JP2012083157A - Outdoor feature detection system, program for the same, and record media of program for the same - Google Patents

Abstract

PROBLEM TO BE SOLVED: To specify a point group included in a target feature based on the point group obtained as a result of a laser survey as well as a designated piece of a camera image, and to detect a position and a size of the feature from the specified point group.SOLUTION: The present invention comprises: a camera image data storage section (31) which stores an image of a periphery of a target feature; a point group data storage section (32) which stores a laser survey result of the periphery of the target feature as a point group; an image drawing section (21) which displays the image of the periphery of the target feature on a monitor; a point designation section (12) which designates at least one point included in a range of the target feature from the image of the periphery of the target feature; and a feature detection processing sections (22 to 25) which detect a position and a size of the range of the feature by deciding the range of the feature on the image based on at least the one designated point, selecting the point group included in the decided range from the point group storage section and separating the point group corresponding to the feature through grouping the point groups having distances therebetween equal to or less than a certain value.

Description

  The present invention relates to an outdoor feature detection system that performs outdoor feature detection using a laser point cloud and a camera image, an outdoor feature detection system program, and an outdoor feature detection system program recording medium.

  In recent years, mobile mapping system technology using a vehicle equipped with a GPS satellite, a gyro laser, or the like has been developed. Since this mobile mapping system can continuously acquire information on the surroundings where the vehicle has traveled, it is mainly expected to be applied to techniques such as high-precision map creation.

  Here, for application to high-precision map creation, it is necessary to detect features such as outdoor buildings and advertisements from information acquired by the mobile mapping system and extract necessary information.

  Laser surveying mainly determines the distance from the vehicle, from which the position (point) is determined. Since it is impossible to determine “what the laser hit” using only the points thus obtained, it is necessary to determine which object the point is in contact with by other information.

  Camera images are often used to identify the relationship between a point cloud (a set of points obtained by laser surveying) and an object. Specifically, a point cloud candidate hitting the feature can be found from the position of the object (feature) detected by the camera image on the image and the relationship between the camera image shooting position and the point cloud. .

  On the other hand, such point cloud candidates alone are insufficient to narrow down the point cloud that hits the feature. FIG. 15 is a graph in which the latitude and longitude of the point group included in the range of the feature (signboard) on the image is plotted. The plot as shown in FIG. 15 also includes features before and after the target signboard (for example, building walls, utility poles, electric wires, etc.), and these removal processes are necessary.

  As a method of selecting a point cloud of a specific feature, there is a method of selecting from point cloud candidates based on two camera images (see, for example, Patent Document 1). In this Patent Document 1, two images are used for the same feature, point cloud candidates are extracted from each of them, and a point cloud having a high degree of coincidence among the point clouds included in both is extracted. It is considered as a point cloud.

JP 2009-76096 A

However, the prior art has the following problems.
When the point cloud of the target feature is obtained from the overlap of the point clouds in the two images as in the technique described in Patent Document 1, the overlap range is determined by the positional relationship in which the images are acquired. Different. FIG. 16 and FIG. 17 are explanatory diagrams showing a case where the overlapping range is different depending on the positional relationship in which two images are acquired.

  The traveling direction of the mobile mapping system is the direction indicated by the arrows in FIGS. Here, as shown in FIG. 16, when the interval at which two images are taken is short, the overlap of the point groups in the two images increases, and simply overlapping does not hit the target feature. There is a possibility of picking up a point cloud. On the other hand, as shown in FIG. 17, when the distance at which two images are photographed is long, the overlap of the point clouds in the two images is small, but the point cloud that hits the surroundings of the feature is also present. There is a high possibility of selecting.

  Moreover, in patent document 1, grouping by the size of the feature is also performed. However, in order to perform such grouping, it is necessary to know the size of the feature in advance, and there is a problem that the feature cannot be grouped for a feature whose size is unknown.

  The present invention has been made to solve the above-described problems, and is included in a target feature based on a point cloud that is a laser survey result and a specified one camera image. An outdoor feature detection system, an outdoor feature detection system program, and an outdoor feature detection system program recording medium capable of identifying a point cloud and detecting the position and size of the feature from the identified point cloud The purpose is to obtain.

  An outdoor feature detection system according to the present invention stores an image around a target feature in an outdoor feature detection system that detects an outdoor feature using a point cloud as a laser survey result and a camera image. A camera image data storage unit, a point cloud data storage unit that stores laser survey results around the target feature as a point cloud, and an image around the target feature stored in the camera image data storage unit on the monitor Specified by the point drawing unit, the point designating unit for designating at least one point included in the region of the feature from the image around the target feature displayed on the monitor, and the point designating unit Based on at least one point, the region of the feature on the image is determined, the point group included in the determined region is extracted from the point cloud data storage unit, and the distance between the point groups is less than a certain distance. Feature by grouping A feature point detection processing unit for separating a corresponding point group and detecting the position and size of a feature region including at least one point designated by the point designation unit from the position and range of the separated point cloud It is.

  In addition, the outdoor feature detection system program according to the present invention includes a screen drawing means for displaying an image around a target feature stored in the camera image data storage unit on the monitor, and a display on the monitor. A point designating unit for designating at least one point included in the region of the feature among the images around the target feature to be processed, and a feature on the image based on at least one point designated by the point designating unit. The area of the object is determined, the point cloud included in the determined area is extracted from the point cloud data storage unit that stores the laser survey results around the target feature as the point cloud, and the distance between the point clouds is a fixed distance The point group corresponding to the feature is separated by grouping the following, and the position and size of the feature region including at least one point designated by the point designation means from the position and range of the separated point group Feature detection to detect It is intended to function as a management unit.

  Furthermore, the recording medium for the outdoor feature detection system program according to the present invention is a computer-readable recording medium in which the program for the outdoor feature detection system is recorded, and is stored in the camera image data storage unit in the computer. A screen rendering step for displaying on the monitor an image around the target feature that has been set, and at least one point included in the region of the feature from among the images around the target feature displayed on the monitor A point designation step to be designated, and a region of the feature on the image is determined based on at least one point designated in the point designation step, and a point group included in the decided region is a laser around the target feature. The point cloud corresponding to the feature is separated by extracting the survey results from the point cloud data storage unit that stores them as point clouds, and grouping the distances between the point clouds below a certain distance. Reading by a computer for executing processing including a feature detection processing step for detecting the position and size of a feature region including at least one point specified in the point specification step from the position and range of the separated point cloud This is a recording medium for a program for a possible outdoor feature detection system.

  According to the outdoor feature detection system, the program for the outdoor feature detection system, and the recording medium for the program for the outdoor feature detection system according to the present invention, the feature region on the image determined based on the designated input point By specifying the point cloud corresponding to the feature by extracting the point cloud included in the group and grouping the distance between the point clouds below a certain distance, the point cloud that is the laser survey result is specified. An outdoor feature detection system capable of identifying a point group included in a target feature based on a single camera image and detecting the position and size of the feature from the identified point group, A recording medium for the feature detection system program and the outdoor feature detection system program can be obtained.

It is an example of a screen of the outdoor feature detection system in Embodiment 1 of this invention. It is explanatory drawing of an example which acquires a point cloud with the outdoor feature detection system in Embodiment 1 of this invention. It is a block diagram of the outdoor feature detection system in Embodiment 1 of this invention. It is the flowchart which showed a series of operation | movement of the input of a point cloud extraction part in Embodiment 1 of this invention, a process, and an output. It is the flowchart which showed the specific process of step S400 of FIG. 4 in Embodiment 1 of this invention. It is explanatory drawing regarding the calculation method of the position on the image of the point cloud by the point cloud extraction part in Embodiment 1 of this invention. It is the flowchart which showed the specific process of step S500 of FIG. 4 in Embodiment 1 of this invention. It is the flowchart which showed the specific process of step S510 of FIG. 7 in Embodiment 1 of this invention. It is the flowchart which showed the specific process of FIG.7 S520 in Embodiment 1 of this invention. It is the flowchart which showed the specific process of step S5220 of FIG. 9 in Embodiment 1 of this invention. It is the flowchart which showed the specific process of step S5221 of FIG. 10 in Embodiment 1 of this invention. It is explanatory drawing which shows an example of the external shape of the point cloud calculated by the point cloud extraction part in Embodiment 1 of this invention. It is explanatory drawing for removing the influence of an electric wire etc. by the process of the point cloud extraction part in Embodiment 1 of this invention. It is explanatory drawing which plotted the point group of the utility pole extracted by the point group extraction part in Embodiment 1 of this invention, and an electric wire. It is the graph which plotted the latitude and longitude of the point group contained in the range of the feature (signboard) on the image. It is explanatory drawing which shows the case where the range of overlap differs by the positional relationship which acquired two images. It is explanatory drawing which shows the case where the range of overlap differs by the positional relationship which acquired two images.

  Hereinafter, preferred embodiments of an outdoor feature detection system, an outdoor feature detection system program, and an outdoor feature detection system program recording medium according to the present invention will be described with reference to the drawings. The outdoor feature detection system according to the present invention extracts a laser point group corresponding to a desired feature based on a designation input that specifies the position of the desired feature in the camera image, thereby obtaining the desired feature. The technical feature is to obtain the region of the object and detect the position and size of the feature.

  Moreover, the program for outdoor feature detection systems of this invention is a program for performing the function required for the outdoor feature detection system of this invention mentioned above with a computer. The recording medium for the outdoor feature detection system program is a computer-readable recording medium in which the program is recorded.

Embodiment 1 FIG.
First, the outline of the “method for selecting a point group” in the present invention will be described with reference to FIGS. FIG. 1 is a screen example of the outdoor feature detection system in Embodiment 1 of the present invention. In the present invention, “the system user designates a feature on the screen (here, a sign is designated)” to designate the position and size of a desired feature.

  At this time, the designation means may be a “user designates an area of a feature” method or a “determine area by image recognition using a position designated by the user” method.

  In the present invention, attention is paid to the fact that “a desired feature on the screen is recognized by the system user” based on the “designation of the feature on the screen by the system user”. That is, the present invention considers the premise that the system user can recognize the presence of the feature and the degree of size, although it may be somewhat hidden behind other features.

  Therefore, it is considered that such a premise is applied to a method for selecting a point cloud. Suppose that the point cloud is grouped in some form, and the “point cloud by feature” is put together, and the positional relationship between the point cloud and the mobile mapping system is considered. At this time, in view of the premise that “the system user can recognize the target feature” described above, the “feature is closest to the desired feature to be detected by the mobile mapping system”. Conceivable.

  However, at this time, it is necessary to remove the obstacle in front of the feature. In such a removal process, since the system user can recognize the feature, “the size of the obstacle is small (compared to the feature region) and does not become an obstacle to feature recognition”. Use. That is, when the point cloud is mapped on the screen, it is determined how much the feature area is covered, and if it is below the threshold value, it can be determined that it is not a feature, and as a result, obstacles can be removed. Become.

  In addition, regarding the grouping of point clouds, it is possible to use the fact that a point cloud that hits the same feature exists nearby. Specifically, for example, all the distances between the point groups are equal to or less than a certain distance.

  FIG. 2 is an explanatory diagram of an example of acquiring point clouds by the outdoor feature detection system according to Embodiment 1 of the present invention. As shown in FIG. 2, even when “a point cloud cannot be acquired by a previous feature at a certain timing”, since the laser transmission frequency is high, it can be acquired at another timing, and therefore, the influence of the obstacle is small. That is, in the example of FIG. 2, even if a part of the feature 1 is not shown because the feature 2 is obstructed by the camera, a place that could not be obtained by the camera due to laser transmission from another place. The point cloud can be acquired.

  By the series of processes described above, it is possible to select a point cloud that hits a specific feature. Furthermore, if the size of the feature is known in advance, the size can be calculated using the obtained point cloud, and a check can be made as to whether or not the condition is met. However, since the point cloud does not necessarily cover the end points of the feature, it must be noted that there is an error in the calculated size.

Next, a specific configuration and operation of the outdoor feature detection system of the present invention for realizing the above-described “point cloud selection method” will be described in detail below with reference to FIGS. .
FIG. 3 is a configuration diagram of the outdoor feature detection system according to Embodiment 1 of the present invention. The outdoor feature detection system according to the first embodiment includes a designation input unit 10, a processing unit 20, and a data storage unit 30.

  The designation input means 10 is a part that receives input data designated by an operator, and includes an image designation unit 11 and a point designation unit 12. The processing unit 20 is a part that performs data processing based on the received input data and the data stored in the data storage unit 30, and includes a screen drawing unit 21, an image region determination unit 22, and a point group extraction unit 23. A real plane calculation unit 24 and a region calculation unit 25. Here, the image region determination unit 22, the point group extraction unit 23, the real plane calculation unit 24, and the region calculation unit 25 correspond to a feature detection processing unit.

  Further, the data storage means 30 is a storage unit in which input / output data is accumulated, and includes a camera image data storage unit 31, a point cloud data storage unit 32, a vehicle / camera parameter storage unit 33, and a region output result storage unit 34. Have.

Next, the function of each component will be described in detail.
First, functions in the designation input means 10 will be described. The image designation unit 11 is a unit that designates an image to be displayed from among the images in the camera image data storage unit 31. Specifically, as the image specifying unit 11, for example, as shown in FIG. 1, the thumbnails of the images in the camera image data storage unit 31 are displayed on a part of the screen, and from among them, A desired item can be selected and displayed at the same magnification.

  The point designating unit 12 is a means for designating one point on the image designated by the image designating unit 11, and this point is included in the feature on the image. Specifically, it can be performed by designating one point on the image selected and displayed on the screen by a pointing operation using a mouse or the like.

Next, data stored in the data storage unit 30 will be described.
The camera image data storage unit 31 is a storage unit that stores camera images taken from a vehicle. At this time, the time at the time of photographing is stored together with each image.

  The point cloud data storage unit 32 is a storage unit that stores laser survey data measured from a vehicle. The survey data stores the time when the laser survey was performed together with the coordinate values of each point group. The coordinate system of each point group is a plane rectangular coordinate system (x, y, z coordinate), where the x coordinate is the latitude direction, the y coordinate is the distance from the origin in the longitude direction, and the z coordinate is the altitude. (For example, see Non-Patent Document 1: Geospatial Information Authority of Japan, Plane Cartesian Coordinate System (Ministry of Land, Infrastructure, Transport and Tourism Notification No. 9 in 2004), http://www.gsi.go.jp/LAW/heimencho.html) .

  The vehicle / camera parameter storage unit 33 is a storage unit that stores the camera parameters and camera parameters when the camera image data in the camera image data storage unit 31 and the point cloud data in the point cloud data storage unit 32 are acquired. is there. Specifically, data such as the position / orientation of the vehicle at the time of camera shooting, the arrangement of the camera on the vehicle, and the camera focal length are stored.

  The region output result storage unit 34 is a storage unit that stores the position and size of the feature calculated by the region calculation unit 25 described later.

  Next, functions in the processing means 20 will be described. The screen drawing unit 21 performs a series of displays accompanying image designation by the image designation unit 11, a series of displays accompanying point designation by the point designation unit 12, and a region calculated by the region calculation unit 25.

  The image region determination unit 22 is a unit that determines a region on the image of the feature from one point specified by the point specification unit 12. As this determination means, for example, a method using learning such as Adaboost is used (for example, Non-Patent Document 2: Yoav Freund and Robert E. Schapier. “A decision-theoretic generalization of on-line learning and camp. boosting ". Journal of Computer and System Sciences, 55 (1): 119-139, August 1997).

  The point cloud extraction unit 23 stores an image designated by the image designation unit 11, a feature region decided by the image region decision unit 22, point cloud data stored in the point cloud data storage unit 32, and vehicle / camera parameter storage. Using the vehicle / camera parameters stored in the unit 33, a point group necessary for processing in the subsequent real plane calculation unit 24 is extracted (details will be described later).

  The real plane calculation unit 24 calculates a real plane including the feature from the point group extracted by the point group extraction unit 23. By this calculation, the feature and the actual position (point cloud) are linked.

  Then, the area calculation unit 25 obtains the range of the target feature on the real plane calculated by the real plane calculation unit 24. The width and height of the feature can be calculated from the result of the range calculation.

Next, a series of operations of the outdoor feature detection system according to the first embodiment will be described.
By designating a desired image in the camera image data storage unit 31 by the image designation unit 11, the image and its image shooting time are uniquely determined. The screen drawing unit 21 displays this information and the range calculated by the region calculation unit 25 in an overlapping manner on the screen. When the range of the region extends outside the image, the screen drawing unit 21 draws only the range that fits within the image.

  In addition, when the area calculation by the area calculation unit 25 is not performed, the area is not displayed. Further, the image designation unit 11 may designate whether or not to display the area. If it is specified that the area is not displayed, the screen drawing unit 21 displays only the image even after the area calculation is performed.

  After the image is displayed on the screen by the screen drawing unit 21, the operator designates one point on the image via the point designation unit 12.

  Next, the image region determination unit 22 determines the region of the feature on the image (four vertices of the minimum rectangular shape covering the feature) based on one point on the image specified by the point specification unit 12. . If the image area determination unit 22 determines that no feature exists in the range including the specified point, the image region determination unit 22 waits for input from the point specification unit 12 without performing the subsequent processing.

  Next, the point cloud extraction unit 23 extracts a point cloud of the feature based on the feature region determined by the image region determination unit 22. FIG. 4 is a flowchart showing a series of input, processing, and output operations of the point cloud extraction unit 23 according to Embodiment 1 of the present invention.

  First, the point cloud extraction unit 23 inputs a point cloud (other than the target point cloud) including the target point cloud (corresponding to input 1), a region range R (corresponding to input 2), and an area S (corresponding to input 3). To do. In step S400, the point group extraction unit 23 obtains a point group within the range R.

  FIG. 5 is a flowchart showing specific processing of step S400 of FIG. 4 in Embodiment 1 of the present invention. First, in step S410, the point cloud extraction unit 23 extracts only point cloud data measured within a certain time range from the point cloud data storage unit 32 from the camera image capturing time tc. At this time, the data mainly used is a point near the camera photographing time.

Next, in step S420, the point group extraction unit 23 performs position calculation when each extracted point group is mapped on the camera image. Specifically, when calculating the position of the point cloud on the image, the point cloud extraction unit 23 first calculates the camera position of each point in the point cloud based on the data in the vehicle / camera parameter storage unit 33. Relative positions (x ₁ , y ₁ , z ₁ ) are calculated from the vehicle position / orientation at the time of camera image capturing and the camera position / orientation on the vehicle. Thereafter, the position of the point cloud on the image is calculated using the camera focal length f, the distance p _m per pixel of the camera, and the optical axis center coordinates (xc, yc) of the camera.

FIG. 6 is an explanatory diagram relating to a method for calculating the position of the point cloud on the image by the point cloud extraction unit 23 according to Embodiment 1 of the present invention. As shown in FIG. 6, the position (x, y) on the image can be calculated by the following equation (1).
(X, y) = (xc + dx, yc−dy)
= (Xc + (f · z ₁ ) / (p _m · x ₁ )
, Yc- (f · y ₁ ) / (p _m · x ₁ )) (1)

  Next, in step S430, the point group extraction unit 23 extracts the points included in the region determined by the image region determination unit 22 from the positions on the image of the point group calculated in the previous step S420. . As a result, a point cloud that may have hit the target feature is obtained, and narrowing down is performed from here.

  Returning to FIG. 4, in step S <b> 500, the point group extraction unit 23 further performs point group limitation and outputs the result. FIG. 7 is a flowchart showing specific processing of step S500 of FIG. 4 in Embodiment 1 of the present invention. First, in step S510, the point cloud extraction unit 23 groups input point clouds and performs “point cloud separation”.

  FIG. 8 is a flowchart showing specific processing of step S510 in FIG. 7 according to Embodiment 1 of the present invention. The point cloud extraction unit 23 sorts the input point clouds according to the distance between the position of the point cloud and the “position of the mobile mapping system at the time of image capture” (step S5101), and the closest point is moved to the vicinity. A certain point group is made into one group (step S5102-step S5108).

  As a specific process, for all the points closest to the mobile mapping system, a point group having a distance less than or equal to a threshold value α (for example, 28 cm) is selected (step S5103). A point group at a distance is selected (step S5106). The process is continued, and when the points no longer increase, the obtained group is set as one group (step S5105).

  Next, the same processing is performed for points that have not entered the group, and the processing is continued until processing is performed for all input point groups (steps S5102 to S5109). Finally, the groups obtained in this way are numbered in the order of the obtained groups (step S5110). By such a series of processes, the “point group separation” in the previous step S510 is completed.

  Next, referring back to FIG. 7, in step S520, the point group extraction unit 23 performs “optimal group selection from each group” for selecting features from the point group grouped by the “point group separation” process. That is, the point group extraction unit 23 determines whether or not the grouped point groups are target features.

  FIG. 9 is a flowchart showing specific processing of step S520 in FIG. 7 according to Embodiment 1 of the present invention. In order to perform “optimal group selection from each group”, the point cloud extraction unit 23 selects one point cloud group closest to the mobile mapping system (step S5210), and the point cloud is determined by the image region determination unit 22. Calculate how much of the determined area is occupied (step S5220), and if the value is equal to or greater than a threshold β (for example, 0.9) (step S5230), the point cloud group is determined to be the target feature. (Step S5240).

  However, since the point cloud is merely position information, the point cloud extraction unit 23 calculates the area obtained by “the outline of the outline formed by the point cloud” as the calculation of the range S ′ occupied by the point cloud in the previous step S5220. Is used. FIG. 10 is a flowchart showing specific processing of step S5220 of FIG. 9 in Embodiment 1 of the present invention.

  In FIG. 10, first, an outline point group is acquired to obtain an outer shape obtained by the point group (step S5221). FIG. 11 is a flowchart showing specific processing of step S5221 of FIG. 10 according to Embodiment 1 of the present invention.

  By performing a series of processing from step S1101 to step S1114 shown in FIG. 11, the point cloud can be obtained after selecting the point cloud in such a way as to take the outer circumference of the point cloud as much as possible.

  FIG. 12 is an explanatory diagram showing an example of the outline of the point cloud calculated by the point cloud extraction unit 23 according to Embodiment 1 of the present invention. In FIG. 12, a black point (♦) is a point group mapped on the image, and a white point (□) is an outline of the point group calculated from the black point (♦). By performing the series of processing shown in FIG. 11, the point cloud extraction unit 23 can extract the outer shape as shown as white dots (□) in FIG. 12.

  Returning to FIG. 10, the point group extraction unit 23 next removes the point group constituting the outer shape obtained in the previous step S5221 from the input point group in step S5222. Further, in step S5223, the point group extraction unit 23 calculates the outer shape from the remaining point group by the same processing as in the previous step S5221.

  The outline calculation in step S5223 is effective for removing the influence of electric wires and the like. FIG. 13 is an explanatory diagram for removing the influence of an electric wire or the like by the processing of the point cloud extraction unit 23 according to Embodiment 1 of the present invention. The image shown on the left side of FIG. 13 shows a situation where a utility pole and an electric wire are present in front of a signboard that is a desired feature.

  The point group shown in the upper right of FIG. 13 corresponds to a utility pole and an electric wire, and the point group shown in the lower right of FIG. 13 corresponds to a signboard. As can be seen from FIG. 13, the electric wire or the like widely covers a signboard area, which is a desired feature, while no point cloud is present in the center.

  FIG. 14 is an explanatory diagram in which the utility poles and the point groups of the electric wires extracted by the point group extraction unit 23 according to the first embodiment of the present invention are plotted, and corresponds to the upper right diagram of FIG. As shown in FIG. 14, the amount of point clouds actually looks smaller than the “range covered with the created outline”, but this is detected as a thin object such as an electric wire covers the upper and lower parts of the area. That's what happens. On the other hand, since the point cloud corresponding to such a thin wire is about one row at most, the influence of the wire can be removed by removing the point cloud after creating the outer shape.

  Returning to FIG. 10 again, in step S5224, the point group extraction unit 23 calculates the area of the outer shape obtained as the processing result of steps S5221 to S5223. By such a series of processing, the point cloud extraction unit 23 can perform area calculation of the range covered with the point cloud.

  Then, as described above, the point group extraction unit 23 performs the processing of Step S5230 and Step S5240 in FIG. 9, and the area (S) of the region determined by the image region determination unit 22 and the previous Step S5220. By obtaining the ratio to the area (S ′) obtained in step 1, a group of target point groups can be determined.

  When the number of points obtained by the point group extraction unit 23 is less than 3, the subsequent processing is not performed as an error, and the image designation unit 11 returns to the image designation. On the other hand, when three or more points are obtained by the point group extraction unit 23, the actual plane calculation unit 24 calculates a plane including the obtained points (see FIG. 3 above).

  Actually, since the obtained point has an error, it is rare that a plane on which all the points lie can be calculated. Therefore, the real plane calculation unit 24 calculates the closest plane using a least square method or the like. The obtained plane is a plane including the specified feature.

  After obtaining the plane by the real plane calculation unit 24, the region calculation unit 25 determines the target range (see FIG. 3 above). For this purpose, using the plane obtained by the real plane calculation unit 24 and the four vertices of the rectangle defining the feature region determined by the image region determination unit 22, “the obtained plane has four points. Calculate “mapped location”.

  This calculation is uniquely determined because the above equation (1) and the position of the point to be calculated (x, y, z) satisfy the equation of the plane obtained by the real plane calculator 24.

  As shown in FIG. 3, after the area calculation by the area calculation unit 25, the process returns to the screen drawing unit 21, draws an area on the image, and waits for image selection from the image designation unit 11. In addition, the area calculation unit 25 stores the width and height of the feature and the four vertexes of the rectangle in the area output result storage unit 34.

  As described above, according to the first embodiment, the position / size of the feature can be calculated based on the camera image and the laser point group. Furthermore, by drawing the specified region of the feature on the image, confirmation by human eyes can be performed at the same time.

  Further, when another image is designated by the image designation unit, the area is redrawn on the designated image as described above. The correctness of the calculated area can be confirmed by superimposing the calculated area on an image showing a feature other than the image for which the area has been calculated.

  By doing this, for each feature that appears in multiple images, calculate each one and superimpose it on another image, and determine what the person best overlaps, You can select the exact position and size of the feature. By doing in this way, the error which cannot be picked up only by calculation can be absorbed through visual confirmation on a screen.

Embodiment 2. FIG.
In the first embodiment, the case where the feature is detected by designating and inputting one point on the image has been described. On the other hand, in the second embodiment, a case will be described in which the feature input is detected with two points to be designated and input instead of one point.

  In the second embodiment, the system user designates and inputs two diagonal diagonal points covering the desired feature. Thereby, the image area determination part 22 can limit the range determined based on two points of this diagonal, and can reduce misdetection. At this time, of the positions calculated by the area calculation unit 25, two points are positions input by the user, and the remaining two points are positions determined by the image area determination unit 22.

  As described above, according to the second embodiment, by specifying two points for specifying and specifying a feature, in addition to the effect of the first embodiment, the feature image can be displayed on the image of the feature. It is possible to reduce false detection when determining a region, and to further improve performance.

  In addition, when the position of the feature is detected and the range of the feature in the image is obtained by the first and second embodiments described above, the range of the feature is determined from the image as shown in FIG. It may be cut out and saved separately. In this way, by separately displaying the position and the image, there is an additional advantage that the operator can easily confirm the feature detection status.

  DESCRIPTION OF SYMBOLS 10 designation | designated input means, 11 image designation | designated part, 12 point designation | designated part, 20 processing means, 21 screen drawing part, 22 image area determination part, 23 point group extraction part, 24 real plane calculation part, 25 area | region calculation part, 30 data storage Means 31 Camera image data storage unit 32 Point cloud data storage unit 33 Vehicle / camera parameter storage unit 34 Area output result storage unit

Claims (6)

In an outdoor feature detection system that detects outdoor features using a point cloud and camera image that are laser survey results,
A camera image data storage unit for storing an image around the target feature;
A point cloud data storage unit for storing laser survey results around the target feature as a point cloud;
A screen drawing unit for displaying on the monitor an image around the target feature stored in the camera image data storage unit;
A point designating unit for designating at least one point included in the region of the feature from the image around the target feature displayed on the monitor;
Based on the at least one point designated by the point designating unit, a region of the feature on the image is determined, a point group included in the determined region is extracted from the point group data storage unit, A point group corresponding to the feature is separated by grouping those whose distance is equal to or less than a certain distance, and the ground including the at least one point designated by the point designation unit from the position and range of the separated point group An outdoor feature detection system comprising: a feature detection processing unit that detects the position and size of an object region. The outdoor feature detection system according to claim 1,
The feature detection processing unit calculates an outer shape for each group of point groups, calculates an area from the calculated outer shape, and occupies a region of the feature on the image determined based on the at least one point An outdoor feature detection system, wherein a point group having an area ratio equal to or greater than a predetermined threshold is specified as a point group included in the target feature. In the outdoor feature detection system according to claim 1 or 2,
A region output result storage unit for storing a detection result by the feature detection processing unit;
The feature detection processing unit cuts out an image corresponding to the position and size of the detected feature from images around the target feature, and stores the image in the region output result storage unit. Outdoor feature detection system. In the outdoor feature detection system according to claim 3,
The screen drawing unit superimposes the detection result by the feature detection processing unit stored in the region output result storage unit on another image stored in the camera image data storage unit and displays it on the monitor. An outdoor feature detection system characterized by that. A program for an outdoor feature detection system,
Computer
Screen drawing means for displaying on the monitor an image around the target feature stored in the camera image data storage unit;
Point designating means for designating at least one point included in the region of the feature from the image around the target feature displayed on the monitor;
Based on the at least one point designated by the point designating means, a region of the feature on the image is determined, and a point group included in the determined region is pointed to a laser survey result around the target feature. Extracting from the point cloud data storage unit stored as a group, separating the point group corresponding to the feature by grouping the distance between the point groups below a certain distance, the position of the separated point group and An outdoor feature detection system program for functioning as a feature detection processing means for detecting the position and size of a feature region including at least one point designated by the point designation means from a range. A computer-readable recording medium in which a program for an outdoor feature detection system is recorded,
On the computer,
A screen drawing step for displaying on the monitor an image around the target feature stored in the camera image data storage unit;
A point designating step of designating at least one point included in the region of the feature from the image around the target feature displayed on the monitor;
Based on the at least one point designated by the point designating means, a region of the feature on the image is determined, and a point group included in the determined region is pointed to a laser survey result around the target feature. Extracting from the point cloud data storage unit stored as a group, separating the point group corresponding to the feature by grouping the distance between the point groups below a certain distance, the position of the separated point group and A computer-readable outdoor feature for executing processing including: a feature detection processing step for detecting a position and a size of a feature region including the at least one point specified by the point specifying means from a range. Recording medium for detection system program.

Images