JP5356269B2 - Laser data filtering method and apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method for filtering laser data with high reliability. <P>SOLUTION: The method for filtering the laser data takes, as an object, laser data groups 5, 5, ..., which are formed by computing three-dimensional coordinates of a reflecting point 4, based on the timing of acquisition of reflected light 3 obtained by sweeping a prescribed area on the ground with beams of laser light 2 from a flying body 1 flying high in the sky, and on the position and direction of emission of each beams of laser light 2, and which contain the three-dimensional coordinates of a plurality of reflecting points 4 obtained from the single laser irradiation light 2, while a digital surface model 8 having three-dimensional coordinates of the prescribed area is formed by stereo matching processing which uses overlapping photographed images 7 obtained by photographing the prescribed area from the flying body 1 by a camera in an overlapping manner and the photographing position and direction of each image. Then, an elevation coordinate value of the digital surface model 8 and that of each laser data 5 are compared with each other. The laser data 5 having the elevation coordinate value within a prescribed range in relation to that of the digital surface model 8 are removed from the laser data groups 5, 5, ..., and ground surface candidate data 6 are extracted. <P>COPYRIGHT: (C)2011,JPO&amp;INPIT

Description

  The present invention relates to a laser data filtering method and apparatus.

  As a laser data filtering method, a method described in Patent Document 1 has been known. In this conventional example, the laser data is transmitted from the laser scanner mounted on the aircraft, and the distance from the laser scanner is measured by the time difference until the wave reflected on the ground (reflected light) is detected. It is generated by obtaining a measurement point group having a predetermined density. The aircraft has a three-dimensional position and orientation measured by GPS and an inertial measurement device, and the three-dimensional coordinates of each measurement point group on the ground are calculated by combining with the above-described measurement distance.

  When the ground is forest, the laser beam is reflected by the crown, or part of it is reflected by the ground by sewing between the trees, and further by the trunk of the tree and the undergrowth under the tree. Therefore, if the distribution of the measurement point group of laser data is shown in the vertical cross-sectional direction, depending on the season, it forms a group in the upper layer corresponding to the tree canopy, and the lower layer corresponding to the ground or undergrowth, and the middle corresponding to the trunk, etc. It is distributed like dots in the layer. The above laser data filtering is performed by selecting a point having the lowest altitude locally from the laser data. By using the lowest elevation point selected in this way, the lower surface layer that seems to correspond to the ground surface is approximated to a curved surface, and the ground height is estimated.

JP 2006-3332 A (page 3-4)

  However, the conventional example described above has a drawback of low reliability.

  In other words, the above-described filtering is based on the premise that the laser data having the lowest elevation locally indicates the ground surface, but the laser light is set at a pinpoint aiming at the ground surface. In the first place, it is not guaranteed that it has reached the ground surface in the first place, and even if it is a point that locally has the lowest elevation, the possibility that it is not the ground surface cannot be denied. This possibility can be reduced by setting a wide local area for extracting the minimum elevation, and therefore the local area tends to be wide if the above-mentioned assumptions are considered. In this case, undulations on the ground surface that are smaller than a widely set region are ignored.

  The present invention has been made to solve the above drawbacks, and an object of the present invention is to provide a highly reliable laser data filtering method. Another object of the present invention is to provide a highly reliable laser data filtering apparatus.

  In general, the filtering of laser data is performed by, for example, performing a secondary process automatically performed by a computer as in the conventional example described above, and then performing a secondary process in which a human makes a correction based on an empirical rule while interpreting the terrain. It consists of multiple stages of processing. Therefore, if the existence of the subsequent filtering process is considered, the filtering process performed before that is more than the direct attempt to extract the laser data indicating the ground surface as the final product as in the conventional example described above. On the contrary, by extracting laser data other than the ground surface with high reliability and removing this, ground surface candidate data can be extracted indirectly.

The present invention was made in consideration of the above points,
The three-dimensional coordinates of the reflection point 4 are determined by the acquisition timing of the reflected light 3 obtained by sweeping the laser light 2 from the flying object 1 flying over the sky to a predetermined area on the ground, and the emission position and emission direction of each laser light 2. Laser data filtering method for extracting ground surface candidate data 6 from laser data groups 5, 5,... Generated by calculation and including three-dimensional coordinates of a plurality of reflection points 4 obtained from a single laser irradiation light 2 Because
A numerical value having the three-dimensional coordinates of the predetermined region by stereo matching processing using an overlapping photographed image 7 obtained by photographing the predetermined region from the flying object 1 with a camera and the photographing position and photographing direction of each image 7 Generate a surface model 8
Next, the altitude coordinate values of the numerical surface layer model 8 and each laser data 5 are compared, and the laser data 5 having an altitude coordinate value within a predetermined range R with respect to the altitude coordinate value of the numerical surface layer model 8 is the laser data. The above-described object is achieved by providing a laser data filtering method for extracting ground surface candidate data 6 by removing from the groups 5, 5,.

  According to the present invention, the filtering of the laser data is performed using an image captured by a camera that is a measurement element separate from the laser beam 2, and other than the ground surface by using the reliability of the measurement result of the separate measurement element. This is done by extracting the laser data 5 indicating that with high accuracy and removing it. In other words, since the captured image of the camera can capture the surface of the natural feature, the numerical surface model 8 (DSM: Digital Surface Model) generated based on this image has the three-dimensional coordinates of the surface of the natural feature. By removing the laser data 5 that is reflected and has the corresponding three-dimensional coordinates, it is possible to remove the laser data 5 that clearly shows a natural feature, thereby extracting the ground surface candidate data 6 Can do.

  The case where the laser beam 2 is irradiated from above to a forested area where plants as natural features flourish will be specifically considered below. In this case, as shown in FIG. 3 (c), for one laser irradiation light 2 (irradiation pulse), for example, reflected light 3A (reflection pulse, in particular, reflected first from a tree crown or the like) is returned. In addition to “first pulse”, the reflected light 3B (reflected pulse, particularly “last pulse”, which is reflected on the ground surface (candidate of the ground surface) by passing through a gap between branches and leaves and returning to the end. In addition, it is also referred to as reflected light 3C (reflected pulse, “2nd pulse”, “3rd pulse”, etc.) reflected by the undergrowth between the crown and the ground surface. , Collectively referred to as “intermediate pulse”). That is, the reflected light 3 reflected from the surface layer other than the crown of the surface layer can also be obtained by passing the laser light 2 through the gaps between the branches and leaves or between the trees. On the other hand, in camera photography, it is difficult to capture a shadowed portion such as the back of a gap between branches and leaves, and therefore, only the surface crown can be captured.

  Therefore, when the numerical surface layer model 8 is generated based on the camera photographing, the three-dimensional coordinates indicate the position of the crown, more precisely, the outer edge of the crown, and the laser data 5 having the three-dimensional coordinates corresponding to this is represented. If the object is to be removed, the laser data 5 indicating the crown, that is, the laser data 5 indicating other than the ground surface can be removed. As also shown in the above-described conventional example, since the ratio of the reflected light 3 reflected by the tree crown generally tends to be high, the above removal removes the laser data 5 indicating other than the ground surface at a high ratio. It is possible to narrow down the laser data groups 5, 5,. Further, since it is ensured that the removed laser data 5 does not indicate the ground surface based on the camera photographing, when the filtering process is further repeated thereafter, for example, the remaining laser data group It is only necessary to delete an appropriate number of specific laser data 5 that can be determined as other than the ground surface based on empirical rules from 5, 5,. For this reason, for example, the verification process of the filtering process considering the possibility that there is the ground surface candidate data 6 that has been excessively removed as in the prior art becomes unnecessary, and the work efficiency is greatly improved.

  Specifically, the above filtering processing is performed by stereo matching processing for the generation of the numerical surface layer model 8 based on the camera image, and for the determination of the laser data 5 to be removed, as described above, the laser light 2 This can be done by comparing whether or not it has passed between branches, that is, by comparing the elevation coordinate values of the numerical surface model 8 and the laser data 5. In this comparison, it is clear that when the three-dimensional coordinates of the numerical surface layer model 8 and the laser data 5 have a certain degree of precision, it is clear that perfect coincidence is not possible as a numerical value in the first place. The predetermined allowable range R is set, and the determination can be made depending on whether the altitude coordinate value of the laser data 5 is within the predetermined range R with respect to the altitude coordinate value of the numerical surface layer model 8. Further, if the camera image 7 for generating the above numerical surface model 8 is mounted on the flying object 1 together with the laser distance measuring device and acquired simultaneously with the laser data 5, the measurement operation can be efficiently advanced. Further, for example, it is possible to reduce the difference in ground resolution from the laser data 5 by acquiring the flight altitude differently and acquiring it separately from the laser data 5.

  Further, the numerical surface layer model 8 can be specifically configured as, for example, an irregular triangle network (TIN), and in this case, in principle, a plane created by a triangle whose apex is a coordinate measurement point Thus, the elevation coordinates between coordinate measurement points are interpolated. Therefore, as a whole, the area where the altitude coordinates are specified by interpolation occupies the majority, and while coordinate comparison with the laser data 5 can be performed according to this resolution, enormous calculation processing is required for the comparison. End up. At this point, a mesh 9 based on the plane coordinates is set in the numerical surface layer model 8, and the altitude coordinate value of the mesh representative point 10 is set to the altitude coordinate value in each mesh 8 with the center point of each mesh 9 as the mesh representative point 10. When the coordinates are compared with the laser data 5 on the assumption that they coincide with each other, the arithmetic processing can be reduced, and the elevation coordinate value changes too much by using the center point of the mesh 9 as a reference. Since it is difficult to occur, the position accuracy is not excessively inferior to the altitude coordinate value estimated by interpolation as described above. In this case, specifically, the elevation coordinate value of the laser data 5 having a plane coordinate value within half of the mesh size a with respect to the mesh representative point 10 in each of the coordinate axes X and Y directions orthogonal to each other on the plane coordinate. Can be compared with the altitude coordinate value of the mesh representative point 10 for calculation processing.

  When the position accuracy of the numerical surface layer model 8 and the laser data 5 is different, the measurement error can also be reduced by setting the size of the mesh size a in consideration of this difference.

  In the above, the case where the numerical surface layer model 8 generated using the captured image 7 of the camera, which is a measurement element separate from the laser beam 2, is used. However, when an appropriate numerical surface layer model 8 is available, etc. In such a case, if such a numerical surface model 8 is used as it is, it is possible to reduce the trouble of stereo matching processing and the like. In this case, there is a possibility that the obtained numerical surface layer model 8 is not generated using the overlapping photographed image 7 of the camera. However, from the laser data groups 5, 5,. The reliability of extraction depends on the reliability of the obtained numerical surface layer model 8 itself, and can be ensured as appropriate by ensuring this.

The above filtering of the laser data 5 is
The three-dimensional coordinates of the reflection point 4 are determined by the acquisition timing of the reflected light 3 obtained by sweeping the laser light 2 from the flying object 1 flying over the sky to a predetermined area on the ground, and the emission position and emission direction of each laser light 2. Laser data filtering device that extracts ground surface candidate data 6 from laser data groups 5, 5,... Generated by calculation and including three-dimensional coordinates of a plurality of reflection points 4 obtained from a single laser irradiation light 2 Because
An overlapped captured image 7 obtained by capturing the predetermined region from the flying object 1 with a camera, an input unit 11 for inputting a capturing position and a capturing direction of each image 7;
A DSM generating unit 12 for generating a numerical surface layer model 8 having the three-dimensional coordinate data of the predetermined region by performing stereo matching processing on the overlapping captured image 7 using the imaging position and the imaging direction;
The altitude coordinate values of the numerical surface model 8 and each laser data 5 are compared, and laser data 5 having an altitude coordinate value within a predetermined range R with respect to the altitude coordinate value of the numerical surface layer model 8 is the laser data group 5. 5... Can be realized by using a laser data filtering device having a filtering processing unit 13 that extracts ground surface candidate data 6 from the ground surface candidate data 6.

  As is clear from the above description, according to the present invention, a reliable laser data filtering method can be provided, and a numerical model of the ground surface excluding natural features can be efficiently generated. it can. In addition, it is possible to provide a highly reliable laser data filtering device, and it is possible to enhance facilities for efficiently generating a numerical model of the ground surface excluding natural features.

It is a hardware block diagram of the DTM production | generation apparatus which includes the filtering apparatus based on this invention in part. It is a hardware block diagram of the measurement system mounted in a flying body. It is a figure explaining the ground measurement state, (a) is a figure which shows the scanning pattern by a laser distance measuring device, (b) is a figure which shows the imaging | photography condition with a camera corresponding to (a), (c) is natural ground It is a figure explaining the reflected light from a thing. It is a figure explaining the distribution state of a laser data, (a) is an image figure of an altitude direction, (b) is an image figure of the plane direction of the 4B part of (a), and the numerical surface layer model which set the mesh was displayed superimposed. Is. It is a detailed image diagram in the elevation direction showing the relative positional relationship with the topographic shape of the laser data, (a) is for the data before filtering processing as the laser data, overlaying the numerical surface layer model with the mesh set The displayed figure, (b) is a figure which shows after a filtering process, (c) is a figure which shows a modification corresponding to (a). It is a flowchart of the filtering process which concerns on this invention.

  FIG. 1 to FIG. 6 show an outline of DTM (Digital Terrain Model) generation for a predetermined area of a forest area. FIG. 1 shows a DTM generating apparatus comprising a computer partially including a filtering apparatus according to the present invention. The input of the DTM generating apparatus is used for generating a DTM in an area where natural vegetation is densely grown. Laser ranging data, GPS / IMU data, captured image data, and ground control point (GCP) data are input to the unit 11. The laser distance measurement data and the like are acquired while flying over the predetermined area as the DTM generation target area with the aircraft (aircraft) 1 equipped with the measurement system 20 shown in FIG. .

  The measurement system 20 includes a control rack device 21 composed of a computer, a laser distance measuring device 22, a photographing device 23, and a GPS / IMU device 24. The control rack device 21 includes a control unit 21a that controls each of the laser distance measuring device 22, the imaging device 23, and the GPS / IMU device 24, and controls and acquires the data acquired by controlling these control target devices at the acquisition time. The information is recorded in the recording unit 21b in association with it.

  The laser distance measuring device 22 amplifies the laser medium output from the laser output unit 22a by the laser oscillation unit 22b and oscillates it as the laser beam 2, and detects the reflected light 3 of the laser beam 2 thus oscillated. The laser receiver 22c is provided. The laser light 2 emitted from the laser oscillating unit 22b is distributed in the traveling direction by the repetitively rotating mirror of the scanning mechanism unit 22d, and the laser receiving unit 22c detects the reflected light 3 through the scanning mechanism unit 22d.

  The above laser beam 2 emission and mirror rotation are controlled by the control unit 21a described above, and the laser beam 2 emission timing and the mirror angle at the time of emission are recorded in the above-described recording unit 21b. When the laser receiver 22c detects the reception of the reflected light 3, the distance measurement unit 22e similarly records the received light detection timing of the reflected light 3 that functions as distance data on the recording unit 21b. The distance measurement unit 22e calculates the time difference between the emission timing of the laser light 2 and the light reception detection timing of the reflected light 3, and the distance from the laser emission point to the reflection point 4 due to this time difference, and causes the recording unit 21b to perform the above-described emission. It is also possible to record the distance data directly instead of the light reception detection timing data.

  The photographing device 23 is a digital camera that performs photographing according to control of photographing timing, focal length, photographing azimuth (angle of depression), etc. by the control unit 21a, and a photographed image 7 acquired by photographing is photographed image data together with the photographing timing described above. Is recorded in the recording unit 21b. This imaging device 23 is configured with a single imaging device (camera), and can also be configured with a plurality of imaging devices in addition to shooting while changing the flight course so that the imaging areas are substantially adjacent. . Specifically, as shown in a shooting situation to be described later, for example, a plurality of cameras (5 in this embodiment) are arranged in different directions so that their shooting areas are almost adjacent to each other. A wide range of images can be acquired at once.

  The GPS / IMU device 24 acquires position information by GPS and posture information by an IMU (Inertial Measurement Unit), a GPS antenna 24a that receives radio waves from GPS satellites, an accelerometer, and a gyro. An IMU sensor unit 24b that measures angular velocity and acceleration with respect to inclination in three directions of rolling (ω), pitching (φ), and heading (κ) is provided. The GPS antenna 24a and the IMU sensor unit 24b are controlled by the GPS / IMU computer unit 24c controlled by the control unit 21a described above, and position information and posture information are recorded in the recording unit 21b described above at predetermined time intervals. Is done.

  Therefore, by flying the aircraft 1 over the DTM generation target region and distributing the laser beam 2 in the direction perpendicular to the flight direction by the scanning mechanism 22d, the laser beam 2 is, for example, as shown in FIG. The DTM generation target area is swept to cover the area, and the emission timing of each laser irradiation light 2, the angle of the mirror at the time of emission, and the light reception detection timing of the reflected light 3 are stored in the recording unit 21 b as laser ranging data. To be recorded. In FIG. 3A, the arrow indicates the flight route of the aircraft 1. 3A shows the case where the scanning is traced in parallel lines, but it is also possible to trace in a zigzag shape, a sine wave shape, or the like.

Further, since the DTM generation region is a forest zone, as shown in FIG. 3C, a single laser irradiation light 2 is reflected by a plurality of reflection points 4 such as leaves 25a of trees 25, undergrass 26, and ground surface 27. In each case, the light is reflected. In this case, the laser receiver 22c detects each of the first pulse 3A, the intermediate pulse 3C, and the last pulse 3B, and the detection timings of the plurality of reflected lights 3 are recorded in the recording unit 21b.

  Similarly, by flying the aircraft 1 over the DTM generation target area and continuously capturing images with the imaging device 23 at short time intervals, for example, positions slightly deviated in the flight direction as shown in FIG. A plurality of photographed images 7 that are photographed in the above and whose photographing region is continuous in the flight direction are acquired and recorded as photographed image data in the recording unit 21b. In FIG. 3B, an arrow indicates a flight route of the aircraft 1 and a white circle indicates a shooting point. In addition, by setting an overlap where a part of the shooting area overlaps between the shot images 7 adjacent in the flight direction, the same area is shot at different shooting positions. Prior to this shooting, a ground reference point (not shown) whose position coordinates are specified by measurement in advance is set or set in the DTM generation target area, and the ground reference point is simultaneously shot in the above-described captured image 7. Is done.

  In FIG. 3B, in consideration of the resolution and the imaging efficiency, five cameras are used as the imaging device 23 so that the imaging area becomes longer in the direction perpendicular to the flight direction. The case where the captured image 7 is acquired at the same time is shown, and in this case, the captured images 7 and 7 adjacent to each other in the direction orthogonal to the flight direction are set with side wraps in which portions of the captured region overlap. Further, instead of the above five cameras, it is also possible to take a picture with a single camera, or take a picture with a so-called area sensor or line sensor provided with a plurality of image sensors for aerial photogrammetry. Also, an analog camera can be used as the photographing device 23. In this case, the photographed image 7 data is generated by reading the photographed photograph with a scanner.

  Similarly, the GPS / IMU device 24 records the position information and attitude information of the aircraft 1 flying over the DTM generation target area in the recording unit 21b. In this embodiment using the GPS positioning by the continuous kinematic method, the above-mentioned GPS / IMU data is obtained from the ground ground base station (not shown) in addition to the position information and the attitude information of the aircraft 1 thus obtained. It is configured by adding location information acquired at the same time using a GPS antenna.

  Laser distance measurement data, GPS / IMU data, and captured image data acquired by the measurement system 20 and the ground reference station as described above are recorded on the hard disk or the appropriate storage medium that constitutes the recording unit 21b described above, and generated as a DTM. The input unit 11 of the apparatus includes a connection terminal to these hard disks and the like. In addition, the input unit 11 includes a connection line to the Internet, and by using browser software included in a calculation unit (to be described later) of the DTM generation device, the position coordinates of the ground reference point described above can be obtained from, for example, the web page of the Geographical Survey Institute. It is input as ground control point data by downloading.

  The DTM generating apparatus to which the above various data is input from the input unit 11 calculates the three-dimensional coordinates of the reflection points 4 of all the laser beams 2 using the input data, and obtains the laser data groups 5, 5,. The calculation unit 29 includes a laser data calculation processing unit 28 to be generated and a DSM generation unit 12 that generates a numerical surface layer model 8 of the DTM generation target region using the captured images 7 in which the imaging regions overlap. The laser data calculation processing unit 28 uses the GPS / IMU analysis processing unit 30 that calculates the position and orientation of the aircraft 1 on the geographical coordinates and the measurement result by the laser distance measuring device 22 to relate to the DTM generation target region. A laser / GPS data combination processing unit 31 that calculates three-dimensional coordinates based on the laser data 5, and the GPS / IMU analysis processing unit 30 captures the three-dimensional coordinates related to the DTM generation target region by the imaging device 23. The DSM generation unit 12 is configured together with the captured image / GPS data combination processing unit 32 calculated based on the image 7.

  The GPS / IMU analysis processing unit 30 is composed of commercially available software such as GPS / IMU analysis processing software of Applanix, for example. Specifically, the aircraft GPS data acquired via the GPS antenna 24a, the IMU data acquired by the IMU sensor unit 24b, and the above-described ground reference point data are combined on the basis of time, so that the aircraft 1 flight position and swing at the time of flight are calculated, and aircraft position / attitude angle data in which the position and attitude angle of the aircraft 1 are recorded is generated. The position of the aircraft 1 can be calculated by performing baseline analysis processing on the GPS data of the aircraft 1 and the ground reference point data.

  The laser / GPS data combination processing unit 31 uses the aircraft position / attitude angle data acquired from the GPS / IMU analysis processing unit 30 and the laser ranging data acquired from the input unit 11 to reflect all the reflection points of the laser beam 2. The three-dimensional coordinates of 4 are calculated to calculate the laser data groups 5, 5,. Specifically, the coordinate value of the reflection point 4 is based on the position of the aircraft 1 when the laser beam 2 is emitted, and the emission direction of the laser beam 2 is determined by the rolling angle (ω), the pitching angle (φ), and the heading angle ( It is calculated using the distance from the laser distance measuring device 22 to the reflection point 4 which is obtained from the three directions (κ) and obtained by using the time difference between the emission / light reception timing of the laser light 2. The rolling (ω) direction is calculated from the tilt of the aircraft 1 and the mirror angle of the laser distance measuring device 22, and the pitching (φ) and heading (κ) directions are calculated from the measured values of the IMU. FIG. 4 shows an image in which the coordinates of each reflection point 4 are represented by dots in the elevation direction and the plane direction for the laser data groups 5, 5,... Generated as described above. In addition, the laser / GPS data combination processing unit 31 performs coordinate conversion from the geodetic coordinate system to the planar rectangular coordinate system for the three-dimensional coordinates of the laser data groups 5, 5,. The ellipsoidal height to altitude conversion process is performed, and the calculation unit 29 has a noise removal unit (not shown) before being output from the laser / GPS data combination processing unit 31 and input to the filtering processing unit 13 described later. The noise data is removed from the laser data groups 5, 5,.

  The captured image / GPS data combination processing unit 32 generates a numerical surface layer model 8 based on the aircraft position / attitude angle data, the captured image data and the ground reference point data input from the input unit 11. Specifically, it is so-called photogrammetry, and the captured images 7 and 7 having parallax by being adjacent to each other in the flight direction are associated with each other by the shooting position and the shooting direction and the orientation point appropriately set in the shooting overlap area. The three-dimensional coordinates are given using the ground reference point. The shooting position and shooting direction of each shot image 7 are calculated from the depression angle of the camera at the time of shooting and the aircraft 1 position / attitude angle data included in the shot image data, and these shooting positions and shooting directions, the above-mentioned orientation points and A stereo model (numerical surface layer model 8) is generated by performing a stereo matching process using the ground reference point and the focal length included in the captured image data. Note that the above-mentioned orientation points are generated by automatically extracting feature points at the periphery of each captured image 7.

  In addition, the stereo model generated by the captured images 7 arranged in the flight direction as described above is generated by each of the plurality of captured images 7 arranged in the direction orthogonal to the flight direction as described above. Then, these stereo models are connected to each other at the orientation points, and the numerical surface layer model 8 of the entire DTM generation area is generated. The numerical surface layer model 8 is configured as an irregular triangle network, for example, by using a tuning point used for stereo matching between the captured images 7 and 7.

  Further, the arithmetic unit 29 described above uses the numerical surface layer model 8 generated by the DSM generation unit 12 as described above, and performs filtering for extracting the ground surface candidate data 6 from the laser data groups 5, 5,. In addition to the processing unit 13, a mesh setting unit 33 that sets the mesh 9 in the numerical surface layer model 8 is included as preprocessing of filtering processing by the filtering processing unit 13. The mesh setting unit 33 sets the mesh 9 based on the plane coordinates in the numerical surface layer model 8, and the mesh size a is set to, for example, the same size as an area corresponding to a single pixel of the captured image 7. The Further, the mesh setting unit 33 extracts the center point of each mesh 9 as the mesh representative point 10 and calculates the elevation coordinate value of the mesh representative point 10 based on the irregular triangle network to obtain the elevation representative value for each mesh 9. Set. The above mesh 9, mesh size a, and mesh representative point 10 will be illustrated using a part of FIG.

  The filtering processing unit 13 compares the numerical surface layer model 8 in which the elevation representative value is set in units of mesh 9 and the laser data 5 of the laser data groups 5, 5,. Then, the laser data 5 of three-dimensional coordinates corresponding to the numerical surface layer model 8 is removed, and the ground surface candidate data 6 is extracted from the laser data groups 5, 5,. The comparison of the three-dimensional coordinate values between the laser data 5 and the mesh representative point 10 described above is based on the mesh size a described above for each of the X and Y axis directions with respect to the plane coordinate value (x, y) of each mesh representative point 10. The laser data 5 having a plane coordinate value within the allowable range is extracted, and the altitude coordinate value of the laser data 5 is compared with the altitude coordinate value of the mesh representative point 10. The In addition, when comparing the altitude coordinate values, a predetermined allowable error (allowable range) R is set, and this allowable error R is an allowable error for each mesh 9 input from the keyboard or the like provided in the input unit 11 shown in FIG. The upper limit side allowable value β and the lower limit side allowable value α of R are configured.

  In this embodiment, the upper limit side allowable value β of the allowable error R is based on, for example, an error in the elevation direction in laser measurement and an error in the elevation direction in camera shooting, and for example, an integration error of both is adopted. Further, the lower limit side allowable value α is determined in consideration of the height of the vegetation that grows in the mesh 9. For example, when the height is the same as the above-described accumulation error, the height dimension is adopted as it is, and the height is the accumulation error. In the case of exceeding, the height and the integration error are further integrated.

  The height of the vegetation can be applied by selecting a table in which the vegetation type and the average height are associated with each other in the calculation unit 29, for example. In this selection, the color appearing in the camera image 7 and the reflection intensity of the laser light 2 that can be acquired by the laser distance measuring device 22 can be referred to.

  FIG. 5 explains the removal processing of the laser data 5 by the filtering processing unit 13 described above. In the figure, hatching indicates the terrain 34, the two-dot chain line indicates the outer edge 35 of the tree crown, and the bar graph-shaped portion indicates the altitude for each mesh 9. A numerical surface layer model 8 in which coordinate values are expressed by their heights, black circles and white circles represent laser data 5, and arrows represent a tolerance range R set in the mesh 9. It should be noted that the allowable error range R represents only those corresponding to some of the meshes 9 so as not to complicate the figure. FIG. 5A shows a state before the removal processing of the laser data 5, white circles indicate the laser data 5 whose altitude coordinate value is within the allowable error R, and black circles indicate the laser data 5 (ground surface candidate) that is outside the allowable error R. Data 6) is shown. FIG. 5B shows a state after the removal process of the laser data 5. As can be seen from these drawings, the laser data 5 indicating the surface other than the ground surface is removed at a high rate by the removal processing of the laser data 5 based on the numerical surface model 8, and the ground surface candidate data 6 is extracted.

  FIG. 5C shows a modified example in which the lower limit allowable value α of the allowable error R is set more strictly. In this case, the double that was not removed in FIG. 5B described above. Laser data 5 other than the ground surface indicated by a circle is removed. The strict setting of the lower limit side allowable value α of the allowable error R described above is performed, for example, by setting the height of the vegetation based on actual measurement.

  Further, the calculation unit 29 includes a correction processing unit 35 and a mesh processing unit 36 in order to generate a DTM based on the laser data groups 5, 5,. The correction processing unit 35 further removes an appropriate number of the laser data 5 from the laser data groups 5, 5,... After the filtering processing by an input operation from an operator using a mouse or the like (not shown) of the input unit 11. For example, when the distribution of the laser data 5, 5,... In the elevation direction is displayed on a monitor outside the figure, and the operator who observes the distribution inputs the designation of the laser data 5 to be removed via the input unit 11. The laser data 5 is removed. The laser data group 5, 5,... After this correction processing is composed only of laser data 5 having a high probability of being the ground surface.

  The mesh processing unit 36 generates a DTM based on the laser data 5 output from the correction processing unit. In this generation, the laser data groups 5, 5,. Interpolation processing is performed on mesh data having a predetermined density interval, and an irregular triangle network as a DTM is generated based on the mesh data. The interpolation process can be performed by, for example, the nearest neighbor method or the inverse distance weighting method.

  The generated DTM can be output to a monitor or an appropriate storage medium via the output unit 37 of the DTM generating apparatus shown in FIG. In FIG. 1, reference numeral 38 denotes a filtering processing device that shares the filtering processing of the laser data 5 in the DTM generating device.

  Therefore, in the DTM generation processing by the above DTM generation apparatus, first, the laser data groups 5, 5,... Are generated by the calculation unit 29 based on the laser distance measurement data and GPS / IMU data input from the input unit 11. At the same time, the calculation unit 29 generates the numerical surface layer model 8 based on the captured image data, the ground reference point data, and the GPS / IMU data described above, which are similarly input from the input unit 11. Further, meshes 9 are set in the numerical surface layer model 8 generated in this way, and the center point of each mesh 9 is set as a mesh representative point 10. Then, the laser data groups 5, 5,... Are compared with the mesh representative point 10 and the altitude coordinate value, the laser data 5 having the three-dimensional coordinates corresponding to the mesh representative point 10 is removed, and the ground surface candidate data 6 is obtained. Extracted. Thereafter, the ground surface candidate data 6 is subjected to correction processing (secondary filtering processing) and then interpolated into mesh data, and a DTM is generated from the mesh data.

  Further, as the filtering processing by the above-described filtering processing apparatus, as shown in FIG. 6, after the generation of the numerical surface layer model 8 (step S1) and the setting of the mesh 9 (step S2), the laser data 5 and the mesh are processed. The elevation coordinate values of the representative point 10 are compared (step S3). Specifically, this comparison is performed based on the mathematical formula shown in FIG. 6, and thereafter, the removal (step S4-1) or adoption (step S4-2) of the compared laser data 5 is determined. The above comparison / determination process is repeated until all of the laser data 5 is performed (step S5), whereby only the ground surface candidate data 6 is extracted, and the filtering process ends. This filtering process can be configured by a computer program in which the above-described computer execution procedure is described.

  In the above-described embodiment, the case where manual processing based on the operator's judgment is performed as the secondary filtering processing has been described. However, when appropriate filtering parameters can be set, the processing is configured as automatic processing by a computer program. It is also possible to do. Similarly, the case where the laser data 5 is acquired for a forest area has been shown, but it is also possible to target an area where some vegetation is included. For example, it is possible to generate ground surface candidate data of a wider area by combining the obtained ground surface candidate data 6 of only the vegetation area with other appropriate ground surface candidate data.

DESCRIPTION OF SYMBOLS 1 Aircraft 2 Laser light 2A Single laser irradiation light 3 Reflected light 4 Reflection point 5 Laser data 6 Ground surface candidate data 7 Photographed image 8 Numerical surface layer model 9 Mesh 10 Mesh representative point 11 Input part 12 DSM generation part 13 Filtering process Part R Tolerance (allowable range)
Each coordinate axis that is orthogonal on the X and Y plane coordinates a Mesh size

Claims (4)

It is generated by calculating the three-dimensional coordinates of the reflection point according to the acquisition timing of the reflected light obtained by sweeping the laser light from the flying object flying over the sky to a predetermined area on the ground and the emission position and emission direction of each laser light A laser data filtering method for extracting ground surface candidate data from a laser data group including three-dimensional coordinates of a plurality of reflection points obtained from a single laser irradiation light,
A numerical surface layer model having three-dimensional coordinates of the predetermined region by a stereo matching process using an overlapping captured image obtained by photographing the predetermined region from the flying object with a camera and a photographing position and a photographing direction of each image. Generate
Next, the altitude coordinate values of the numerical surface layer model and each laser data are compared, and laser data having an altitude coordinate value within a predetermined range with respect to the altitude coordinate value of the numerical surface layer model is removed from the laser data group. A laser data filtering method for extracting ground surface candidate data. After setting a mesh based on plane coordinates in the numerical surface layer model,
The center point of each mesh is extracted as a mesh representative point, and the elevation coordinate value of the laser data having a plane coordinate value within half the mesh size with respect to the mesh representative point in each of the coordinate axis directions orthogonal to each other on the plane coordinate. The laser data filtering method according to claim 1, wherein the method is compared with an elevation coordinate value of a mesh representative point. It is generated by calculating the three-dimensional coordinates of the reflection point according to the acquisition timing of the reflected light obtained by sweeping the laser light from the flying object flying over the sky to a predetermined area on the ground and the emission position and emission direction of each laser light A laser data filtering method for extracting ground surface candidate data from a laser data group including three-dimensional coordinates of a plurality of reflection points obtained from a single laser irradiation light,
After obtaining a numerical surface layer model having three-dimensional coordinates of the predetermined area and setting a mesh based on plane coordinates in the numerical surface layer model, the center point of each mesh is extracted as a mesh representative point,
Next, the elevation coordinate value of the laser data having a plane coordinate value within half the mesh size with respect to the mesh representative point in each of the coordinate axis directions orthogonal to each other on the plane coordinate is compared with the elevation coordinate value of the mesh representative point, A laser data filtering method of extracting ground surface candidate data by removing laser data having an elevation coordinate value within a predetermined range with respect to a mesh representative point from the laser data group. It is generated by calculating the three-dimensional coordinates of the reflection point according to the acquisition timing of the reflected light obtained by sweeping the laser light from the flying object flying over the sky to a predetermined area on the ground and the emission position and emission direction of each laser light A laser data filtering device for extracting ground surface candidate data from a laser data group including three-dimensional coordinates of a plurality of reflection points obtained from a single laser irradiation light,
An overlapped photographed image obtained by overlapping the predetermined area from the flying object and photographed by the camera, an input unit for inputting a photographing position and a photographing direction of each image;
A DSM generation unit that generates a numerical surface layer model having three-dimensional coordinate data of the predetermined region by performing stereo matching processing on a duplicate captured image using the imaging position and the imaging direction;
The elevation surface coordinate value of each laser data is compared with the numerical surface layer model, and laser data having an elevation coordinate value within a predetermined range with respect to the elevation coordinate value of the numerical surface layer model is removed from the laser data group to obtain the ground surface. A laser data filtering apparatus comprising: a filtering processing unit that extracts candidate data.

Images