CPU 101 specifies the position and orientation of the vehicle. This is realized by storing in the storage device 102.

  The image acquisition unit 3 is realized by storing an image captured by the camera 111 in the storage device 102 via the external device interface 104 and the microcomputer interface 103.

  The attention direction receiving unit 4 displays the image stored in the storage device 102 in the image acquisition unit 3 on the video output device 108 and the user received by the HMI (Human Machine Interface) device 110. Is input to the storage device 102 via the microcomputer interface 103, and the CPU 101 calculates the direction angle of interest using this input information and stores it in the storage device 102.

  The shape recognition unit 5 is realized by the CPU 103 using the image stored in the storage device 102 in the image acquisition unit 3 to extract the shape feature information and store it in the storage device 102.

  The model construction unit 6 uses the own vehicle position and orientation stored in the storage device 102 in the positioning unit 2, and the map data and building data stored in the storage device 102 in the map building database 1, so that the CPU 103 generates a three-dimensional model. This is realized by constructing and storing in the storage device 102.

  The building specifying unit 7 stores the attention direction angle stored in the storage device 102 in the attention direction receiving unit 4, the geometric feature information stored in the storage device 102 in the shape recognition unit 5, and the storage unit 102 in the model construction unit 6. It implement | achieves because CPU103 specifies a building and stores it in the memory | storage device 102 using the stored solid model.

  In the guide information acquisition unit 8, the CPU 103 selects the guide information corresponding to the building stored in the storage device 102 in the building specifying unit 7 from the building data stored in the storage device 102 in the map building database 1. This is realized by storing in another area of the storage device 102.

  The display unit 9 uses the guidance information stored in the storage device 102 in the guidance information acquisition unit 8 and the guidance video and guidance voice created by the CPU 103, and the microcomputer interface 103 mediates the video output device 108 and the acoustic output device 109. This is realized by outputting to.

  Next, the building specifying method of the navigation apparatus configured as described above and the operation of the building specifying unit 7 will be described. In order to specify a building, first, an on-vehicle camera captures an image of the periphery of the vehicle, shows it on a display device, and the user designates a certain building of interest in the display image. Next, the building specifying unit 7 is a real two-dimensional image of the building specified by the user in the real space, the shape feature information extracted by the shape recognition unit 5 based on the two-dimensional image, and the map building Compare the virtual two-dimensional image of each building in the three-dimensional model (virtual space) created by the model building unit 6 from the object database 1 and select the building in the virtual space closest to the real image, The building is identified as the building designated by the user. The operation will be described with reference to the flowchart shown in FIG.

  First, in step 71, it is determined whether or not evaluation values have been calculated for all buildings in the surrounding area in the three-dimensional model. The evaluation value here is a numerical value of the possibility that the building is designated by the user. If there is still a building for which an evaluation value has not been calculated, one of those buildings is selected and the process proceeds to step 72. If the evaluation values of all buildings have been calculated, the process proceeds to step 74.

  In step 72, a projection image (virtual two-dimensional image) based on the building selected in step 71 is created (see FIGS. 4 and 5). Here, the projection image is created based on a three-dimensional model, which is an image that can be seen from the vehicle position when the selected building exists at the direction angle of interest. That is, the image is created so that the attention direction angle in the real space is the same as the direction angle of the selected building in the virtual space. Alternatively, it is also an image created by moving the entire three-dimensional model so that the direction to the building designated on the live-action screen is the same as the direction to the selected building in the three-dimensional model. Since the selected building has a width, the direction to the selected building is, for example, the center of the selected building from the origin, or the selected building when moving the entire 3D model to create an image. There is a method to make the place where the amount of movement is the smallest among them.

  The projection image is created as follows, for example. Since the origin is the vehicle position in the stereo model, the projection center of the projection image is set as the origin of the stereo model. Next, the projection direction of the projection image is determined so that the direction angle of interest matches the direction angle of the selected building. That is, the angle (= attention direction angle) formed by the direction from the own vehicle position in the real space to the designated building (= attention direction vector) and the direction in which the vehicle is facing (= three-dimensional model) in the virtual space The projection direction is determined so as to coincide with the angle formed by the direction from the origin to the selected building and the projection direction (= the direction angle of the selected building). The perspective transformation is performed using the projection center and the projection direction thus determined. A transformation using camera parameters is applied to an image obtained by perspective transformation and used as a projection image.

  In step 72a, the visibility of the building selected in step 71 is calculated using the projection image created in step 72. Here, the visibility is a parameter whose value increases as the size of the selected building that can be seen from the position and orientation of the host vehicle increases. For the visibility, for example, the area of the selected building in the projection image is used.

  In step 72b, the degree of direction coincidence of the building selected in step 71 is calculated using the projection image created in step 72. Here, the degree of direction coincidence is a parameter whose value increases as the direction angle at which the selected building should be seen from the position and orientation of the vehicle and the attention direction angle are closer. Alternatively, it is calculated from the movement angle obtained by moving the entire three-dimensional model when creating the projection image. As the degree of direction coincidence, for example, an inner product of a vector whose direction coincides with the depth direction (= direction in which the vehicle faces) of the three-dimensional model and a vector whose direction coincides with the projection direction in step 72 is used. However, the length of the vector used here is normalized.

  In step 72c, the shape similarity of the building selected in step 71 is calculated using the projection image created in step 72. Here, the shape similarity is a parameter whose value increases as the shape feature information of the selected building in the scenery that should be seen from the position and orientation of the host vehicle is similar to the actually captured image. For the shape similarity, for example, the shape feature information extracted from the projection image (virtual two-dimensional image) and the shape feature information extracted by the shape recognition unit 5 from the actual two-dimensional image are collated and compared. When an edge is used as the shape feature information, for each edge extracted by the shape recognition unit 5, the nearest edge on the projection image is searched and the distance is obtained. The sum of the distances obtained for each edge is obtained, and the reciprocal of the value divided by the number of edges may be used as the shape similarity. The edges used here may be limited to those in the vertical direction, and noise due to unnecessary edges may be removed. The calculation may be performed in consideration of the positional relationship between the attention direction point (the point on the image corresponding to the attention direction angle) and each edge.

  In step 73, the evaluation value of the building selected in step 71 is calculated using each parameter calculated in steps 72a to 72c. As the evaluation value, for example, the sum of all parameters is used. Here, a weighting coefficient for each parameter may be set in accordance with the user's personality and driving situation, and the sum of the weighted parameters may be used as the evaluation value. For example, when a user who tends to focus on a building that looks large is operated, a weighting factor for visibility is set to a large value. When GPS satellite radio waves cannot be received and positioning accuracy is low, a weight for direction matching is set. Suppose that the coefficient is reduced. Here, the weight setting according to the user's personality may be performed by investigating the usage status of a plurality of users and setting an average setting as an initial setting, or by analyzing the usage status of the user actually used. The optimum setting may be obtained. Also, the evaluation value calculation formula need not be a linear combination of the parameters. For example, since the influence of visibility is considered to be small when the degree of direction coincidence is low, the sum of the product of these two parameters and the shape similarity may be used as the evaluation value.

  In step 74, the evaluation values of the buildings obtained in steps 71 to 73 are compared, and the building having the highest evaluation value is specified as the target building designated by the user. If there is no significant difference in the evaluation values for multiple buildings with higher evaluation values here, use the distance from the vehicle position of those buildings, the type of building facilities, or the name of the building. Specific. The facility type is assumed to be stored in the map building database 1 in association with each building. Convenience stores, supermarkets, and other facility types that users would not normally pay attention to should be set to a low priority, and tourist facilities, stations, and other facility types that users are likely to focus on should be set to a high priority. . The candidate buildings may be compared according to this setting. At this time, the priority set for the facility type may be adjusted according to the traveling state. For example, the priority of the gas station is increased when the gasoline is low, and the priority of the restaurant is decreased immediately after stopping at the restaurant. In accordance with the priority set in this way, a building belonging to a facility type that is considered not to be noticed by the user is excluded from the candidates for the building of interest. Further, when using the name recognition, for example, the name of the building is searched on the external network, and the number of hits may be used as an index.

  FIG. 6 shows an outline of an example of the parameter calculation described above. The photographed image in FIG. 6 is an example of a simplified photograph of the photographed image acquired by the image processing unit 3. The shape feature information is a result of recognizing the shape feature information from an example of a photographed image. Circles in the figure indicate attention direction points (points on the image corresponding to attention direction angles). The planar map is a plan view of the exemplified solid model from above. A square in the map indicates a building, a triangle indicates a vehicle position, and a broken line extending from the triangle indicates a direction vector of interest from the vehicle position (origin). The projection images a to c are projection images obtained by moving the entire three-dimensional model so that the directions from the origin to the buildings corresponding to A to C on the planar map (selected buildings) and the attention direction vector overlap each other. Alternatively, it is an image created by setting the projection direction so that the attention direction angle in the real space and the direction angle of the building (selected building) corresponding to A to C in the virtual space are the same.

  In this example, the building designated by the user is B, but in the projection image with the depth direction as the projection direction, the attention direction vector and the building A overlap. In particular, when the buildings A and B are only floor information and there is no accurate height information, the accuracy of the building data is low. If it is low, the height information and position information of the projection images of the buildings A and B become ambiguous, and the building may be erroneously specified only by matching the three-dimensional model with the direction angle of interest. For this reason, it is necessary to consider the shape feature information extracted from the actual photographed image. Even if the height information and the position information are ambiguous, for example, if the distance from the attention direction point to the left and right vertical edges is used as the shape feature information, the building A and B can be distinguished because the width of the building is different. . As a result, the projection image a is determined to have a high degree of direction coincidence because the depth direction of the stereo model and the projection direction coincide with each other, but the shape similarity is medium. Although the direction coincidence degree of the projected image b is slightly lower than a, it is determined that the shape similarity is higher than a. It is determined that the direction coincidence of the projected image c is extremely low and the shape similarity is also low. These determination results are digitized and the building is specified as a parameter of the evaluation function.

  According to the navigation device configured as described above, the image acquisition unit 3 and the attention direction receiving unit 4 are provided, so that the building on which the user pays attention can be easily specified on the display screen. In addition, because it has a positioning unit, a map building database, a model building unit, and a building specifying unit, it builds a three-dimensional model around the current position and collates this with a live-action image. improves. Furthermore, by providing the shape recognizing unit 5, even if the information regarding the three-dimensional shape of the map and the building is not sufficient or the position and orientation of the own vehicle cannot be measured with high accuracy, it is extracted from the actual photographed image. Since the geometric feature information is collated, it is possible to improve the identification accuracy of the building. As a result, building information can be easily acquired, and the burden on the user can be reduced.

Embodiment 2. FIG.
In Embodiment 2, the accuracy of the building is not sufficient, such as when the current position accuracy by GPS is not sufficient due to the valley of the building, or when the building of the map building database 1 is only the floor information and there is no accurate height information. The method of identifying a building when there is not enough or there is an error. FIG. 7 is a system configuration diagram showing the navigation apparatus in the second embodiment for carrying out the present invention. In the figure, the map building database 1, the positioning unit 2, the image acquisition unit 3, the attention direction receiving unit 4, the shape recognition unit 5, the guidance information acquisition unit 8 and the display unit 9 have the same functions as those of the first embodiment. To do.

  The traveling state acquisition unit 10 receives the traveling state of the host vehicle related to positioning accuracy from the vehicle and various sensors. Examples of the running situation include how to turn the steering wheel, the strength of the brake and accelerator, the reception situation of GPS radio waves, and the vibration of the vehicle.

  The model construction unit 6a constructs a three-dimensional model of the vehicle surrounding area by the same method as the model construction unit 6 of the first embodiment. Then, an error range is added to the three-dimensional model using the accuracy of the three-dimensional shape information of the building possessed by the map building database and the traveling state of the host vehicle acquired by the traveling state acquisition unit 10. The error range is set by the following method. The building sets an error range of the model shape in accordance with the accuracy and detail level of the solid shape information. For example, when the height information has only the floor number, the height range corresponding to the floor number is set. For the origin position and the depth direction, an error range is set according to the positioning accuracy. For example, if the GPS radio wave reception condition is poor, the estimated value of the position of the own vehicle is likely to shift, so the error range of the origin position is set large. In addition, since the estimated value of the vehicle posture is likely to be shifted at the moment when the steering wheel is suddenly turned, the error range in the depth direction is set to be large.

  A hardware configuration for realizing the second embodiment of the present invention will be described with reference to FIG. The map building database 1, the positioning unit 2, the image acquisition unit 3, the attention direction receiving unit 4, the shape recognition unit 5, the model construction unit 6a, the building identification unit 7a, the guidance information acquisition unit 8, and the display unit 9 are embodiments. 1 is realized with the same hardware configuration as in FIG.

  The travel status acquisition unit 10 includes a travel status received by the vehicle interface 105 from the vehicle, a vehicle position calculated by the GPS receiver 106 receiving a radio wave from a GPS satellite, and a vehicle angular velocity detected by the gyro sensor 107. This is realized by mediating and storing in the storage device 102.

  The operation of the building measuring unit 7a of the navigation device configured as described above will be described with reference to the flowchart shown in FIG. The building specifying unit 7a specifies the building of interest by the same method as the building specifying unit 7 in the first embodiment, but considers the error range added to the three-dimensional model at that time. Steps 71, 72, 72 a to 72 c, 73, 74 perform the same operations as the steps with the same numbers in the first embodiment.

  In step 71a, it is determined whether or not the evaluation value has been calculated under all conditions in the error range. For example, when a height range is set as the error range, another value within the error range is applied as the height of each building of the stereo model, and the stereo model is deformed. In order to recalculate the evaluation values of all the buildings on the three-dimensional model thus deformed, one building is selected and the process proceeds to step 72. Here, if all the values in the range are applied without using the value once applied, the process proceeds to step 74. If a plurality of parameters are set as the error range, all combinations thereof are applied.

  In step 73, the evaluation value of the building selected in step 71 is calculated using the same method as in step 73 in the first embodiment. However, at that time, if the evaluation value of the selected building has already been calculated under another condition, the larger evaluation value is adopted. In addition, the condition that gave the evaluation value of the adopted one is recorded.

  In step 74, the designated building is specified using the same method as in step 74 in the first embodiment. Here, when the origin position and the depth direction are different from the vehicle position / posture obtained by the positioning unit 2 under the condition where the evaluation value of the specified building is given, the positioning accuracy is improved by feeding back these values to the positioning unit 2. It may be improved.

  Here, the traveling state acquisition unit 10 obtains a positioning error of the vehicle, and the model construction unit 6a adds an error in the origin position and the depth direction to the three-dimensional model. However, even without the traveling state acquisition unit 10, the model construction unit 6a adds an error range to the building of the three-dimensional model, and the building identification unit 7a can identify the building.

  According to the navigation device configured in this manner, the model construction unit 6a constructs a three-dimensional model in consideration of errors, thereby increasing the accuracy of building identification even when there is an error in the map information and the building information. be able to. In addition, since the travel condition acquisition unit 10 is provided, a three-dimensional model to which an error is added is constructed in consideration of the error of the own vehicle position and orientation, and thus the accuracy of specifying the building can be further increased.

Embodiment 3 FIG.
Here, a method for improving the specified accuracy of a designated building in consideration of attribute similarity such as the color and pattern of the building will be described. FIG. 9 is a system configuration diagram showing a navigation apparatus according to Embodiment 3 of the present invention. The map building database 1a, the positioning unit 2, the image acquisition unit 3, the attention direction receiving unit 4, the shape recognition unit 5, the guidance information acquisition unit 8 and the display unit 9 in FIG. 9 have the same functions as those of the first embodiment. do.

  The model construction unit 6b constructs a three-dimensional model of the vehicle surrounding area by the same method as the model construction unit 6 in the first embodiment. And the attribute information which the map building database 1a has is linked | related with each point of the constructed stereo model. The attribute information of the building is, for example, color, texture, material, and the like, and it is assumed that the map building database 1a has such information.

  The attribute recognizing unit 11 extracts attribute information of the surrounding building using the photographed image acquired by the image acquiring unit 3. The attribute information extracted here is the same as that possessed by the map building database 1a. The attribute of the building is obtained by extracting a plane from the image acquired by the image acquisition unit 3 using a luminance value or the like, and further performing texture analysis or the like on the extracted plane. As methods of plane extraction and texture analysis, for example, “Introduction to Image Processing” ISBN 4-7898-1834-9, P.I. The method described in 50-62 is used.

  A hardware configuration for realizing the third embodiment of the present invention will be described with reference to FIG. The map building database 1a, the positioning unit 2, the image acquisition unit 3, the attention direction receiving unit 4, the shape recognition unit 5, the model construction unit 6b, the building identification unit 7b, the guidance information acquisition unit 8, and the display unit 9 are the embodiments. 1 is realized with the same hardware configuration as in FIG. The attribute recognition unit 11 is realized by the CPU 103 using the image stored in the storage device 102 in the image acquisition unit 3 to extract attribute information and store it in the storage device 102.

  The operation of the building measuring unit 7b of the navigation apparatus configured as described above will be described with reference to the flowchart shown in FIG. The building specifying unit 7b specifies the building of interest by the same method as the building specifying unit 7 in the first embodiment, but considers the attribute information extracted by the attribute recognizing unit 11 at that time. Steps 71, 72, 72 a to 72 c, 73, 74 operate in the same manner as the steps with the same numbers in the first embodiment.

  In step 72d, the attribute similarity of the building selected in step 71 is calculated using the projection image created in step 72. Here, the attribute similarity is a parameter whose value increases as the attribute information of the selected building in the scenery to be seen from the position and orientation of the host vehicle is similar to the actually captured image. For the attribute similarity, for example, the attribute information extracted from the projection image of the selected building in the virtual space is compared with the attribute information of the designated building in the real space extracted by the attribute recognition unit 11. When color is used as the attribute information, the difference between the luminance value of each pixel in the live-action image extracted by the attribute recognition unit 11 and the luminance value of the corresponding pixel in the projection image is obtained. What is necessary is just to obtain | require the sum total of the difference of the luminance value calculated | required about each pixel, and let the reciprocal number of the value divided by the pixel number be attribute similarity.

  In addition, when the color of a building changes with conditions, such as dusk, it is good to use a color difference, for example. As a formula for obtaining the color difference, for example, a color difference formula of CIE 1976 L * a * b * is used. Compare the value obtained by applying this formula to the adjacent building in the image with the value obtained by applying this formula to the adjacent building in the 3D model. It is determined that the degree is high. Moreover, you may make it obtain | require the color difference with what averaged the color of the whole image instead of the color difference with an adjacent building.

  In step 74, the designated building is specified using the same method as in step 74 in the first embodiment.

  According to the navigation device configured in this manner, the building identifying unit 7b can identify the building by the evaluation function considering the attribute information by providing the attribute recognition unit 11 and the three-dimensional model associated with the attribute information. Even when there is an error in the vehicle position and orientation, the map information, and the building information, the building identification accuracy can be increased. Moreover, even if the attribute information of the building changes depending on the situation, the building can be specified appropriately.

Embodiment 4 FIG.
Here, a technique for improving the accuracy of the three-dimensional model by recognizing the undulations of the ground and improving the specified accuracy of the designated building will be described. FIG. 11 is a system configuration diagram showing a navigation apparatus according to Embodiment 4 of the present invention. In this figure, the map building database 1, positioning unit 2, image acquisition unit 3, attention direction receiving unit 4, shape recognition unit 5, building identification unit 7, guidance information acquisition unit 8 and display unit 9 Works the same as Form 1. In FIG. 11, if the map building database 1 does not have detailed elevation information of the map, the three-dimensional model constructed by the model construction unit 6c becomes inaccurate, and the building identification accuracy by the building identification unit 7 is low. Become.

  The undulation recognition unit 12 recognizes the undulation of the ground around the own vehicle from the photographed image acquired from the image acquisition unit 3. For the undulations around the own vehicle, for example, the vanishing point is obtained by extracting the road surface from the photographed image, and the inclination of the surrounding road with respect to the own vehicle posture is obtained from the position of the vanishing point on the image. As a method for determining the vanishing point, for example, “Introduction to Image Processing” ISBN 4-7898-1834-9, P.I. The method described in 50-62 is used.

  The model construction unit 6c uses the host vehicle position / posture acquired from the positioning unit 2 and sets a peripheral region with the host vehicle position as the origin and the direction of the host vehicle posture as the depth direction. Then, the three-dimensional shape information of the building located in the peripheral region is acquired from the map building database 1. Further, the surrounding relief is obtained from the relief recognition unit 11. Using the site shape included in the acquired three-dimensional shape information of the building, the position on the map of the point on the road that the building faces is calculated. Specifically, a perpendicular line is drawn from the center of the site with respect to the road closest to the vehicle among the roads adjacent to the site of the building. The intersection of the perpendicular and the road becomes the desired point. Then, the elevation difference between the vehicle position and the point is calculated using the surrounding undulations, and a three-dimensional model may be constructed together with the three-dimensional shape information of the building. Subsequent building identification is performed in the same manner as in the first embodiment.

  A hardware configuration for realizing the fourth embodiment of the present invention will be described with reference to FIG. The map building database 1, the positioning unit 2, the image acquisition unit 3, the attention direction receiving unit 4, the shape recognition unit 5, the model construction unit 6c, the building identification unit 7, the guidance information acquisition unit 8, and the display unit 9 are embodiments. 1 is realized with the same hardware configuration as in FIG. The undulation recognition unit 12 is realized by the CPU 103 using the image stored in the storage device 102 in the image acquisition unit 3 to calculate the peripheral undulation and store it in the storage device 102.

  According to the navigation device configured as described above, since the undulation recognition unit 12 is provided, the model construction unit 6c can construct a highly accurate three-dimensional model, and the map information includes no elevation information. Even when it is insufficient, it is possible to increase the accuracy of building identification.

Embodiment 5 FIG.
Depending on the driving situation, the driver may not be able to operate the navigation device, and may go too far over the building, thus failing to obtain appropriate information. Here, a description will be given of what enables a building to be identified from past images by saving the position and orientation of a vehicle and a live-action image when there is room. FIG. 12 is a system configuration diagram showing a navigation apparatus according to Embodiment 5 of the present invention. In this figure, the map building database 1, the positioning unit 2, the image acquisition unit 3, the building identification unit 7, the guidance information acquisition unit 8 and the display unit 9 perform the same functions as those in the first embodiment.

  The storage unit 13 stores the position and orientation of the host vehicle acquired from the positioning unit 2 and the actual captured image acquired from the image acquisition unit 3 in association with the acquired time. Then, the stored real image is played back and selected by the user, and the image is transferred to the attention direction receiving unit 4 and the shape recognition unit 5. Further, the vehicle position / posture associated with the same time as the image is passed to the model construction unit 6. Here, only an image suitable for use in the present method may be stored. Examples of inappropriate images include images that are too bright, images that are too dark, images that are blurred, images in which surrounding buildings are not visible due to the preceding vehicle, and images that are far from the current location. By reducing the number of images to be saved, the memory usage can be suppressed and the searchability of the images can be improved. Further, as a method for improving the searchability of the image, a method of hierarchizing the images to be stored may be taken with the time point of passing through the intersection as a break.

  The attention direction receiving unit 4 and the shape recognizing unit 5 function in the same manner as the attention direction receiving unit 4 and the shape recognizing unit 5 in the first embodiment, using the captured image acquired from the storage unit 13. The model building unit 6 functions in the same manner as the model building unit 6 in the first embodiment using the own vehicle position / posture acquired from the storage unit 13.

  A hardware configuration for realizing the fifth embodiment of the present invention will be described with reference to FIG. The map building database 1, the positioning unit 2, the image acquisition unit 3, the attention direction receiving unit 4, the shape recognition unit 5, the model construction unit 6, the building identification unit 7, the guidance information acquisition unit 8, and the display unit 9 are embodiments. 1 is realized with the same hardware configuration as in FIG. The storage unit 13 is a separate region of the storage device 102 in which the image stored in the storage device 102 in the image acquisition unit 3 and the own vehicle position and orientation stored in the storage region 102 in the positioning unit 2 are secured with a sufficient size. This is realized by storing in.

  According to the navigation device configured as described above, since the storage unit 13 is provided, the user can designate a building from a past image. Therefore, the user can safely and effectively edit the building according to the driving situation. Information can be acquired and the burden on the user can be reduced.

BRIEF DESCRIPTION OF THE DRAWINGS It is a system block diagram of the navigation apparatus which shows Embodiment 1 of this invention. It is an example of the hardware block diagram which shows the implementation | achievement method of the navigation apparatus which shows Embodiment 1-5 of this invention. It is a flowchart of a process of the building specific | specification part 7 which shows Embodiment 1 of this invention. It is explanatory drawing of the navigation apparatus which shows Embodiment 1 of this invention. It is explanatory drawing of the navigation apparatus which shows Embodiment 1 of this invention. It is a schematic diagram which shows the image at the time of the building specific | specification part 7 which shows Embodiment 1 of this invention calculating an evaluation value. It is a block diagram which shows the navigation apparatus which shows Embodiment 2 of this invention. It is a flowchart of a process of the building specific | specification part 7a which shows Embodiment 2 of this invention. It is a block diagram which shows the navigation apparatus which shows Embodiment 3 of this invention. It is a flowchart of the process of the building specific | specification part 7b which shows Embodiment 3 of this invention. It is a block diagram which shows the navigation apparatus which shows Embodiment 4 of this invention. It is a block diagram which shows the navigation apparatus which shows Embodiment 5 of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1, 1a Map building database 2 Positioning part 3 Image acquisition part 4 Attention direction receiving part 5 Shape recognition part 6, 6a, 6b, 6c Model construction part 7, 7a, 7b Building specification part 8 Guidance information acquisition part 9 Display part DESCRIPTION OF SYMBOLS 10 Running condition acquisition part 11 Attribute recognition part 12 Unevenness recognition part 13 Storage part 101 CPU
102 Storage Device 103 Microcomputer Interface 104 External Device Interface 105 Vehicle Interface 106 GPS Receiver 107 Gyro Sensor 108 Video Output Device 109 Sound Output Device 110 HMI Device 111 Camera

Claims (6)

A positioning unit for determining the current position of the vehicle;
A map building database that stores map and building information;
A model building unit for building a three-dimensional model around the current position from the current position information from the positioning unit and the map and building information from the map building database;
An image acquisition unit that captures the vicinity of the current position and acquires a real image;
In a display device that displays a live-action image from the image acquisition unit on a screen, an attention direction receiving unit that receives position information on a screen of a designated building and obtains a direction to the designated building with respect to a current position; ,
A shape recognizing unit that extracts shape characteristic information of a surrounding building from a photographed image from the image acquiring unit;
The building specified on the display screen using the three-dimensional model from the model building unit, the direction from the attention direction receiving unit, and the geometric feature information from the shape recognition unit is any building in the three-dimensional model. A building identification part that identifies whether there is,
A guide information acquisition unit for acquiring the guide information of the building specified by the building specifying unit from the map building database;
A navigation device comprising: a display unit that outputs the guidance information from the guidance information acquisition unit in at least one of audio and video. Based on the virtual image created by moving the 3D model so that the direction to the building specified on the display screen is the same as the direction to each building in the 3D model ,
Visibility calculated from the size of each building in the virtual image,
Direction coincidence calculated from the movement angle of the three-dimensional model;
An evaluation value based on at least one of the shape similarities calculated by comparing the shape feature information of the designated building from the shape recognition unit and the shape feature information of each building in the virtual image In search of
The navigation device according to claim 1, wherein the highest evaluation value is specified as the designated building. The navigation apparatus according to claim 1, wherein the building specifying unit specifies a designated building using a three-dimensional model in which an error range from the model building unit is set. The building information in the map building database includes location, site shape, number of floors, color and pattern,
An attribute recognizing unit for extracting the color and pattern of the building from the live-action image from the image acquiring unit;
The building specifying unit moves the 3D model so that the direction to the building specified on the display screen is the same as the direction to each building in the 3D model associated with the color and pattern information. Based on the virtual image created
Attribute similarity calculated by comparing color and pattern information from the attribute recognition unit with color and pattern information from the virtual image,
Visibility calculated from the size of each building in the virtual image,
Direction coincidence calculated from the movement angle of the three-dimensional model;
An evaluation value based on at least one of the shape similarities calculated by comparing the shape feature information of the designated building from the shape recognition unit and the shape feature information of each building in the virtual image In search of
The navigation device according to claim 1, wherein the highest evaluation value is specified as the designated building. Further comprising a undulation recognition unit for obtaining the undulations around the vehicle from the actual image from the image acquisition unit,
The model construction unit constructs a three-dimensional model around the current position from the current position information from the positioning unit and the map and building information from the map building database and the undulation information from the undulation recognition unit,
The building specifying unit uses the 3D model from the model building unit, the direction from the target direction receiving unit, and the shape feature information from the shape recognition unit to identify the building specified on the display screen in the 3D model. The navigation apparatus according to claim 1, wherein the navigation apparatus identifies which building. The navigation apparatus according to claim 1, further comprising a storage unit that stores information on the positioning unit and the image acquisition unit.

Images