Point cloud data enhancement method, system and device and storage medium
By adjusting the number and distribution of point cloud data, the problem that point cloud data enhancement methods in the existing technology cannot improve the accuracy rate of target recognition models is solved, and the model can efficiently identify the real targets and avoid low recall risks.
Patent Information
- Application Number
- CN202411055248.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-01
- Publication Date
- 2024-12-17
AI Technical Summary
Existing point cloud data enhancement methods cannot improve the accuracy rate of target recognition models, especially when the target position density changes, resulting in a decrease in model recall.
By adjusting the number and distribution of point cloud data according to the target's location, setting a mask and adding point cloud data for the newly added target, we will enhance the network learning data set, adapt to different spatial locations, and avoid low recall risk.
It effectively improves the error detection rate of the target recognition model for the real target and improves the recognition accuracy of the model.
Smart Images
Figure CN119151807A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method, system, device and storage medium for enhancing point cloud data. Background Art
[0002] In the field of point cloud deep learning, common ways to enhance point clouds include: Dropout, Swap, Mix, Sparse, Noise, PA-AUG, etc. However, these methods are all aimed at improving the recall rate of target detection and reducing the missed detection rate. That is, according to the point cloud data enhanced by these methods, the training of the target recognition model usually improves the recall rate of target detection and reduces the missed detection rate, but it cannot improve the precision rate of the model for target recognition, that is, the false detection rate. For example, the point cloud enhanced by the above methods does not improve the model's ability to reduce false detections. In particular, the actual position of the target is directly positively correlated with the density of the point cloud. For example, the distribution of density. For laser point cloud sensors, the point cloud on the target surface will change in direct proportion to the distance of the target from the sensor. If the position of the target is directly translated without considering the changes in the internal point cloud distribution of the target, it will cause distortion of the positive sample and reduce the recall rate of the learned network. Summary of the invention
[0003] Based on this, it is necessary to provide a point cloud data enhancement method, system, device and storage medium for the above technical problems. After the point cloud data is enhanced, the number and distribution of point clouds inside the target are adjusted according to the position of the newly added target, thereby achieving the purpose of increasing the positive samples of the network learning data set, so that the newly added target can adapt to different spatial positions, thereby avoiding the risk of low network recall. Training the target recognition model based on the enhanced point cloud data can effectively improve the false detection rate of the target recognition model for real target recognition.
[0004] In a first aspect, a method for enhancing point cloud data is provided, comprising:
[0005] According to the point cloud data scanned by the vehicle's sensor, the plane range of the ground point is obtained;
[0006] According to the point cloud data of the real target, the distance of the real target and the depth of the location where the new target is added, the point cloud situation of the new target is obtained;
[0007] According to the depth of the location where the new target is added, a mask is set within the ground point plane, and according to the point cloud situation of the new target, the point cloud data of the new target is added within the mask, wherein the point cloud data of the new target serves as a positive sample of the point cloud data representing the real target.
[0008] In some examples, obtaining the point cloud of the newly added target according to the point cloud data of the real target, the distance of the real target, and the depth of the newly added target includes:
[0009] According to the distance of the real target and the depth of the location where the new target is added, the point cloud ratio of the new target is obtained;
[0010] According to the point cloud data of the real target and the point cloud ratio of the newly added target, the quantity and distribution of the point cloud of the newly added target are obtained.
[0011] In some examples, the point cloud ratio of the newly added target is obtained by the following formula:
[0012]
[0013] Wherein, the dense_rate is the point cloud ratio of the newly added target, and the d 0 is the distance of the real target, and z is the depth of the position where the newly added target is added.
[0014] In some examples, setting a mask within the plane of the ground point according to the depth of the location where the newly added target is added, and adding point cloud data of the newly added target within the mask according to the point cloud of the newly added target, includes:
[0015] According to the depth of the newly added target adding position, an anchor position is determined within the plane range of the ground point;
[0016] The mask is set according to the anchor position, and point cloud data of the newly added target is added to the mask according to the point cloud situation of the newly added target.
[0017] In some examples, obtaining the plane range of ground points based on the point cloud data scanned by the sensor of the vehicle includes:
[0018] Obtaining point cloud data scanned by a sensor of the vehicle;
[0019] Converting the point cloud data from the sensor coordinate system to the vehicle body coordinate system;
[0020] Perform ground recognition on the point cloud data in the vehicle body coordinate system to obtain the plane range of the ground points.
[0021] In some examples, after setting a mask within the ground point plane according to the depth of the location where the newly added target is added, and adding point cloud data of the newly added target within the mask according to the point cloud of the newly added target, the method further includes:
[0022] The point cloud data is input into an initial target recognition model, and the initial target recognition model is trained according to the loss between the output of the initial target recognition model and the label of the point cloud data to obtain a trained target recognition model, wherein the label is used to characterize the real target.
[0023] In some examples, the depth of the newly added target adding position is at least one.
[0024] In a second aspect, a point cloud data enhancement system is provided, comprising:
[0025] An acquisition module is used to obtain the plane range of ground points based on the point cloud data scanned by the vehicle's sensor;
[0026] A new point cloud determination module is added, which is used to obtain the point cloud situation of the newly added target based on the point cloud data of the real target, the distance of the real target and the depth of the newly added target position;
[0027] The enhancement module is used to set a mask within the ground point plane according to the depth of the location where the new target is added, and add the point cloud data of the new target within the mask according to the point cloud situation of the new target, wherein the point cloud data of the new target is used as a positive sample of the point cloud data representing the real target.
[0028] In a third aspect, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the point cloud data enhancement method of the first aspect and any possible implementation method of the first aspect are implemented.
[0029] In a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the steps of the point cloud data enhancement method of the above-mentioned first aspect and any possible implementation method of the first aspect are implemented.
[0030] According to the embodiment of the present application, the ground point plane range is obtained based on the point cloud data scanned by the vehicle's sensor, and the point cloud situation of the newly added target is obtained based on the point cloud data of the real target, the distance of the real target and the depth of the newly added target position. Finally, according to the depth of the newly added target position, a mask is set within the ground point plane range, and according to the point cloud situation of the newly added target, the point cloud data of the newly added target is added within the mask, wherein the point cloud data of the newly added target is used as a positive sample of the point cloud data representing the real target. Therefore, after the point cloud data is enhanced, the number and distribution of point clouds inside the target are adjusted according to the position of the newly added target, thereby achieving the purpose of increasing the positive samples of the network learning data set, so that the newly added target can adapt to different spatial positions, thereby avoiding the risk of low network recall. Training the target recognition model based on the enhanced point cloud data can effectively improve the false detection rate of the target recognition model for real target recognition. BRIEF DESCRIPTION OF THE DRAWINGS
[0031] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings:
[0032] Figure 1 A flowchart of a method for enhancing point cloud data provided in an embodiment of the present application;
[0033] Figure 2 A schematic diagram of a newly added target point cloud in the point cloud data enhancement method provided in an embodiment of the present application;
[0034] Figure 3 A structural block diagram of a point cloud data enhancement system provided in an embodiment of the present application;
[0035] Figure 4 A structural block diagram of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0036] The present application is further described in detail below in conjunction with the embodiments and drawings. It is to be understood that the specific embodiments described herein are only used to explain the relevant application, rather than to limit the application. It is also necessary to explain that, for ease of description, only the parts related to the application are shown in the drawings.
[0037] It should be noted that, in the absence of conflict, the embodiments of the present application, that is, the features of the embodiments, can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.
[0038] The following describes in detail the point cloud data enhancement method, system, device and storage medium according to the embodiments of the present application in conjunction with the accompanying drawings.
[0039] Figure 1FIG. 1 is a flow chart of a method for enhancing point cloud data according to an embodiment of the present application. Figure 1 As shown, the method for enhancing point cloud data according to an embodiment of the present application includes the following steps:
[0040] S101: Obtaining a plane range of ground points based on point cloud data scanned by a sensor of the vehicle.
[0041] Taking the radar on the vehicle to perceive the surrounding environment as an example, such as the vehicle-mounted laser radar, the vehicle-mounted laser radar is a mobile scanning system that can transmit and receive laser beams, analyze the return time after the laser encounters the target object, calculate the relative distance between the target object and the vehicle, and use the three-dimensional coordinates, reflectivity and other information of a large number of dense points on the surface of the target object collected to quickly rebuild the three-dimensional model of the target and various map data, establish a three-dimensional point cloud map, and draw an environmental map to achieve the purpose of environmental perception. The object information collected by the laser radar presents a series of scattered points with accurate angle and distance information, which is called a point cloud. That is, the laser radar scans and perceives the surroundings of the vehicle, such as the front of the vehicle, and obtains the point cloud data scanned by the radar.
[0042] After the on-board laser radar scans the point cloud data around the vehicle, the plane range of the ground point (ie, the point cloud on the ground plane) can be obtained based on the point cloud data.
[0043] In a specific example, the plane range of ground points is obtained according to the point cloud data scanned by the sensor of the vehicle, including: obtaining the point cloud data scanned by the sensor of the vehicle; converting the point cloud data from the sensor coordinate system to the vehicle body coordinate system; performing ground recognition on the point cloud data in the vehicle body coordinate system to obtain the plane range of the ground points. Taking the vehicle-mounted laser radar as an example, the point cloud data scanned by the laser radar is in the laser radar sensor coordinate system. First, the point cloud data in the laser radar sensor coordinate system is converted to the vehicle body coordinate system, that is, the vehicle body coordinate system defined based on the IMU. For example: according to the external parameters of the laser radar sensor, the position and coordinate system of the laser radar sensor are determined, and then the point cloud data is converted to the vehicle body coordinate system, so that the origin and ground height in the vehicle body coordinate system can be determined. Then, after the point cloud data is converted to the vehicle body coordinate system, point cloud ground detection is performed to obtain the ground point detection result, that is, the plane range of the ground point.
[0044] In the above example, ground recognition is performed on the point cloud data in the vehicle body coordinate system to obtain the plane range of the ground points, for example, including: segmenting or clustering the point cloud data to obtain a foreground point cloud; using the statistical distribution characteristics of the foreground point cloud, ground fitting is performed on each foreground point cloud to obtain the plane range of the ground points. Specifically, point cloud ground detection refers to identifying and extracting ground information from point cloud data. Methods for point cloud ground detection include direct methods, ground fitting methods, and conditional screening methods. In an embodiment of the present application, for example, the ground fitting method RANSAC is used for ground point detection, for example: ground points are detected by fitting the normal vector of the ground. First, the foreground point cloud is separated from the background point cloud by segmenting or clustering the point cloud. Then, using the statistical distribution characteristics of the foreground point cloud, ground fitting is performed on each point to determine whether it is a ground point.
[0045] It should be noted that in order to improve the quality of point cloud data, after the point cloud data is scanned by the vehicle's sensor, it can also include: correcting the point cloud data, thereby improving the quality of the point cloud data, thereby enhancing the quality of the data samples after subsequent point cloud data enhancement.
[0046] S102: Obtaining the point cloud situation of the newly added target according to the point cloud data of the real target, the distance of the real target and the depth of the position where the newly added target is added.
[0047] In one embodiment of the present application, the point cloud situation of the newly added target is obtained based on the point cloud data of the real target, the distance of the real target and the depth of the newly added target, including: obtaining the point cloud ratio of the newly added target based on the distance of the real target and the depth of the newly added target position; obtaining the number and distribution of the point cloud of the newly added target based on the point cloud data of the real target and the point cloud ratio of the newly added target.
[0048] In this example, the point cloud ratio of the newly added target is obtained by the following formula:
[0049]
[0050] Wherein, the dense_rate is the point cloud ratio of the newly added target, and the d 0 is the distance of the real target, and z is the depth of the newly added target. The mask is set according to dense_rate.
[0051] Since the actual position of the target has a direct positive correlation with the density of the point cloud. For example, the distribution of density. For laser point cloud sensors, the point cloud on the target surface will change in direct proportion to the distance of the target from the sensor. If the position of the target is directly translated without considering the change in the distribution of the internal point cloud of the target, it will cause distortion of the positive samples and reduce the recall rate of the learned network. Therefore, by adding the position of the target to adjust the number and distribution of the point cloud inside the target, the purpose of increasing the positive samples of the network learning data set is achieved, so that the newly added targets can adapt to different spatial positions, thereby avoiding the risk of low recall of the network.
[0052] S103: According to the depth of the location where the newly added target is added, a mask is set within the plane range of the ground point, and according to the point cloud situation of the newly added target, point cloud data of the newly added target is added within the mask, wherein the point cloud data of the newly added target is used as a positive sample of the point cloud data representing the real target.
[0053] In one embodiment of the present application, Figure 2 As shown, according to the depth of the position where the new target is added, a mask is set within the ground point plane, and according to the point cloud situation of the new target, the point cloud data of the new target is added within the mask, including: according to the depth of the position where the new target is added, an anchor position is determined within the ground point plane; according to the anchor position, the mask is set, and according to the point cloud situation of the new target, the point cloud data of the new target is added within the mask.
[0054] It should be noted that if Figure 2 As shown, the depth of the newly added target position is at least one, that is, targets at multiple positions can be added, thereby enhancing the point cloud data.
[0055] After setting a mask within the ground point plane according to the depth of the location where the new target is added, and adding the point cloud data of the new target within the mask according to the point cloud situation of the new target, the method also includes: inputting the point cloud data into an initial target recognition model, and training the initial target recognition model according to the loss between the output of the initial target recognition model and the label of the point cloud data to obtain a trained target recognition model, wherein the label is used to characterize the real target.
[0056] The point cloud data representing the real target in the point cloud data sample is a positive sample. That is, in the final point cloud data sample, there will be point clouds of targets at multiple locations. Therefore, the target recognition model can be trained based on these samples to improve the recognition accuracy of the target model for the real target.
[0057] For example, in the field of point cloud deep learning, common ways to enhance point clouds include Dropout, Swap, Mix, Sparse, Noise, PA-AUG, etc. However, these methods are all aimed at improving the recall rate of target detection and reducing the missed detection rate. That is, according to the point cloud data enhanced by these methods, the training of the target recognition model usually improves the recall rate of target detection and reduces the missed detection rate, but it cannot improve the precision rate of the model for target recognition, that is, the false detection rate. For example, the point cloud enhanced by the above methods can effectively improve the precision rate for target recognition.
[0058] According to the point cloud data enhancement method of the embodiment of the present application, the ground point plane range is obtained according to the point cloud data scanned by the vehicle's sensor, and the point cloud situation of the newly added target is obtained according to the point cloud data of the real target, the distance of the real target and the depth of the newly added target position. Finally, according to the depth of the newly added target position, a mask is set within the ground point plane range, and according to the point cloud situation of the newly added target, the point cloud data of the newly added target is added within the mask, wherein the point cloud data of the newly added target is used as a positive sample of the point cloud data representing the real target. Therefore, after the point cloud data is enhanced, the number and distribution of point clouds inside the target are adjusted according to the position of the newly added target, thereby achieving the purpose of increasing the positive samples of the network learning data set, so that the newly added target can adapt to different spatial positions, thereby avoiding the risk of low recall of the network. Training the target recognition model based on the enhanced point cloud data can effectively improve the false detection rate of the target recognition model for real target recognition.
[0059] Figure 3 is a structural block diagram of a point cloud data enhancement system according to an embodiment of the present application. Figure 3 As shown, according to an embodiment of the present application, a point cloud data enhancement system includes: an acquisition module 310, a new point cloud determination module 320 and an enhancement module 330, wherein:
[0060] The acquisition module 310 is used to obtain the plane range of the ground point according to the point cloud data scanned by the sensor of the vehicle;
[0061] A newly added point cloud determination module 320 is used to obtain the point cloud situation of the newly added target according to the point cloud data of the real target, the distance of the real target and the depth of the newly added target position;
[0062] The enhancement module 330 is used to set a mask within the ground point plane according to the depth of the location where the new target is added, and to add the point cloud data of the new target within the mask according to the point cloud situation of the new target, wherein the point cloud data of the new target serves as a positive sample of the point cloud data representing the real target.
[0063] According to the point cloud data enhancement system of the embodiment of the present application, the ground point plane range is obtained according to the point cloud data scanned by the vehicle's sensor, and the point cloud situation of the newly added target is obtained according to the point cloud data of the real target, the distance of the real target and the depth of the newly added target position. Finally, according to the depth of the newly added target position, a mask is set within the ground point plane range, and according to the point cloud situation of the newly added target, the point cloud data of the newly added target is added within the mask, wherein the point cloud data of the newly added target is used as a positive sample of the point cloud data representing the real target. Therefore, after the point cloud data is enhanced, the number and distribution of point clouds inside the target are adjusted according to the position of the newly added target, thereby achieving the purpose of increasing the positive samples of the network learning data set, so that the newly added target can adapt to different spatial positions, thereby avoiding the risk of low recall of the network. Training the target recognition model based on the enhanced point cloud data can effectively improve the false detection rate of the target recognition model for real target recognition.
[0064] For the specific definition of the point cloud data enhancement system, please refer to the definition of the point cloud data enhancement method mentioned above, which will not be repeated here. The various modules of the above-mentioned point cloud data enhancement system can be implemented in whole or in part by software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.
[0065] In one embodiment, a computer device is provided. Figure 4 This is a block diagram of the computer device provided in the embodiment of the present application, refer to Figure 4 The computer device includes a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the aforementioned point cloud data enhancement method embodiment when executing the computer program. For example, the following is performed: obtaining the plane range of the ground point according to the point cloud data scanned by the sensor of the vehicle;
[0066] According to the point cloud data of the real target, the distance of the real target and the depth of the location where the new target is added, the point cloud situation of the new target is obtained;
[0067] According to the depth of the location where the new target is added, a mask is set within the ground point plane, and according to the point cloud situation of the new target, the point cloud data of the new target is added within the mask, wherein the point cloud data of the new target serves as a positive sample of the point cloud data representing the real target.
[0068] The present application also provides a computer-readable storage medium, which stores a computer program. When the processor executes the computer program, the aforementioned point cloud data enhancement method embodiment is implemented. For example, the following is performed: according to the point cloud data scanned by the vehicle's sensor, the plane range of the ground point is obtained;
[0069] According to the point cloud data of the real target, the distance of the real target and the depth of the location where the new target is added, the point cloud situation of the new target is obtained;
[0070] According to the depth of the location where the new target is added, a mask is set within the ground point plane, and according to the point cloud situation of the new target, the point cloud data of the new target is added within the mask, wherein the point cloud data of the new target serves as a positive sample of the point cloud data representing the real target.
[0071] The present application embodiment provides a computer program product, which includes instructions. When the instructions are executed, the method described in the embodiment of the present application is executed. For example, it can be executed Figure 1 The various steps of the point cloud data enhancement method shown are, for example, performed as follows: obtaining a plane range of ground points based on the point cloud data scanned by a sensor of the vehicle;
[0072] According to the point cloud data of the real target, the distance of the real target and the depth of the location where the new target is added, the point cloud situation of the new target is obtained;
[0073] According to the depth of the location where the new target is added, a mask is set within the ground point plane, and according to the point cloud situation of the new target, the point cloud data of the new target is added within the mask, wherein the point cloud data of the new target serves as a positive sample of the point cloud data representing the real target.
[0074] Those of ordinary skill in the art can understand that all or part of the processes in the methods for implementing the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).
[0075] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0076] The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the patent application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent application shall be subject to the attached claims.
Claims
1. A method for enhancing point cloud data, characterized in that: include: According to the point cloud data scanned by the vehicle's sensor, the plane range of the ground point is obtained; According to the point cloud data of the real target, the distance of the real target and the depth of the location where the new target is added, the point cloud situation of the new target is obtained; According to the depth of the location where the new target is added, a mask is set within the ground point plane, and according to the point cloud situation of the new target, the point cloud data of the new target is added within the mask, wherein the point cloud data of the new target serves as a positive sample of the point cloud data representing the real target.
2. The point cloud data enhancement method according to claim 1, characterized in that: The step of obtaining the point cloud of the newly added target according to the point cloud data of the real target, the distance of the real target and the depth of the newly added target includes: According to the distance of the real target and the depth of the location where the new target is added, the point cloud ratio of the new target is obtained; According to the point cloud data of the real target and the point cloud ratio of the newly added target, the quantity and distribution of the point cloud of the newly added target are obtained.
3. The point cloud data enhancement method according to claim 2, characterized in that: The point cloud ratio of the newly added target is obtained by the following formula: Wherein, the dense_rate is the point cloud ratio of the newly added target, the d0 is the distance of the real target, and the z is the depth of the position where the newly added target is added.
4. The method for enhancing point cloud data according to claim 1, characterized in that: The step of setting a mask within the plane range of the ground point according to the depth of the location where the newly added target is added, and adding point cloud data of the newly added target within the mask according to the point cloud situation of the newly added target, comprises: According to the depth of the newly added target adding position, an anchor position is determined within the plane range of the ground point; The mask is set according to the anchor position, and point cloud data of the newly added target is added to the mask according to the point cloud situation of the newly added target.
5. The point cloud data enhancement method according to claim 1, characterized in that: The point cloud data scanned by the vehicle sensor is used to obtain the plane range of the ground point, including: Obtaining point cloud data scanned by a sensor of the vehicle; Converting the point cloud data from the sensor coordinate system to the vehicle body coordinate system; Perform ground recognition on the point cloud data in the vehicle body coordinate system to obtain the plane range of the ground points.
6. The point cloud data enhancement method according to any one of claims 1 to 5, characterized in that: After setting a mask within the plane range of the ground point according to the depth of the location where the newly added target is added, and adding the point cloud data of the newly added target within the mask according to the point cloud situation of the newly added target, the method further includes: The point cloud data is input into an initial target recognition model, and the initial target recognition model is trained according to the loss between the output of the initial target recognition model and the label of the point cloud data to obtain a trained target recognition model, wherein the label is used to characterize the real target.
7. The point cloud data enhancement method according to claim 1, characterized in that: The depth of the newly added target adding position is at least one.
8. A point cloud data enhancement system, characterized in that: include: An acquisition module is used to obtain the plane range of ground points based on the point cloud data scanned by the vehicle's sensor; A new point cloud determination module is added, which is used to obtain the point cloud situation of the newly added target based on the point cloud data of the real target, the distance of the real target and the depth of the newly added target position; The enhancement module is used to set a mask within the ground point plane according to the depth of the location where the new target is added, and add the point cloud data of the new target within the mask according to the point cloud situation of the new target, wherein the point cloud data of the new target is used as a positive sample of the point cloud data representing the real target.
9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the point cloud data enhancement method according to any one of claims 1 to 7 is implemented.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the point cloud data enhancement method according to any one of claims 1 to 7 is implemented.