Information processing device, information processing method, and program

JP2024162856APending Publication Date: 2024-11-21NEC CORP

Patent Information

Application Number
JP2023078793
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-11
Publication Date
2024-11-21

AI Technical Summary

Benefits of technology

【0011】 本開示は、以上のように構成されることにより、三次元点群データから精度よくノイズを除去することができる。

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024162856000001_ABST
    Figure 2024162856000001_ABST
Patent Text Reader

Abstract

To accurately remove noises from three-dimensional point group data.SOLUTION: An information processing device 100 includes: a classification unit that classifies three-dimensional point group data including a distance value to an object in a predetermined area into one or more clusters based on the distance value; and a determination unit that determines a cluster to be adopted, based on a distribution of the distance value of the three-dimensional point group data included in each of the classified clusters.SELECTED DRAWING: Figure 10
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present disclosure relates to an information processing device, an information processing method, and a program. [Background technology]

[0002] It is generally known that 3D point clouds acquired by LiDAR (Light Detection And Ranging) contain noise point clouds.Causes of noise point clouds include those due to the distance measurement accuracy of the LiDAR device, objects moving during measurement, reflections on metal surfaces, transmission through glass surfaces, and erroneous distance measurement due to noise signals being recognized as reflected pulses due to factors such as low light intensity when the light pulse transmitted from the LiDAR device is reflected from the target object and received.

[0003] In particular, when the received light intensity is low, it is believed that erroneous ranging occurs through the following process. In a LiDAR device, the received reflected light pulse is photoelectrically converted by an optical receiver inside the device and processed as an electrical signal. The calculation unit of the LiDAR device calculates the distance to the target object from this electrical signal. However, when the received reflected light intensity is low, noise signals from the device or external factors are added to the small signal waveform, distorting the signal waveform, which can cause erroneous ranging or reduced ranging accuracy.

[0004] On the other hand, such noise point clouds can be removed by a point cloud processing method such as an isolated point removal filter. For example, Patent Document 1 proposes a method of removing points from point cloud data whose Euclidean distance to neighboring points is equal to or greater than a threshold value as noise points. Also proposed is a method of clustering point clouds based on the variation in laser reflection intensity, and isolating and removing reflection points other than those of the target object. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Patent Publication 2018-173749 Summary of the Invention [Problem to be solved by the invention]

[0006] However, in the technology described in the above-mentioned prior art document, in a situation where noise point groups are continuous and dense in an area where a large number of noise point groups are likely to occur, the problem occurs that such noise point groups cannot be completely removed and remain. In addition, in the clustering method based on the reflection intensity of the laser, it is difficult to distinguish noise point groups caused by the distance measurement accuracy of the LiDAR device, and in the case of measuring a distant object, the reflection signal from the object and the noise signal are both weak, so it is difficult to classify the two. Thus, the technology in the prior art document has a problem that it is difficult to accurately remove noise from three-dimensional point cloud data.

[0007] Therefore, an object of the present disclosure is to provide an information processing device that can solve the above-mentioned problem that it is difficult to remove noise from three-dimensional point cloud data with high accuracy. [Means for solving the problem]

[0008] An information processing device according to an embodiment of the present disclosure includes: A classification unit that classifies three-dimensional point cloud data including distance values ​​to objects in a predetermined region into one or more clusters based on the distance values; a determination unit that determines the cluster to be adopted based on a distribution of the distance values ​​of the three-dimensional point cloud data included in the classified cluster; Equipped with The structure is as follows.

[0009] In addition, an information processing method according to an embodiment of the present disclosure includes: classifying three-dimensional point cloud data including distance values ​​to objects in a predetermined region into one or more clusters based on the distance values; determining the cluster to be adopted based on a distribution of the distance values ​​of the three-dimensional point cloud data included in the classified cluster; The structure is as follows.

[0010] In addition, a program according to an embodiment of the present disclosure includes: classifying three-dimensional point cloud data including distance values ​​to objects in a predetermined region into one or more clusters based on the distance values; determining the cluster to be adopted based on a distribution of the distance values ​​of the three-dimensional point cloud data included in the classified cluster; Have a computer carry out the process, The structure is as follows. Effect of the Invention

[0011] With the above-described configuration, the present disclosure can accurately remove noise from three-dimensional point cloud data. [Brief description of the drawings]

[0012] [Figure 1] FIG. 2 is a diagram showing how three-dimensional point cloud data is acquired. [Diagram 2] FIG. 2 is a diagram showing how three-dimensional point cloud data is acquired. [Diagram 3] FIG. 1 is a block diagram showing a configuration of a point cloud processing device according to a first embodiment of the present disclosure. [Figure 4] FIG. 4 is a diagram showing a state of processing by the point cloud processing device disclosed in FIG. [Diagram 5] FIG. 4 is a diagram showing a state of processing by the point cloud processing device disclosed in FIG. [Figure 6] 4 is a flowchart showing the operation of the point cloud processing device disclosed in FIG. 3. [Figure 7] FIG. 11 is a diagram showing a processing state of a point cloud processing device according to a second embodiment of the present disclosure. [Figure 8] 11 is a flowchart showing the operation of a point cloud processing device according to a second embodiment of the present disclosure. [Figure 9] FIG. 11 is a block diagram showing a hardware configuration of an information processing device according to a third embodiment of the present disclosure. [Figure 10]FIG. 11 is a block diagram showing a configuration of an information processing device according to a third embodiment of the present disclosure. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] <Embodiment 1> A first embodiment of the present disclosure will be described with reference to Fig. 1 to Fig. 6. Fig. 1 and Fig. 2 are diagrams for explaining how three-dimensional point cloud data is acquired. Fig. 3 is a diagram for explaining the configuration of a point cloud processing device, and Fig. 4 to Fig. 6 are diagrams for explaining the processing operation of the point cloud processing device.

[0014] The point cloud processing device in the present disclosure is used to remove point clouds that may be determined as noise point clouds from three-dimensional point cloud data including distance values ​​to an object acquired using a measurement technique such as LiDAR (Light Detection And Ranging). Here, LiDAR is a technique for irradiating a laser light to an object and measuring the distance to the object and the shape of the object based on the information of the reflected light. Therefore, using LiDAR, three-dimensional point cloud data including distance values ​​to each point located on the surface of the object can be acquired, and can be used to grasp the shape of structures such as terrain and buildings. However, the three-dimensional point cloud data is not limited to being acquired by LiDAR, and may be acquired by measuring and acquiring using any measurement device.

[0015] Here, with reference to Figs. 1 and 2, a manner in which three-dimensional point cloud data is acquired using a LiDAR device will be described, along with a situation in which noise point clouds are likely to occur when acquiring three-dimensional point cloud data.

[0016] As shown in Fig. 1, the laser light emitted from the LiDAR device A1 has a finite beam diameter, and depending on the direction of scanning by the LiDAR, a situation may occur in which the laser light is irradiated onto objects at different distances located at the edge of the object, etc. Here, the locations indicated by the symbols T1 and T2 in Fig. 1 indicate the measurement objects located in front of and behind the measurement object, respectively.

[0017] The LiDAR device A1 then irradiates the objects T1 and T2 with a ranging pulse laser. Here, the symbols B1 and B2 in FIG. 1 indicate the areas illuminated by the ranging pulse laser irradiated from the LiDAR device A1 on the objects T1 and T2, respectively. At this time, the surface indicated by the symbol T3 in FIG. 1 is approximately parallel to the line of sight direction, which is the direction of irradiation of the laser light of the LiDAR device A1, and is a blind spot for the LiDAR device A1, so it is an area where a point cloud based on the ranging of the LiDAR device A1 cannot be obtained. Note that the symbol B3 in FIG. 1 is an area on the surface indicated by the symbol T3, and is an area where there is a difference in distance between the objects T1 and T2 to be measured on the front side and the back side.

[0018] In the situation shown in FIG. 1, the reflected pulse light is incident on the LiDAR device A1 independently from each of the objects T1, T2, and T3, so the received light intensity is reduced. Since the accuracy of distance measurement is reduced due to the reduction in the light intensity of the received pulse, the area around the object T3 indicated by the symbol B3 is particularly prone to noise point clouds. In addition, since the beam diameter of the distance measurement pulse generally increases as the distance from the LiDAR device A1 increases, the occurrence rate of noise point clouds increases as the distance measurement is performed at a longer distance. Here, FIG. 2 illustrates an example of a point cloud obtained by the LiDAR device A1. At this time, the symbols P1, P2, and P3 shown in FIG. 2 are schematic diagrams of the point clouds of the measurement objects T1, T2, and T3, respectively. The symbol C1 shown in FIG. 2 indicates the shape of the laser light with the LiDAR device A1 at its apex, and indicates a cone-shaped area. In this embodiment, as described below, the distribution of distance values ​​for the point clouds in the cone-shaped area is used to remove point clouds that can be determined to be third-order noise point clouds.

[0019] [composition] The point cloud processing device 1 in this embodiment is composed of one or more information processing devices each having a calculation device and a storage device. As shown in Fig. 3, the point cloud processing device 1 includes a point cloud extraction means 11, a point cloud classification means 12, and a point cloud generation means 13. Each function of the point cloud extraction means 11, the point cloud classification means 12, and the point cloud generation means 13 can be realized by the calculation device executing a program for realizing each function stored in the storage device. Each configuration will be described in detail below.

[0020] The point cloud extraction means 11 has a function of receiving input of point cloud data from a point cloud input device (not shown) and extracting and cutting out the point cloud within a specified three-dimensional region. The input point cloud data is three-dimensional point cloud data including distance values ​​to an object measured using a measurement device such as the LiDAR device A1, as described with reference to Figs. 1 and 2. The point cloud extraction means 11 specifies a cone-shaped region, such as a cone region having a certain solid angle with the LiDAR device A1 as the apex, and cuts out the point cloud. The shape of the specified region is arbitrary, and may be specified as a rectangular parallelepiped region or a spherical region of infinite length in the distance direction.

[0021] The point cloud classification means 12 (classification unit) receives the point cloud from the point cloud extraction means 11, performs clustering on the point cloud, and classifies each point cloud into at least one cluster so that it belongs to one of the clusters. The clustering method classifies the point cloud using a classification method such as the k-means method. More specifically, the classification method may be, for example, a classification method based on the Euclidean distance using three-dimensional position information of the point cloud, or a classification method based on the distance in the line of sight with the LiDAR device A1 as the reference. In particular, in this embodiment, a classification method based on the distribution of the point cloud based on a histogram of the distance values ​​of the point cloud with the LiDAR device A1 as the reference is used.

[0022] Here, FIG. 4 shows an example of a histogram of distance values ​​of point clouds. The histogram in FIG. 4 shows a histogram of distance values ​​of point clouds P1, P2, and P3 cut out by the cone-shaped region C1 in FIG. 2. The point cloud classification means 12 sets clusters of point clouds according to distance values ​​using the histogram of distance values. For example, point clouds whose distance values ​​are located within a predetermined range from each other and point clouds whose distance values ​​exist continuously are set as one cluster. In the example of FIG. 4, the point clouds are classified into clusters indicated by symbols H11, H12, and H13 with the distance value at which the frequency becomes 0 as a boundary. In the example shown in FIG. 2, the distances to objects T1, T2, and T3 are different from each other, and the distance values ​​of each point cloud are concentrated near the distance to each object, so that they can be classified into clusters H11, H12, and H13 as shown in FIG. 4. Note that cluster H11 can correspond to object T1, cluster H12 to object T2, and cluster H13 to object T3.

[0023] The point cloud generating means 13 (decision unit) receives the point clouds classified into each cluster from the point cloud classification means 12, and determines whether to adopt or not adopt the point clouds belonging to each cluster. That is, the point cloud generating means 13 determines whether to adopt the point clouds belonging to a cluster as the point cloud of the target object, or to reject, i.e., remove, them as noise point clouds. When adopting a cluster, the point cloud generating means 13 may recalculate the three-dimensional position information of the point cloud and generate a new point cloud.

[0024] Specifically, the point cloud generating means 13 judges whether to adopt or not adopt a point cloud belonging to a cluster, for example, according to the distribution of the distance values ​​of the input point cloud of the cluster. Here, the distance values ​​of the point cloud indicate variations due to the distance measurement accuracy of the LiDAR device A1 and other measurement conditions, and the shape of the histogram of the distance values ​​of the point cloud usually shows a Gaussian distribution shape centered on the distance value to the measurement object. On the other hand, when the distance measurement result is calculated from a noise signal, the distance values ​​show a random distribution, so the shape of the histogram of the distance values ​​of the point cloud shows a certain frequency distribution regardless of the distance. Therefore, it is possible to judge whether to adopt the point cloud as the point cloud of the object or not to adopt it as the noise point cloud from the shape of the histogram of the distance values ​​of the point cloud.

[0025] 4, the shape of the histogram of distance values ​​in clusters H11 and H12 shows a Gaussian approximation curve with a peak. Therefore, the point cloud generation means 13 judges that the clusters H11 and H12, which have a Gaussian distribution shape in the histogram of distance values, are clusters of point clouds of the target object, and adopts the point clouds of such clusters H11 and H12. On the other hand, in FIG. 4, the histogram of distance values ​​in cluster H13 shows a certain frequency distribution shape. Therefore, the point cloud generation means 13 judges that the cluster H13 is a cluster of noise point clouds, and does not adopt the point cloud of such cluster H13.

[0026] Also, the point cloud generating means 13 may determine whether or not to adopt a point cloud belonging to a cluster depending on whether the number of point clouds in the cluster, that is, the frequency of distance values, is equal to or greater than a threshold value. Here, it is assumed that the number of point clouds classified as a cluster of objects is greater than the number of point clouds classified as a cluster of noise points. Therefore, the point cloud generating means 13 may adopt the point cloud of a cluster when the number of point clouds included in the cluster is equal to or greater than a preset threshold value Th1. Also, in the histogram of the distance values ​​of the point clouds as shown in FIG. 4, the point cloud of the cluster of the object shows a frequency peak at a specific distance value. Therefore, the point cloud generating means 13 may adopt the point cloud of the cluster when the number of point clouds included in a width dR indicating a range of a predetermined distance value in the cluster is equal to or greater than a threshold value Th2 in the histogram of the distance values ​​of each cluster.

[0027] In addition, in determining the point cloud to be adopted, due to the principle of LiDAR, the area behind the measurement target is a blind spot area, and the point cloud of the object existing in that area cannot be acquired. For this reason, the point cloud generating means 13 may extract only the cluster closest to the LiDAR device A1 from among the clusters determined to be adopted as described above.

[0028] In this way, the point cloud generating means 13 uses the histogram of distance values ​​of the point cloud data and the frequency of distance values ​​to determine whether to accept or reject the point cloud for each cluster. In other words, the point cloud generating means 13 determines whether the point cloud is an object or noise based on the distribution of distance values ​​for each cluster, making the determination easy.

[0029] The point cloud generating means 13 may generate a new point cloud from the point cloud of the cluster determined to be adopted. For example, the point cloud generating means 13 may extract the center of gravity of the cluster as a representative point and generate a new point cloud. Here, the point cloud acquired by LiDAR may generate points before and after the distance at which the target object actually exists due to its distance measurement accuracy. Since the distribution of the distance values ​​of this point cloud shows a normal Gaussian distribution centered on the distance at which the object exists when randomly extracted, when the point cloud generating means 13 recalculates the distance, it is possible to adopt the average value of the distance values ​​for the extracted cluster, the center of gravity of the point cloud, or the Gaussian center position obtained by regression analysis. In this way, it is expected that the accuracy of the point cloud data obtained by LiDAR can be improved by calculating the coordinates of the point cloud to be adopted by regression analysis of the distribution of the point cloud and extraction of a representative value such as the average value. As a result, for example, the approximation curves L11 and L12 of the adopted clusters H11 and H12 shown in FIG. 4 are expected to show a Gaussian distribution with small variance like the approximation curves L21 and L22 shown in FIG. 5 by regenerating the point clouds.

[0030] [Operation] Next, the operation of the above-mentioned point cloud processing apparatus 1 will be described mainly with reference to the flowchart of FIG.

[0031] First, the point cloud extraction means 11 receives input of point cloud data from an input device (not shown) (step S1). Note that the point cloud data input in step S1 holds viewpoint information of the LiDAR device A1.

[0032] The point cloud extraction means 11 specifies an area for extracting a point cloud from the input point cloud data (step S2). Then, the point cloud extraction means 11 extracts a point cloud existing in the specified area and outputs it to the point cloud classification means 12 (step S3). For example, as shown in Fig. 2, the point cloud extraction means 11 specifies a cone-shaped area C1 with the LiDAR device A1 as the apex, and extracts a point cloud within the area C1.

[0033] The point cloud classification means 12 classifies the point cloud input from the point cloud extraction means 11 into at least one cluster, and outputs it to the point cloud generation means 13 (step S4). At this time, the point cloud classification means 12 sets the cluster of the point cloud according to the distance value, for example, using a histogram of the distance value. This allows the point cloud to be classified into clusters H11, H12, and H13, as shown in FIG.

[0034] The point cloud generating means 13 judges whether to adopt or not adopt the point cloud of each cluster for the clustered point cloud input from the point cloud classification means 12. For example, the point cloud generating means 13 judges whether to adopt or not adopt the point cloud belonging to the cluster according to the distribution of the distance values ​​of the point cloud of the input cluster. As an example, when the distance values ​​of the point cloud are histograms as shown in FIG. 4, the clusters H11 and H12 have a Gaussian distribution shape, so it is judged that they are clusters of point clouds of the target object, and the point clouds of the clusters H11 and H12 are adopted. On the other hand, the cluster H13 has a certain frequency distribution shape, so it is judged that the cluster H13 is a cluster of noise point cloud, and the point cloud of the cluster H13 is not adopted. Then, the point cloud generating means 13 calculates the representative points of the point cloud of the adopted cluster, generates it as a new point cloud, and outputs the generated point cloud to an output device not shown (step S5).

[0035] As described above, according to the point cloud processing device 1 of this embodiment, even if the point clouds are densely packed within the area extracted by the point cloud extraction means 11, it is possible to accurately remove point cloud data determined to be noise point clouds. In addition, by calculating position information for the point cloud data adopted by the point cloud generation means 13, it is possible to improve distance measurement accuracy.

[0036] <Embodiment 2> Next, a second embodiment of the present disclosure will be described with reference to Fig. 7 to Fig. 8. Fig. 7 to Fig. 8 are diagrams for explaining the processing operation of the point cloud processing device. Note that the point cloud processing device 1 in this embodiment has a configuration almost similar to that described in the above-mentioned first embodiment, so hereinafter, functions of the point cloud processing device 1 that are different from those in the above-mentioned first embodiment will be described mainly with reference to the flowchart in Fig. 8.

[0037] The point cloud processing device 1 in this embodiment performs noise removal on the entire point cloud by applying the processing on a certain specific region described in embodiment 1 to each part of the point cloud. Specifically, similar to the above-mentioned embodiment 1, the point cloud extraction means 11 first receives input of point cloud data (step S1), specifies a region from which the point cloud is to be extracted from the input point cloud data (step S2), and extracts the point cloud existing within the specified region (step S3).

[0038] Next, the point cloud classification means 12 classifies the point cloud input from the point cloud extraction means 11 into at least one or more clusters (step S4). Here, the point cloud to be clustered by the point cloud classification means 12 is clustered at least once. In a situation where the point cloud clustered by the point cloud classification means 12 is separated into multiple regions in terms of distance, the clustering may be performed in more detail.

[0039] FIG. 7 shows an example of a histogram of distance values ​​when two or more clustering steps are required. FIG. 7 shows a case where the point cloud is classified into three clusters H31, H32, and H33. In this case, when the object on the near side, which is the side closer to the LiDAR device A1, shows a complex structure, it is assumed that the distribution shows a distribution in which many Gaussian distributions overlap. For this reason, the cluster H31 located on the near side has a distribution shape of the histogram, that is, an approximation curve, in which multiple Gaussian distributions overlap, as shown in L311, L312, and L313, respectively. Moreover, the cluster H32 corresponds to the point cloud of the object on the far side, and the approximation curve of the histogram of the distance values ​​is as shown in L32. Note that the cluster H33 corresponds to a noise point cloud.

[0040] In the above-mentioned situation, whether or not more detailed clustering is required can be determined from the variance of the point cloud based on the distance measurement accuracy of the LiDAR. This is because each Gaussian distribution shown by the point cloud obtained by reflection from the target object shows a variance corresponding to the distance measurement accuracy. For this reason, the point cloud classification means 12 performs more detailed classification by clustering for each cluster classified in step S4 when the variance of the cluster is equal to or greater than a threshold value. In this way, when the point cloud classification means 12 determines that reclassification is required based on the distribution of distance values ​​of the point cloud included in the classified cluster, it reclassifies the point cloud of the cluster into more detailed clusters.

[0041] In the example of FIG. 7, the point cloud classification means 12 judges that the variance in the histogram of the cluster H31 is equal to or greater than the threshold value and that further clustering is required (Yes in step S4'), and clusters the points of the cluster H31 again (step S4). Note that the point cloud classification means 12 changes the clustering criteria when reclassifying. For example, the point cloud classification means 12 sets the points located within a narrower range of distance values ​​in the histogram of distance values ​​to one cluster. In this way, the cluster H31 shown in FIG. 7 is reclassified into each cluster of each Gaussian distribution shape shown in each of the approximation curves L311, L312, and L313. Note that, when the variance value of the clustered points is less than the threshold value and it is judged that further clustering is not required (No in step S4), the point cloud classification means 12 outputs the clustered points to the point cloud generation means 13.

[0042] Next, the point cloud generation means 13, as described above, determines whether to accept or reject the point cloud of each cluster for the clustered point cloud input from the point cloud classification means 12, and outputs the point cloud of the cluster to be accepted (step S5).

[0043] Thereafter, when the series of operations from steps S2 to S5 is completed, the point cloud processing device 1 checks whether any other areas remain that require point cloud processing, and if so (Yes in step S5'), it again specifies a different area and repeats the same processing described above to perform point cloud processing on each part of the input point cloud.

[0044] <Embodiment 3> Next, a third embodiment of the present disclosure will be described with reference to Fig. 9 to Fig. 10. Fig. 9 to Fig. 10 are block diagrams showing the configuration of an information processing device in the third embodiment. Note that this embodiment shows an outline of the configuration of the point cloud processing device described in the above-mentioned embodiment.

[0045] First, a hardware configuration of the information processing device 100 in this embodiment will be described with reference to Fig. 9. The information processing device 100 is configured as a general information processing device, and is equipped with the following hardware configuration, as an example. ·CPU(Central Processing Unit)101(Arithmetic unit) ROM (Read Only Memory) 102 (storage device) RAM (Random Access Memory) 103 (storage device) · Programs 104 loaded into RAM 103 A storage device 105 for storing a program group 104 A drive device 106 that reads and writes data from a storage medium 110 external to the information processing device A communication interface 107 for connecting to a communication network 111 outside the information processing device Input / output interface 108 for inputting and outputting data Bus 109 connecting each component

[0046] 9 shows an example of the hardware configuration of the information processing device which is the information processing device 100, and the hardware configuration of the information processing device is not limited to the above-mentioned case. For example, the information processing device may be configured with a part of the above-mentioned configuration, such as not having the drive device 106. Moreover, the information processing device may use a GPU (Graphic Processing Unit), a DSP (Digital Signal Processor), an MPU (Micro Processing Unit), an FPU (Floating point number Processing Unit), a PPU (Physics Processing Unit), a TPU (Tensor Processing Unit), a quantum processor, a microcontroller, or a combination thereof, instead of the above-mentioned CPU.

[0047] The information processing device 100 can be equipped with the classification unit 121 and the determination unit 122 shown in Fig. 10 by the CPU 101 acquiring the program group 104 and executing it. The program group 104 is stored in the storage device 105 or the ROM 102 in advance, for example, and the CPU 101 loads the program group 104 into the RAM 103 and executes it as necessary. The program group 104 may be supplied to the CPU 101 via the communication network 111, or may be stored in the storage medium 110 in advance, and the drive device 106 may read out the program and supply it to the CPU 101. However, the above-mentioned classification unit 121 and the determination unit 122 may be constructed of dedicated electronic circuits for realizing such means.

[0048] The classification unit 121 classifies three-dimensional point cloud data including distance values ​​to objects in a predetermined region into one or more clusters based on the distance values. The determination unit 122 determines the cluster to be adopted based on a distribution of the distance values ​​of the three-dimensional point cloud data.

[0049] By being configured as described above, the present disclosure determines which cluster to adopt based on the distribution of distance values ​​of the three-dimensional point cloud data, making it possible to accurately remove point cloud data that is determined to be a noise point cloud even when the point clouds are densely packed.

[0050] The above-mentioned program can be stored and supplied to a computer using various types of non-transitory computer readable media. The non-transitory computer readable medium includes various types of tangible storage media. Examples of the non-transitory computer readable medium include magnetic recording media (e.g., flexible disks, magnetic tapes, hard disk drives), magneto-optical recording media (e.g., magneto-optical disks), CD-ROM (Read Only Memory), CD-R, CD-R / W, and semiconductor memory (e.g., mask ROM, PROM (Programmable ROM), EPROM (Erasable PROM), flash ROM, RAM (Random Access Memory)). The program may also be supplied to a computer by various types of transitory computer readable media. Examples of the transitory computer readable medium include electric signals, optical signals, and electromagnetic waves. The transitory computer readable medium can supply the program to a computer via a wired communication path such as an electric wire and an optical fiber, or via a wireless communication path.

[0051] Although the present disclosure has been described above with reference to the above-mentioned embodiments, the present disclosure is not limited to the above-mentioned embodiments. Various modifications that can be understood by a person skilled in the art can be made to the configuration and details of the present disclosure within the scope of the present disclosure. In addition, at least one or more functions of the above-mentioned classification unit 121 and determination unit 122 may be executed by an information processing device installed and connected anywhere on a network, that is, they may be executed by so-called cloud computing.

[0052] <Additional Notes> A part or all of the above-described embodiments may be described as follows: Below, an overview of the configurations of an information processing device, an information processing method, and a program according to the present disclosure will be described. However, the present disclosure is not limited to the following configurations. (Appendix 1) A classification unit that classifies three-dimensional point cloud data including distance values ​​to objects in a predetermined region into one or more clusters based on the distance values; a determination unit that determines the cluster to be adopted based on a distribution of the distance values ​​of the three-dimensional point cloud data included in the classified cluster; An information processing device comprising: (Appendix 2) 2. The information processing device according to claim 1, The determination unit determines the cluster to be adopted based on a histogram of the distance values ​​of the three-dimensional point cloud data included in the classified cluster. Information processing device. (Appendix 3) The information processing device according to claim 2, The determination unit determines the cluster to be adopted based on a shape of the histogram. Information processing device. (Appendix 4) The information processing device according to claim 3, The determination unit determines the cluster having the histogram having a Gaussian distribution shape as the cluster to be adopted. Information processing device. (Appendix 5) An information processing device according to any one of claims 1 to 4, The determination unit determines the cluster to be adopted based on a frequency of the distance values ​​of the three-dimensional point cloud data included in the classified cluster. Information processing device. (Appendix 6) The information processing device according to claim 5, The determination unit determines, as the cluster to be adopted, a cluster in which a frequency of the distance values ​​of the three-dimensional point cloud data included in the classified cluster is equal to or greater than a preset threshold value. Information processing device. (Appendix 7) 7. The information processing device according to claim 1, the determination unit calculates a representative value of the distance values ​​based on the distance values ​​of the three-dimensional point cloud data included in the cluster to be adopted; Information processing device. (Appendix 8) An information processing device according to any one of claims 1 to 7, the classification unit reclassifies the three-dimensional point cloud data included in the classified cluster into a plurality of clusters based on a distribution of the distance values ​​of the three-dimensional point cloud data included in the classified cluster; the determination unit determines the cluster to be adopted based on a distribution of the distance values ​​of the three-dimensional point cloud data included in the reclassified cluster. Information processing device. (Appendix 9) classifying three-dimensional point cloud data including distance values ​​to objects in a predetermined region into one or more clusters based on the distance values; determining the cluster to be adopted based on a distribution of the distance values ​​of the three-dimensional point cloud data included in the classified cluster; Information processing methods. (Appendix 10) classifying three-dimensional point cloud data including distance values ​​to objects in a predetermined region into one or more clusters based on the distance values; determining the cluster to be adopted based on a distribution of the distance values ​​of the three-dimensional point cloud data included in the classified cluster; A program that causes a computer to execute a process. [Explanation of symbols]

[0053] 1 Point cloud processing device 11 Point cloud extraction method 12 Point cloud classification means 13 Point cloud generation means 100 Information processing device 101 CPU 102 ROM 103 RAM 104 Programs 105 Storage device 106 Drive device 107 Communication Interface 108 Input / Output Interface 109 Bus 110 Storage medium 111 Communication Network 121 Classification Department 122 Decision Section

Claims

1. A classification unit that classifies three-dimensional point cloud data including distance values ​​to objects in a predetermined region into one or more clusters based on the distance values; a determination unit that determines the cluster to be adopted based on a distribution of the distance values ​​of the three-dimensional point cloud data included in the classified cluster; An information processing device comprising:

2. 2. The information processing device according to claim 1, The determination unit determines the cluster to be adopted based on a histogram of the distance values ​​of the three-dimensional point cloud data included in the classified cluster. Information processing device.

3. 3. The information processing device according to claim 2, The determination unit determines the cluster to be adopted based on a shape of the histogram. Information processing device.

4. 4. The information processing device according to claim 3, The determination unit determines the cluster having the histogram having a Gaussian distribution shape as the cluster to be adopted. Information processing device.

5. 2. The information processing device according to claim 1, The determination unit determines the cluster to be adopted based on a frequency of the distance values ​​of the three-dimensional point cloud data included in the classified cluster. Information processing device.

6. 6. The information processing device according to claim 5, The determination unit determines, as the cluster to be adopted, a cluster in which a frequency of the distance values ​​of the three-dimensional point cloud data included in the classified cluster is equal to or greater than a preset threshold value. Information processing device.

7. 2. The information processing device according to claim 1, the determination unit calculates a representative value of the distance values ​​based on the distance values ​​of the three-dimensional point cloud data included in the cluster to be adopted; Information processing device.

8. 2. The information processing device according to claim 1, the classification unit reclassifies the three-dimensional point cloud data included in the classified cluster into a plurality of clusters based on a distribution of the distance values ​​of the three-dimensional point cloud data included in the classified cluster; the determination unit determines the cluster to be adopted based on a distribution of the distance values ​​of the three-dimensional point cloud data included in the reclassified cluster. Information processing device.

9. classifying three-dimensional point cloud data including distance values ​​to objects in a predetermined region into one or more clusters based on the distance values; determining the cluster to be adopted based on a distribution of the distance values ​​of the three-dimensional point cloud data included in the classified cluster; Information processing methods.

10. classifying three-dimensional point cloud data including distance values ​​to objects in a predetermined region into one or more clusters based on the distance values; determining the cluster to be adopted based on a distribution of the distance values ​​of the three-dimensional point cloud data included in the classified cluster; A program that causes a computer to execute a process.

Citation Information

Patent Citations

  • JP173749A

Cited By

  • Information processing device, information processing method, and program

    JP2026140243A

  • Information processing device, information processing method, and program

    JP7813925B1