Point cloud processing device, point cloud processing method and program

Abstract

To provide a point cloud processing device, a point cloud processing method, and a program that are capable of correcting the positions of point clouds with higher accuracy.SOLUTION: A point cloud processing device includes a control unit. The control unit, as acquisition means, acquires a reference point cloud and a correction target point cloud obtained by measuring the road or sidewalk surface and surrounding features of the road surface with overlapping portions. The control unit, as horizontal position correction means, extracts partial point clouds from the reference point cloud and the correction target point cloud, which are in a range higher than moving objects on the road or sidewalk, and compares the partial point clouds with each other to align the horizontal position of the correction target point cloud to the position of the reference point cloud. The control unit includes vertical position correction means for matching the heights of common areas having common horizontal positions in the reference point cloud and the correction target point cloud aligned by the horizontal position correction means.SELECTED DRAWING: Figure 5

Description

[Technical field]

[0001] The present invention relates to a point cloud processing device, a point cloud processing method, and a program. [Background technology]

[0002] A high-precision three-dimensional road database is being developed for road management and the construction of systems for autonomous vehicle driving. Three-dimensional measurements of roads are performed, for example, by laser measurements using a Mobile Mapping System (MMS). In the MMS, a vehicle traveling on a road irradiates various surrounding locations with laser light and detects the reflected light to obtain a point cloud that represents the three-dimensional positions of the road surface and surrounding features along the road. The three-dimensional position of each point that constitutes the point cloud is determined by receiving radio waves from a positioning satellite related to the Global Navigation Satellite System (GNSS) and performing positioning calculations.

[0003] When a road has many lanes, it is difficult to irradiate the entire road width with high density laser light at once. Therefore, point clouds obtained during multiple runs are joined to obtain a point cloud for the entire road width. When joining, the point clouds are aligned to correct minute positional deviations that occur between the point clouds obtained during multiple runs. Even during a single run, the positions of each point based on the positioning results may fluctuate slightly. Therefore, it is desirable to divide the running direction into multiple sections and perform alignment in the overlapping parts of each section to eliminate the influence of the fluctuation. For alignment, technologies such as ICP (Iterative Closest Point) and LS3D (Least Squares 3D Surface Matching) are known (for example, Non-Patent Document 1). [Prior art documents] [Non-patent literature]

[0004] [Non-Patent Document 1] Koji Mano and 6 others, "Accuracy Evaluation of Point Clouds Obtained by Mobile Measurement System (MMS)", Photogrammetry and Remote Sensing, Japan Society of Photogrammetry and Meteorology, September 7, 2012, Vol. 51, No. 4, pp. 186-200 Summary of the Invention [Problem to be solved by the invention]

[0005] However, when applying such a registration technique to a point cloud measured while moving on a road or sidewalk, it may not be possible to perform proper registration due to passing objects such as other vehicles or pedestrians, which may result in a decrease in positional accuracy.

[0006] An object of the present invention is to provide a point cloud processing device, a point cloud processing method, and a program that can correct the positions of point clouds with higher accuracy. [Means for solving the problem]

[0007] In order to achieve the above object, the present disclosure provides: An acquisition means for acquiring a reference point cloud and a correction target point cloud obtained by measuring a road or sidewalk surface and surrounding features of the road surface with overlapping portions therebetween; a horizontal position correction means for extracting partial point clouds from the reference point cloud and the correction target point cloud in a range higher than a moving object on the road or sidewalk, and for comparing the partial point clouds with each other to align a horizontal position of the correction target point cloud with a position of the reference point cloud; a vertical position correction means for matching heights of a common area having a common horizontal position in the reference point group and the correction target point group aligned by the horizontal position correction means; The point cloud processing device includes: Effect of the Invention

[0008] According to the present disclosure, there is an effect that the positions of the point cloud can be corrected with higher accuracy. [Brief description of the drawings]

[0009] [Figure 1] FIG. 2 is a block diagram showing a functional configuration of the information processing device. [Diagram 2] FIG. 10 is a diagram showing a schematic example of setting of each block and a center point. [Diagram 3] FIG. 13 is a diagram illustrating horizontal alignment. [Figure 4] 11 is a diagram illustrating setting of an extraction range of a road surface point cloud for calculating a height deviation amount. FIG. [Diagram 5] 11A and 11B are diagrams for explaining an example of height adjustment. [Figure 6] 13 is a flowchart showing a control procedure for a point cloud position correction process. [Figure 7] FIG. 13 is a diagram illustrating alignment between lanes. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing the functional configuration of an information processing device 1 which is a point cloud processing device according to this embodiment.

[0011] The information processing device 1 includes a control unit 11, a storage unit 12, an input / output interface (I / F) 13, a display unit 14, an operation reception unit 15, and the like.

[0012] The control unit 11 controls the overall operation of the information processing device 1. The control unit 11 has a processor that performs arithmetic processing. The processor may be a single general-purpose CPU (Central Processing Unit), or may have multiple CPUs that perform arithmetic processing in parallel or independently depending on the application. The processor may include a processor specialized for a specific arithmetic processing or image processing. The control unit 11 performs various control processing by reading and executing a program 120 from the storage unit 12.

[0013] The storage unit 12 has a RAM (Random Access Memory) and a non-volatile memory, and stores various data. The RAM provides a working memory space for the control unit 11 and stores temporary data. The non-volatile memory stores and holds the program 120, setting data, and the like. The non-volatile memory is, for example, a flash memory or a HDD (Hard Disk Drive), but is not limited to these. The storage unit 12 may have a ROM (Read Only Memory). An initial control program and the like can be stored in the ROM. The program 120 includes a control program related to the point cloud position correction process described below.

[0014] The input / output interface 13 inputs and outputs data between the information processing device 1 and the outside (including peripheral devices). The input / output interface 13 has a connection terminal 131 and a communication unit 132. The connection terminal 131 includes, for example, a Universal Serial Bus (USB) terminal and a Local Area Network (LAN) connector. The communication unit 132 controls communication in accordance with a communication rule (protocol) related to the LAN, such as TCP / IP.

[0015] The peripheral devices may include a database device 21, which is an auxiliary storage device, and an optical reader 22 that reads portable storage media (optical disks) such as CD-ROM, DVD, and Blu-ray (registered trademark). The portable storage media may include a magnetic tape, and the peripheral devices may include a reader that reads the magnetic tape.

[0016] Data that the information processing device 1 can acquire from the outside via the input / output interface 13 includes measurement data 201 and movement trajectory data 202. The measurement data 201 is data including a point cloud obtained by measuring the three-dimensional positions of the road surface and the surfaces of features (surrounding features) around the road surface at appropriate intervals using a measuring device mounted on a measurement vehicle moving on the road. The measurement data 201 may also be data including a point cloud obtained by measuring the three-dimensional positions of the sidewalk surface and the surfaces of the surrounding features of the road surface at appropriate intervals using a measuring device mounted on a pushcart moving on the sidewalk or a measuring device carried on the back of a measurer moving on the sidewalk. The surrounding features are, for example, buildings and slopes along the road or sidewalk, as well as roadside signs. The point cloud represents the three-dimensional shape of the road surface and the surrounding features.

[0017] The measuring instruments include, for example, a laser scanner, a satellite positioning device, and an attitude measuring device. In this case, the laser scanner emits a laser beam and detects the reflected wave, thereby obtaining relative position information of the reflection point with respect to the emission point. That is, the distance from the emission point to the reflection point is specified by the time until the emitted laser beam is detected and the TOF (Time of Flight) according to the speed of light. Alternatively, the relative position information may be obtained by a phase-shift type laser scanner instead of the TOF type. Also, the direction of the reflection point with respect to the emission point is specified from the emission angle of the laser beam in the laser scanner and the attitude of the laser scanner measured by the attitude measuring device. On the other hand, the absolute position of the emission point, i.e., the geographical position and the emission time, are obtained by the satellite positioning device performing satellite positioning using the GNSS (Global Navigation Satellite System).

[0018] These results are combined to obtain the absolute position of each reflection point for which the relative position information is obtained. The absolute position is expressed as a three-dimensional position in a Cartesian coordinate system, such as a plane Cartesian coordinate system having X-axis and Y-axis mutually orthogonal in the horizontal direction and Z-axis along the vertical direction, or as a three-dimensional position having latitude, longitude and altitude. The laser light is emitted in each direction at appropriate time intervals, and a point cloud is obtained, which is a collection of data representing the three-dimensional positions of multiple discrete points (reflection points) that are measured on the surface of the road and surrounding features.

[0019] The measurement data 201 includes a point cloud obtained by one or more measurement operations. In this point cloud data 201, the measurement operation is associated with each point constituting the point cloud obtained by the measurement operation. For example, a point cloud measured in one run of a measurement vehicle may be stored in one file with a unique file name. Also, for example, a point cloud obtained by a measurement vehicle running on an uphill lane of a section of a road and a point cloud obtained by running on an outhill lane of a section common to the section may be stored in different files.

[0020] In addition, in the measurement data 201, the three-dimensional position of each point constituting the point cloud is associated with the measurement position when the point was measured. This makes it possible to extract points measured at measurement positions within an arbitrary range from the point cloud. In addition, in the measurement data 201, the measurement order of the measurement positions can be specified. The measurement positions arranged in this measurement order represent a movement trajectory. In this embodiment, the movement trajectory corresponding to each point cloud extracted from the measurement data 201 is acquired as the movement trajectory 202.

[0021] The measured position is position information based on the absolute position acquired at appropriate intervals by the satellite positioning. When the acquisition timing of the absolute position by the satellite positioning does not coincide with the measurement timing of each point, the absolute position corresponding to the measurement timing may be acquired by interpolation from the relationship between the absolute position acquired by the satellite positioning before and after the measurement timing and the date and time. The measured position represents, for example, the three-dimensional position of the satellite positioning device when each point is measured (the position itself obtained by the satellite positioning). For example, the measured position may be the three-dimensional position of the emission point of the laser scanner converted based on the relationship between the satellite positioning device and the installation position of the laser scanner. For example, the measured position may be a three-dimensional position in which the value of the height component at the three-dimensional position of the satellite positioning device or the laser scanner is replaced with the value of the road surface height, which is the surface formed in the point cloud at the position where the horizontal component coincides. Alternatively, the measured position may be a two-dimensional position obtained by removing the height component from any of the three-dimensional positions obtained above.

[0022] The three-dimensional position of each point and the measurement position may correspond one-to-one or many-to-one. In the case of one-to-one correspondence, for example, each point cloud file can be a file in which a set of data consisting of the measurement position and the three-dimensional position of each point is arranged in the order of measurement. In the case of many-to-one correspondence, for example, the measurement positions are thinned out at a predetermined distance interval, and each point cloud file can be a file in which a set of data consisting of the remaining measurement positions after thinning out and the three-dimensional positions of multiple points measured between the measurement positions and the next remaining measurement positions is arranged in the order of measurement. Note that the distance interval in this case is set in advance to be shorter than the distance between center points described later, that is, so that each block described later includes multiple measurement positions. In the measurement data 201, the three-dimensional position of each point may be further associated with the measurement date and time of each point, the orientation at the time of measurement, and the like.

[0023] The display unit 14 performs display on the display screen based on the control of the control unit 11. The display screen is, for example, a liquid crystal display or an organic EL (Electro-Luminescent) display, but is not limited to these.

[0024] The operation reception unit 15 receives an input operation from the outside, and outputs an operation signal corresponding to the input operation to the control unit 11. The operation reception unit 15 includes, for example, a keyboard and a pointing device. The pointing device may be a mouse. The display unit 14 and / or the operation reception unit 15 may be peripheral devices of the information processing device 1. In other words, they may be attached to the main body (computer) of the information processing device 1 including the control unit 11, the storage unit 12, and the input / output interface 13.

[0025] Next, the point cloud processing operation of this embodiment will be described.

[0026] As described above, the three-dimensional position of each point constituting the point cloud is calculated based on the absolute position and date and time obtained by satellite positioning. However, this date and time information may be accompanied by slight fluctuations depending on the status of the measuring instrument, such as heat generation status. As a result, the three-dimensional position of each point constituting the point cloud may become inaccurate depending on the fluctuations. Therefore, in this embodiment, the point cloud, which is the three-dimensional position data continuously acquired during the movement of the measuring instrument, is divided into multiple sections, and the point clouds of the divided sections are aligned with each other to correct the point cloud to an accurate one free of the influence of the fluctuations.

[0027] First, a center point is determined for each predetermined distance on the movement trajectory (on a broken line connecting the measurement positions) associated with the point cloud to be processed. The predetermined distance is a length within a range in which the influence of the above-mentioned fluctuation can be ignored, for example, 30 meters. However, the first center point is determined at a position half the predetermined distance from the tip of the movement trajectory. For each of these center points, measurement positions within a distance range of half the predetermined distance before and after the center point on the movement trajectory are extracted, and points associated with the extracted measurement positions are extracted from the point cloud. Alternatively, measurement positions within the distance range for each predetermined distance on the movement trajectory may be extracted, and further points associated with the extracted measurement positions may be extracted from the point cloud, and a center point may be determined at a position half the predetermined distance on the movement trajectory. As a result, blocks representing collections of points extracted for each center point are set consecutively along the movement trajectory. Hereinafter, the distribution range of points extracted for each center point is referred to as the range of each block. The range of each block includes an overlapping portion with the range of the previous and next blocks, and alignment is possible using the points in this overlapping portion (overlapping portion between point groups). The absolute position (geographical coordinates) of the leading block can be manually adjusted.

[0028] 2 is a diagram showing an example of setting center points C1-C3 and blocks A1-A3. For ease of explanation, the extension direction of the road is taken as the X direction, and the width direction of the road is taken as the Y direction. The range of blocks A1-A3 indicates the range related to the road surface, and actually extends wide in the Y direction and includes the surface positions of surrounding features.

[0029] The blocks A1 and A2 have an overlapping portion D12 where they overlap, and the blocks A2 and A3 have an overlapping portion D23 where they overlap. Using the measurement data of the overlapping portions D12 and D23, the range of one of the blocks (e.g., block A1) is set as a reference section, and the range of the other block (e.g., block A2) is set as a correction target section. The amount of correction is specified by calculating the amount of deviation of the positions of the points belonging to the point cloud (correction target point cloud) in the correction target section from the three-dimensional positions of the points belonging to the point cloud (reference point cloud) in the reference section. In other words, the correction amount is a value obtained by inverting the sign of each component of the deviation amount.

[0030] The point cloud in this embodiment is acquired with the road surface and surrounding features as the original measurement targets. However, the point cloud may include moving objects such as vehicles and pedestrians that are not the original measurement targets. Such moving objects become disturbances in the alignment. In this embodiment, the alignment is performed using points (partial point cloud) in the point cloud that are in a range higher than the height of the moving object, thereby preventing a decrease in the accuracy of the alignment due to the moving object. The partial point cloud is a collection of points where the surrounding features are measured. However, since the partial point cloud has few features in the height direction, it is not used for vertical alignment, but is used for horizontal alignment. The vertical alignment is performed separately after the horizontal alignment. This makes it possible to use points that are horizontally aligned for vertical alignment, and alignment can be performed with high accuracy. In particular, since the road surface is measured at high density, it is suitable to use points where the road surface is measured for vertical alignment. In view of this, alignment can be performed with high accuracy by using points on or near the movement trajectory (road surface point cloud) for vertical alignment.

[0031] FIG. 3 is a diagram for explaining alignment in the horizontal direction. The portion of the point cloud shown in FIG. 3(a) obtained on the side of the road includes surrounding objects such as buildings and road installations. For these, first, a representative height of the road surface is obtained for each block as a provisional road surface height. The representative height can be the minimum value, average value, or median value of the heights of points in the point cloud in the block whose three-dimensional positions or horizontal positions are within a predetermined distance from the moving trajectory. In this embodiment, the minimum height of points whose horizontal positions are within 1 meter from the center point is set as the provisional road surface height. A set of points within a certain reference height range from this provisional road surface height is extracted as a partial point cloud, and the horizontal position deviation amount (horizontal position deviation amount) is specified. The reference height range may be set to a height range whose lower limit is a position higher than a moving object on the road or sidewalk, such as the total height of a normal vehicle (including a bicycle or a motorcycle) or the height of a pedestrian or a resident along the road, and whose upper limit is a position several tens of centimeters to several meters higher than the lower limit. Special vehicles such as high-altitude work vehicles and crane trucks during work may be excluded from the normal vehicles referred to here. The reference height range may be changeable depending on the situation of the surrounding features. As for the horizontal position range, instead of extracting the partial point cloud from all points in the block, a range wide enough to ensure the accuracy of alignment may be appropriately set as the source of the partial point cloud, for example, by extracting the partial point cloud from points within a certain horizontal distance from the moving trajectory in the block. In this way, the partial point cloud set in a position range higher than the moving object is less likely to be obstructed by the moving object. Therefore, the horizontal position deviation amount is calculated with high accuracy based on the surrounding features.

[0032] In Fig. 3(a), the black stripe area among the points is the partial point cloud of the reference height range of the reference block, and the dark gray stripe area (partially overlapping with the black points) is the partial point cloud of the reference height range of the block to be aligned.

[0033] Since the points measured in each block are not necessarily at the same position, it is not possible to simply align the points belonging to the partial point groups of each block. On the other hand, the closest points in the strip-like regions of both blocks generally represent the surface positions of the same structure. In this embodiment, for example, an ICP (Iterative Closest Point) matching algorithm is used to calculate the horizontal position shift amount. In ICP matching, (1) a point belonging to the partial point group on the correction target side that is closest to each point of the partial point group on the reference side is searched for. (2) A movement amount for bringing these searched points as a whole closest to each point of the partial point group on the reference side that corresponds to them is calculated. After the partial point group on the correction target side is moved by the calculated movement amount, the above processes (1) and (2) are repeated to asymptotically converge the movement amount to an optimal value.

[0034] The amount of movement calculated by ICP matching is generally the sum of translation and rotation. However, in measurements performed by moving a measuring instrument while performing satellite positioning and attitude measurement at any time, the amount of rotation is often so small that it can be ignored. Therefore, the amount of movement may be calculated only for translation. In addition, the vertical component (height) is not adjusted here. Therefore, the Z component of the position data may simply be ignored and data processing may be performed in two dimensions, or the Z component may be temporarily set to zero and calculation may be performed using a three-dimensional data processing algorithm. That is, the amount of translation (Txi, Tyi) (i is a variable representing the number of processing times) obtained in each iteration of the ICP matching algorithm is calculated as the amount of translation Tx=ΣTxi, Ty=ΣTyi (Σ is the sum for each i). The horizontal position (Xc, Yc) of each point after correction is expressed as Xc=X+Tx and Yc=Y+Ty, respectively, from the horizontal position (X, Y) of each point before correction. The correction targets include not only points within the blocks of the correction target section, but also the corresponding measurement positions constituting the movement trajectory.

[0035] The ICP matching algorithm requires a large amount of processing, and therefore the number of points to be processed must be as small as possible. The above-mentioned reference height range is appropriately determined based on the relationship between the number of points for which the horizontal deviation amount can be obtained with high accuracy and the amount of processing. In particular, since a sufficient number of points representing the shape of surrounding features can often be obtained within a height range of several tens of centimeters to several meters, by setting an upper limit height for the reference height range, the number of points to be processed can be easily and appropriately reduced, and the amount of processing can be reduced. The reference height range may be variably determined depending on the situation.

[0036] In the example of horizontal positional deviation shown in FIG. 3(b), it can be seen that a deviation of road markings, etc. occurs in a direction diagonally at approximately 40 degrees to the right of the road extension direction X. The above-mentioned ICP matching algorithm identifies the amount of positional deviation in two directions in the horizontal plane. If there are not enough surrounding features within the reference height range and it is difficult to identify the amount of horizontal positional deviation with high accuracy, the process of aligning the correction target point group is terminated with an error. In this case, the calculation of the height deviation amount and the alignment thereafter are also canceled.

[0037] Once the horizontal position deviation amount is calculated, the horizontal position of each point in the block to be corrected is corrected and aligned using the horizontal position deviation amount. After that, the height deviation amount of the block to be corrected is calculated. The height deviation amount is calculated using multiple points on the road surface (hereinafter, road surface point cloud). In this case, the positions of the road surface point clouds do not match. Since the height deviation amount is a one-dimensional scalar value, for example, the difference between a representative value such as the average value of the heights of the road surface point clouds that are simply extracted may be calculated. The deviation between the height of the correction target section and the height of the reference section is adjusted by correcting this difference so as to match the height of the road surface in the correction target section to the height of the reference section.

[0038] The road surface point clouds on the road surface of the reference section and the correction target section may be set, for example, within a circle of a predetermined radius (common area with a common horizontal position) with a point on the movement trajectory in the overlapping part of both blocks as the center point (road surface center point) in a planar view. The road surface center point may be set, for example, at a distance of half the length along the movement trajectory from the start or end of the movement trajectory in the common part of both blocks. Within the range of this common area in a planar view, points within the range of the vertical reference difference (predetermined height) based on the provisional road surface height may be extracted. The common area and the vertical reference difference are determined so as to exclude points that are not on the road surface, that is, moving objects such as surrounding vehicles and people. For example, the predetermined radius may be one meter. Also, for example, the vertical reference difference may be set to a range from one meter below the provisional road surface height to 20 centimeters above the provisional road surface height. Points within a cylinder defined by these vertical reference differences, the radius of the circle, and the position of the road surface center point are extracted as candidates for the road surface point cloud.

[0039] In this embodiment, a predetermined number of points may be extracted from the extracted points in ascending order of height. A high-height point on a local road surface may be obtained by measuring the surface of fallen leaves or gravel, for example. By excluding points that do not properly measure the road surface in this way, the accuracy of alignment is improved. The predetermined number may be, for example, 10. In order to increase the accuracy that the road surface is measured without measuring a moving object in the common area, a predetermined number of points must be obtained from both the reference section and the correction target section.

[0040] FIG. 4 is a diagram for explaining setting of an extraction range of a road surface point cloud for calculating a height deviation amount. In the above-mentioned extraction of road surface points, if a moving object such as a vehicle or a person exists near the set road surface center point, a sufficient number of points may not be obtained from within a circle (R1) of a predetermined radius centered on the road surface center point. If the number of extracted points is less than the above-mentioned predetermined number, the above-mentioned predetermined radius may be expanded first and the extraction process may be performed to increase the number of points to be extracted. The expansion of the radius may be, for example, 1.5 times (R2) to 2.0 times (R3) the above-mentioned predetermined radius. The expansion of the radius may be performed once, or may be repeated multiple times while changing the amount of expansion.

[0041] If the number of extracted points does not reach the predetermined number even if the radius is expanded to the upper limit, the center position of the extraction range may be shifted horizontally from the center point of the road surface and the extraction process may be performed again. The direction in which the center position is shifted may be, for example, only the front-rear direction, or may be four directions including front-rear, left-right, and diagonal directions. For example, it may be determined in order whether a predetermined number of points or more can be obtained for a circle R4 whose center position is shifted by a predetermined radius in the traveling direction, and a circle R5 whose center position is shifted by a predetermined radius in the opposite direction to the traveling direction. If the predetermined number of points or more cannot be obtained for either circle, the shift width of the center position may be sequentially expanded to two or three times the predetermined radius, as in circles R6, R7, R8, and R9, until the predetermined number of points or more can be obtained. An upper limit value may be specified for the shift width of the center position. The upper limit value (predetermined distance) of the shift width of the center position is determined within a range in which the position of the point is identified with the required accuracy. The laser light for road surface measurement is most densely emitted directly below the laser light emission position in the vertical direction, and the accuracy decreases as the distance from this point increases. Therefore, a large distance in the Y direction can decrease the accuracy of identifying the amount of positional deviation in the vertical direction.

[0042] If the number of points is not sufficient to extract the desired number of points even after changing the extraction range, it is determined that a registration error has occurred and the calculation of the positional deviation amount for this block is discontinued. In this case, the positional deviation amount of the previous block may be used as the estimated positional deviation amount.

[0043] FIG. 5 is a diagram for explaining an example of height adjustment. In the perspective view of Fig. 5(a), for example, the light gray points are the positions of the reference section measured by the scanner, and the dark gray points are the positions of the correction target section measured by the scanner. The black points closely arranged within the circle R1 are the positions of the reference section and the correction target section measured by the MMS. The grid lines represent the X and Y directions, respectively.

[0044] FIG. 5(b) is a diagram showing the height Z of each point in the circle Rr in the Y direction. The range indicated by the two dotted lines is the height of each point in the circle R1. This result shows that the height of the correction target section tends to be measured higher than the height of the reference section. A predetermined number of points are further extracted from the circle R1 to obtain the heights of the road surface in the reference section and the correction target section in the common area, and the difference between these is calculated as the height deviation amount.

[0045] 6 is a flowchart showing a control procedure of the point cloud position correction process executed by the control unit 11 of the information processing device 1 of this embodiment. The point cloud position correction process including the point cloud processing method of this embodiment is executed by, for example, reading out the program 120 in response to a predetermined input operation to the operation reception unit 15. The input operation may be received by a GUI (Graphical User Interface) as a part of a processing command in application software that displays the measurement result. The point cloud (for example, a point cloud file) that the user wishes to correct among the measurement data 201 may be specified by the input operation, or may be read by generating a setting data file in advance and storing it in a specified folder or the like.

[0046] The control unit 11, as an acquisition means, acquires a point cloud from the measurement data 201, and also acquires the movement trajectory data 202 (S1). The acquisition means sets blocks at predetermined distances along the movement trajectory (S2).

[0047] The acquisition means selects a block with good measurement accuracy from among the set blocks, and sets a reference section (S3). The acquisition means sets the reference section in response to an input operation (selection operation) by the user. This reference section is a section for which the user has confirmed in advance that the measurement accuracy is good, or a section for which the user has manually adjusted the positional deviation of the point cloud in advance. The acquisition means selects a block having an overlapping portion with the reference section, and sets a section to be corrected (S4). The point cloud of the reference section is the reference point cloud. The point cloud of the section to be corrected is the point cloud to be corrected.

[0048] The control unit 11, as a horizontal position correction means, obtains tentative road surface heights of each point cloud in the reference section and the correction target section, and extracts a partial point cloud within the reference height range (S5). The horizontal position correction means calculates the amount of horizontal positional deviation of the partial point cloud in the correction target section from the partial point cloud in the reference section based on an ICP matching algorithm or the like, and corrects the horizontal positions of each point in the correction target section and the measurement position (S6).

[0049] The control unit 11, as a vertical position correction means, specifies a road surface point cloud further extracted from the road surface points located within the cylindrical range of the common area from each point cloud of the reference section and the correction target section (S7). The vertical position correction means calculates the height deviation amount of the correction target section from the reference section, and corrects the vertical position (height) of the correction target section (S8).

[0050] The control unit 11 determines whether or not there are any uncorrected blocks remaining (S9). If it is determined that there are no uncorrected blocks remaining (S9; NO), the control unit 11 ends the point cloud position correction process.

[0051] If it is determined that there are uncorrected blocks remaining (S9; YES), the acquisition means sets a correction target section that overlaps with the range of the uncorrected and corrected blocks, and sets the range of the corrected blocks that overlaps with this correction target section as a reference section (S10). Basically, the reference section may be the block in which the correction target section was set in the previous process, and the new correction target section may be the range of the block on the opposite side of the reference section of the block in which the correction target section was set in the previous process. Then, the processing of the control unit 11 returns to processing S5.

[0052] Of the procedures described with reference to the flowchart in Fig. 6, steps S1 to S4 and S10 are included in the acquisition step in the point cloud processing method of this embodiment. Steps S5 to S6 are included in the horizontal position correction step in the point cloud processing method of this embodiment. Steps S7 to S8 are included in the vertical position correction step in the point cloud processing method of this embodiment.

[0053] In the above, the correction of positional deviation in the direction along the movement trajectory related to the measurement has been described. When the road has many lanes or the road is divided between opposing lanes, the same road may be measured multiple times for one or more lanes. In this case, alignment is required between point clouds from different times. Each measurement is performed so that there is an overlap with at least one other measurement.

[0054] FIG. 7 is a diagram for explaining alignment between lanes. For example, after alignment along the trajectory W1 in one lane is completed in the range Ac, alignment along the trajectory W2 in the other lane is performed in sequence. Center points C4 to C6 and blocks A4 to A6 are set. The setting may be performed so that there is an overlap between the blocks A4 to A6, as in the alignment along the trajectory W1. However, in this case, there does not necessarily have to be an overlap between the blocks A4 to A6. Here, the trajectory W1 and the trajectory W2 face in opposite directions, but when alignment is performed between multiple lanes heading in the same direction, the trajectory W1 and the trajectory W2 face in the same direction.

[0055] Each of the blocks A4 to A6 overlaps with the aligned range Ac at an overlapping portion Dc. Therefore, when the range of each of the blocks A4 to A6 is set as the correction target section, the reference section can always be set to the range Ac. In other words, when aligning a lane different from the aligned lane, the alignment may be performed using the point cloud of the aligned lane as the reference point cloud. The alignment procedure is the same as that shown in FIG. 6, so a detailed description will be omitted.

[0056] When performing alignment for three or more measurements on a road with many lanes, the order of alignment may be determined as appropriate. For example, the first alignment may be performed along a movement trajectory related to a measurement that is determined by the user visually checking the data or by calculation by the control unit 11 to provide the most stable and accurate positioning results, and alignment along other movement trajectories may be performed using the range of blocks related to the movement trajectory that was aligned first as a reference section.

[0057] In addition, the alignment of point clouds obtained by multiple measurements having overlapping parts for the same lane can be performed in a similar manner. In this case, for example, after alignment of one of two point clouds having overlapping parts, a first reference point cloud is set in the overlapping part of the one point cloud, and a first correction target point cloud is set in the overlapping part of the other point cloud, and the alignment of the other point cloud may be performed. In addition, for example, after alignment of one point cloud and the other point cloud, a partial point cloud and a road point cloud may be set in the overlapping part of both point clouds, with one point cloud as the reference point cloud and the other point cloud as the correction target point cloud, and alignment may be performed.

[0058] The present invention is not limited to the above-described embodiment, but may be modified in various ways. For example, in the above horizontal alignment, the amount of movement in a two-dimensional plane is calculated by ignoring the height direction component in ICP matching, but this is not limited to this. The amount of movement may be calculated three-dimensionally in ICP matching, and only the amount of movement in the XY direction components may be reflected in the horizontal alignment.

[0059] Alternatively, other techniques than ICP matching, such as LS3D, may be used for horizontal alignment.

[0060] In the above description, the road surface point cloud is extracted and specified from a cylindrical region during vertical alignment, but this is not limited to the above. A prism region may be set. In this case, the prism may be a square in plan view, or a rectangle with a long side parallel to the direction of the movement trajectory.

[0061] In addition, the axial position of the cylindrical region determined during vertical alignment, i.e., the center position of the circular region in plan view, does not have to be determined on the movement trajectory. For example, the center position may be determined based on a random number within a strip-shaped region within a predetermined distance (within a distance predetermined to be about the width of the vehicle) from the movement trajectory in the overlapping portion. Alternatively, the center position may be determined manually on the road surface of the overlapping portion in response to an input operation to the operation reception unit 15.

[0062] Furthermore, the distance interval (the above-mentioned predetermined distance) between the center points when setting the reference section and the correction section may be appropriately determined. These settings may be performed, for example, based on the evaluation results obtained by a user visually checking the data to evaluate the magnitude and interval of fluctuations in the positioning results.

[0063] Furthermore, the point cloud of the measurement data 201 may be a point cloud measured in whole or in part by a method other than a laser scanner. For example, the point cloud may be a point cloud measured in whole or in part by multi-viewpoint image measurement in which a point cloud is obtained by performing SfM (Structure from Motion) processing on a plurality of images of the road surface and surrounding features continuously captured while a camera is moved. Alternatively, the point cloud may be a point cloud measured by measuring the road surface by a light section method and measuring the surrounding features by a laser scanner.

[0064] In the above description, the storage unit 12 including a non-volatile memory such as an HDD or a flash memory is used as an example of a computer-readable medium for storing the program 120 related to the control of the measurement position correction of the present invention, but is not limited to this. As other computer-readable media, other non-volatile memories such as MRAM and portable recording media such as CD-ROM and DVD disks can be applied. In addition, a carrier wave is also applied to the present invention as a medium for providing the data of the program according to the present invention via a communication line. In addition, the specific configurations, contents and procedures of the processing operations, etc. shown in the above embodiments can be modified as appropriate without departing from the spirit of the present invention. The scope of the present invention includes the scope of the invention described in the claims and its equivalents.

[0065] As described above, the information processing device 1, which is the point cloud processing device of this embodiment, includes the control unit 11. The control unit 11, as an acquisition means, acquires a reference point cloud and a correction target point cloud obtained by measuring the road or sidewalk surface and the surrounding features of the road surface with overlapping portions. The control unit 11, as a horizontal position correction means, extracts partial point clouds from the reference point cloud and the correction target point cloud, respectively, that are in a range higher than the moving objects on the road or sidewalk, and aligns the horizontal position of the correction target point cloud to the position of the reference point cloud by comparing the partial point clouds. The control unit 11, as a vertical position correction means, matches the heights of common areas with common horizontal positions in the reference point cloud and the correction target point cloud aligned by the horizontal position correction means. In this way, by performing two-stage alignment, namely horizontal position alignment using a partial point cloud at a high position and height alignment after the horizontal position alignment, the information processing device 1 can perform accurate alignment even for a point cloud obtained by measuring a road or sidewalk on which a moving object exists. Therefore, the information processing device 1 can correct the position of the point cloud with higher accuracy.

[0066] The vertical position correction means may set a common area on the road surface. By performing vertical position alignment using points on the road surface that are measured at high density, it becomes possible to correct the positions of the point cloud with higher accuracy.

[0067] The acquisition means also acquires the reference point cloud and the correction target point cloud measured by a measuring instrument moving on a road or sidewalk in association with a measurement position, which is the position of the measuring instrument at the time of measurement. The vertical position correction means may set a common area within a predetermined distance from the measurement position. Setting the common area within a predetermined distance from the measurement position increases the accuracy of setting the common area on the road surface. Therefore, the information processing device 1 can perform highly accurate alignment using points on the road surface that are measured at high density.

[0068] The vertical position correction means may set a common area in which a predetermined number of points constituting the reference point group and a predetermined number of points constituting the correction target point group are included within a predetermined height. If the number of points within the predetermined height is insufficient in the common area in which the heights of the reference point group and the correction target point group match, it is considered that the common area does not properly measure the road surface. Therefore, by setting a common area in which a predetermined number of points are included within a predetermined height, a common area that properly measures the road surface can be determined, thereby enabling accurate alignment.

[0069] The point cloud processing method of this embodiment also includes the following steps: (1) an acquisition step of acquiring a reference point cloud and a correction target point cloud measured with overlapping areas of the road or sidewalk surface and surrounding features on the road surface; (2) a horizontal position correction step of extracting partial point clouds from the reference point cloud and the correction target point cloud, respectively, that are in a range higher than moving objects on the road or sidewalk, and comparing the partial point clouds to align the horizontal position of the correction target point cloud to the position of the reference point cloud; (3) a vertical position correction step of matching the heights of common areas with common horizontal positions in the reference point cloud and the correction target point cloud aligned in the horizontal position correction step. According to this point cloud processing method, by performing two-stage alignment, namely horizontal position alignment using a partial point cloud at a high position, and height alignment after the horizontal position alignment, it is possible to perform accurate alignment even for point clouds measured on roads or sidewalks on which moving objects exist. Therefore, this point cloud processing method can correct the positions of point clouds with higher accuracy.

[0070] Moreover, by executing the program 120 relating to the point cloud processing method, the positions of the point cloud can be easily corrected with high accuracy by a general-purpose computer. [Explanation of symbols]

[0071] 1. Information processing device 11 Control section 12 Storage section 120 Programs 13 Input / Output Interface 131 Connection terminal 132 Communications Department 14 Display section 15 Operation reception section 21 Database device 22 Optical reading device 201 Measurement Data 202 Movement trajectory data

Claims

1. An acquisition means for acquiring a reference point cloud and a correction target point cloud obtained by measuring a road or sidewalk surface and surrounding features of the road surface with overlapping portions therebetween; a horizontal position correction means for extracting partial point clouds from the reference point cloud and the correction target point cloud in a range higher than a moving object on the road or sidewalk, and for comparing the partial point clouds with each other to align a horizontal position of the correction target point cloud with a position of the reference point cloud; a vertical position correction means for matching heights of a common area having a common horizontal position in the reference point group and the correction target point group aligned by the horizontal position correction means; A point cloud processing device comprising:

2. The point cloud processing apparatus according to claim 1 , wherein the vertical position correction means sets the common area on the road surface.

3. the acquisition means acquires the reference point cloud and the correction target point cloud measured by a measuring instrument moving on the road or sidewalk in association with a measurement position which is a position of the measuring instrument at the time of measurement; The point cloud processing apparatus according to claim 2 , wherein the vertical position correction means sets the common area within a predetermined distance from the measurement position.

4. 4. The point cloud processing device according to claim 2, wherein the vertical position correction means sets the common area in which a predetermined number or more of points constituting the reference point cloud are included within a predetermined height, and the predetermined number or more of points constituting the correction target point cloud are included.

5. an acquisition step of acquiring a reference point cloud and a correction target point cloud measured while the road or sidewalk surface and surrounding features of the road surface have overlapping portions; a horizontal position correction step of extracting partial point clouds from the reference point cloud and the correction target point cloud in a range higher than a moving object on the road or sidewalk, respectively, and comparing the partial point clouds to align the horizontal position of the correction target point cloud with the position of the reference point cloud; a vertical position correction step of matching heights of common areas having common horizontal positions in the reference point group and the correction target point group aligned in the horizontal position correction step; A point cloud processing method comprising:

6. Computer, An acquisition means for acquiring a reference point cloud and a correction target point cloud measured while a road or sidewalk surface and surrounding features of the road surface have overlapping portions; a horizontal position correction means for extracting partial point clouds from the reference point cloud and the correction target point cloud in a range higher than a moving object on the road or sidewalk, respectively, and for aligning a horizontal position of the correction target point cloud with a position of the reference point cloud by comparing the partial point clouds; a vertical position correction means for matching heights of common areas having common horizontal positions in the reference point group and the correction target point group aligned by the horizontal position correction means; A program that functions as a

Images