Method, apparatus and device for processing point cloud data, and storage medium

Abstract

The present application relates to a kind of for processing the method, device and storage medium of point cloud data, the method comprises: according to the spacing of adjacent point in point cloud, determine the sparse point cloud interval and dense point cloud interval in point cloud;For the dense point cloud interval in point cloud, according to the spacing of adjacent point in the dense point cloud interval, the point in the dense point cloud interval is carried out sparse processing;And / or for the sparse point cloud interval in point cloud, the camera image and point cloud of the same perspective are fused, to generate the fusion chart under the same perspective, and based on the fusion chart, the sparse point cloud interval is carried out density enhancement processing.The scheme of the present application can determine sparse point cloud interval and dense point cloud interval, and can be carried out corresponding processing respectively for sparse point cloud interval and dense point cloud interval, so that the point cloud data after processing is more uniform, to facilitate the precision of target detection is improved.

Description

TECHNICAL FIELD

[0001] The present application generally relates to the technical field of data processing. More particularly, the present application relates to a method, apparatus, device and storage medium for processing point cloud data. BACKGROUND

[0002] This section is intended to provide background information to the embodiments of the present application recited in the claims. The description herein can include concepts that can be pursued, but are not necessarily ones that have been previously conceived or pursued. Therefore, unless otherwise indicated herein, the contents described in this section are not, and are not intended to be, prior art to the present application, merely because of their inclusion in this section.

[0003] The point cloud data generated by the laser radar has a divergence characteristic, so that the distance from the radar itself will have an impact on the density of the point cloud data, etc. Specifically, for the points in the point cloud that are farther away from the radar itself, the spacing between points and the density are smaller, thereby causing certain difficulty in identifying the target, and easily causing the problem of missed detection; while for the points in the point cloud that are closer to the radar itself, the spacing between points and the density are smaller, which will greatly increase the algorithm complexity when processing the point cloud data, and therefore there is an urgent need for a method capable of encryption processing or thinning processing according to the sparsity of the point cloud. SUMMARY

[0004] In view of the above-mentioned technical problems, the technical solutions of the present application provide a method, apparatus, device and storage medium for processing point cloud data in multiple aspects.

[0005] In a first aspect of the present application, a method for processing point cloud data is provided, comprising: determining sparse point cloud intervals and dense point cloud intervals in the point cloud according to the spacing of adjacent points in the point cloud; for the dense point cloud intervals in the point cloud, performing thinning processing on the points in the dense point cloud intervals according to the spacing of adjacent points in the dense point cloud intervals; and / or for the sparse point cloud intervals in the point cloud, fusing the camera image and the point cloud of the same perspective to generate a fusion image under the same perspective, and performing density enhancement processing on the sparse point cloud intervals based on the fusion image.

[0006] In one embodiment, determining the sparse point cloud interval and the dense point cloud interval in the point cloud according to the interval of adjacent points in the point cloud comprises: determining the interval of adjacent points in the point cloud according to distance information between points in the point cloud and the radar and a resolution of the radar; the resolution comprises a horizontal resolution and / or a vertical resolution, and correspondingly, when the resolution comprises the horizontal resolution, the interval of adjacent points comprises a horizontal interval; when the resolution comprises the vertical resolution, the interval of adjacent points comprises a vertical interval; determining the sparse point cloud interval and the dense point cloud interval in the point cloud according to a first ratio of a preset minimum point cloud interval to the interval.

[0007] In another embodiment, determining the sparse point cloud interval and the dense point cloud interval in the point cloud comprises: determining the dense point cloud interval in the point cloud in response to the first ratio being greater than 1; determining the sparse point cloud interval in the point cloud in response to the first ratio being less than 1.

[0008] In yet another embodiment, thinning the points in the dense point cloud interval according to the interval of adjacent points in the dense point cloud interval comprises: in response to the dense point cloud interval comprising points on one row of point cloud or multiple rows of point cloud, determining a horizontal interval of adjacent points in each row of point cloud in the dense point cloud interval according to distance information between points in each row of point cloud and the radar and a horizontal resolution of the radar; and determining a horizontal thinning interval of the dense point cloud interval according to a second ratio of a preset minimum point cloud interval to the horizontal interval, so as to perform thinning processing of at least one point for each horizontal thinning interval in the dense point cloud interval.

[0009] In one embodiment, thinning the points in the dense point cloud interval according to the interval of adjacent points in the dense point cloud interval comprises: in response to the dense point cloud interval comprising points on one column of point cloud or multiple columns of point cloud, determining a vertical interval of adjacent points in each column of point cloud in the dense point cloud interval according to distance information between points in each column of point cloud and the radar and a vertical resolution of the radar; and determining a vertical thinning interval of the dense point cloud interval according to a third ratio of a preset minimum point cloud interval to the vertical interval, so as to perform thinning processing of at least one point for each vertical thinning interval in the dense point cloud interval.

[0010] In another embodiment, determining the horizontal thinning interval of the dense point cloud interval according to the second ratio of the preset minimum point cloud interval to the horizontal interval comprises: determining the horizontal thinning interval of each row of point cloud in the dense point cloud interval according to an average value of the second ratio of each row of point cloud in the dense point cloud interval.

[0011] In yet another embodiment, determining the vertical thinning interval of the dense point cloud interval according to a third ratio of a preset minimum point cloud interval and the vertical interval includes: determining the vertical thinning interval of each column of point cloud in the dense point cloud interval according to an average value of the third ratio of each column of point cloud in the dense point cloud interval.

[0012] In one embodiment, fusing the camera image and the point cloud of the same view angle to generate a fused image of the same view angle includes: mapping the camera image into an image front view or an image top view; projecting the point cloud into a point cloud front view or a point cloud top view; and fusing the image front view and the point cloud front view to generate a fused front view, or fusing the image top view and the point cloud top view to generate a fused top view.

[0013] In another embodiment, performing density enhancement processing on the sparse point cloud interval based on the fused image includes: projecting the sparse point cloud interval into the fused image, judging whether adjacent points in the sparse point cloud interval on the fused image satisfy an interpolation condition, and in response to the adjacent points in the sparse point cloud interval satisfying the interpolation condition, interpolating between the adjacent points in the sparse point cloud interval according to pixel information and distance information of the adjacent points in the sparse point cloud interval.

[0014] In yet another embodiment, the interpolation condition includes: the adjacent points in the sparse point cloud interval are all target points; or there is a non-target point in the adjacent points in the sparse point cloud interval, and a pixel difference value of the adjacent points in the sparse point cloud interval is less than a first threshold value.

[0015] In one embodiment, before judging whether the adjacent points in the sparse point cloud interval on the fused image satisfy the interpolation condition, the method further includes: judging whether each point in the adjacent points in the sparse point cloud interval scans a target object to determine whether each point in the adjacent points in the sparse point cloud interval is a target point.

[0016] In another embodiment, judging whether each point in the adjacent points in the sparse point cloud interval scans a target object includes: for each to-be-judged point in the adjacent points in the sparse point cloud interval, judging whether the to-be-judged point scans the target object according to an angle relationship between the to-be-judged point and a point at a corresponding position of a previous row of the to-be-judged point.

[0017] In yet another embodiment, determining whether the to-be-determined point scans the target object comprises: connecting the to-be-determined point with a point at a corresponding position of a previous row of the to-be-determined point; determining that the to-be-determined point scans the target object in response to an included angle between the connection and a light ray on which the point at the corresponding position of the previous row is located being less than a second threshold; and determining that the to-be-determined point does not scan the target object in response to the included angle between the connection and the light ray on which the point at the corresponding position of the previous row is located being greater than the second threshold.

[0018] In a second aspect of the present application, a device for processing point cloud data is provided, comprising: a determination unit configured to determine a sparse point cloud interval and a dense point cloud interval in a point cloud according to a distance between adjacent points in the point cloud; a thinning unit configured to perform thinning processing on points in a dense point cloud interval in the point cloud according to a distance between adjacent points in the dense point cloud interval; and / or a density enhancement unit configured to fuse a camera image and the point cloud of a same view angle to generate a fusion image of the same view angle for a sparse point cloud interval in the point cloud, and perform density enhancement processing on the sparse point cloud interval based on the fusion image.

[0019] In a third aspect of the present application, a device for processing point cloud data is provided, comprising: at least one processor; and a memory storing program instructions which, when executed by the at least one processor, cause the device to perform the method according to any one of the first aspect of the present application.

[0020] In a fourth aspect of the present application, a computer readable storage medium storing a program for processing point cloud data is provided, which, when executed by a processor, performs the method according to any one of the first aspect of the present application.

[0021] Through the above description of the technical solutions of the present application and the plurality of embodiments thereof, those skilled in the art can understand that the method for processing point cloud data according to the embodiments of the present application can determine a sparse point cloud interval and a dense point cloud interval according to a distance between adjacent points in a point cloud, and can perform corresponding processing on the sparse point cloud interval and the dense point cloud interval respectively, so that the processed point cloud data is more uniform, thereby facilitating improvement of the accuracy and precision of target detection. The method according to the embodiments of the present application can also perform density enhancement processing on the sparse point cloud interval by generating a fusion image of the same view angle, so as to improve the precision of the enhanced point cloud, and can effectively avoid deformation problems that can occur in the transformation of point cloud data. BRIEF DESCRIPTION OF DRAWINGS

[0022] The above and other objects, features and advantages of the present exemplary embodiments will become more apparent from the following detailed description read in conjunction with the accompanying drawings, in which like reference numerals refer to like elements throughout. The accompanying drawings, which are shown by way of example, illustrate several embodiments of the present application and, where appropriate, refer to similar or corresponding elements. In the drawings:

[0023] Figure 1 A flowchart of a method for processing point cloud data according to an embodiment of the present application is schematically shown;

[0024] Figure 2 A flowchart of a method for determining sparse point cloud intervals and dense point cloud intervals according to an embodiment of the present application is schematically shown;

[0025] Figure 3 A flowchart of a method for thinning dense point cloud intervals in a horizontal direction of a point cloud according to an embodiment of the present application is schematically shown;

[0026] Figure 4 A flowchart of a method for thinning dense point cloud intervals in a vertical direction of a point cloud according to an embodiment of the present application is schematically shown;

[0027] Figure 5 A flowchart of a method for density enhancement processing for sparse point cloud intervals according to an embodiment of the present application is schematically shown;

[0028] Figure 6 A schematic diagram of a point scanning to a target object to be judged according to an embodiment of the present application is schematically shown; and

[0029] Figure 7 A schematic diagram of an apparatus for processing point cloud data according to an embodiment of the present application is schematically shown. DETAILED DESCRIPTION

[0030] The technical solutions in the embodiments of the present application will be apparently and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of, rather than all of, the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person skilled in the art without any creative work fall within the protection scope of the present application.

[0031] It should be understood that the terms "first", "second", "third", and "fourth" and the like in the claims, specification, and drawings of the present application are used to distinguish different objects, rather than to describe a particular order. The terms "include" and "contain" used in the specification and claims of the present application indicate the presence of described features, integers, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or sets thereof.

[0032] It should also be understood that the terms used herein are for the purpose of describing particular embodiments and are not intended to limit the application. As used in the specification and claims, the singular forms "a," "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "and / or," as used herein, refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0033] As used in the specification and claims, the term "if" can be interpreted as meaning "when" or "once" or "in response to a determination" or "in response to a detection" depending on the context. Similarly, the phrase "if it is determined" or "if [the recited condition or event] is detected" can be interpreted as meaning "once it is determined" or "in response to a determination" or "once [the recited condition or event] is detected" or "in response to a detection of [the recited condition or event]" depending on the context.

[0034] The present inventors have found that when sparse point cloud intervals are encrypted, the real-time performance of encryption methods based on deep learning is poor, and it is difficult to meet the application requirements of real-time perception systems. The present inventors have also found that there are differences in the perspectives of cameras and radars, and when camera images and point clouds are fused, the commonly used fusion method is to convert the radar coordinate system to the camera coordinate system through internal and external parameter calibration for fusion matching, but such a fusion method will cause a certain deformation of the radar point cloud data, thereby affecting the accuracy and precision of the enhanced point cloud data.

[0035] The present application provides a new and implementable solution to overcome the deficiencies of the prior art. In particular, the method for processing point cloud data of the present application can use a fast data processing method to realize the judgment and targeted processing of sparse point cloud intervals and dense point cloud intervals in the point cloud, thereby improving the real-time performance of the method and reducing the miss detection rate of the target object.

[0036] As will be appreciated by one skilled in the art from the following description, the present application also provides embodiments of fine processing of dense point cloud intervals and / or sparse point cloud intervals in the point cloud. In some embodiments, the method of the present application can determine the horizontal distance and / or vertical distance of adjacent points according to the horizontal resolution and / or vertical resolution of the radar, so as to determine the horizontal interval and / or vertical interval of the dense point cloud. In other embodiments, the camera image can be mapped into an image front view or an image top view, and the point cloud can be projected into a point cloud front view or a point cloud top view, so as to facilitate the fusion of the camera image and the point cloud of the same perspective.

[0037] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0038] Figure 1 A flowchart illustrating a method for processing point cloud data according to an embodiment of the present invention is shown. Figure 1 As shown, a method 100 for processing point cloud data is provided, which may include: in step 101, determining sparse point cloud intervals and dense point cloud intervals in the point cloud based on the distance between adjacent points in the point cloud. A sparse point cloud interval may include multiple points, and the distance between adjacent points among these multiple points is relatively large. A dense point cloud interval may include multiple points, and these multiple points are relatively close together. In some embodiments, in step 101, the distance between all adjacent points can be determined by traversing the entire point cloud, and the sparse point cloud intervals and dense point cloud intervals can be determined by determining whether the data of each point needs to be encrypted or thinned. In other embodiments, the distance between adjacent points can be determined by the distance information between points in the point cloud and the radar, as well as the radar resolution. In still other embodiments, the distance between adjacent points can be compared with a preset distance, and the sparse point cloud intervals and dense point cloud intervals in the point cloud can be determined based on the comparison result.

[0039] It should be noted that step 101 is not limited to simultaneously determining sparse and dense point cloud intervals. Based on the distance between adjacent points, either sparse or dense point cloud intervals can also be determined. For example, in some application scenarios, there may be point cloud intervals with large distances between adjacent points, but no point cloud intervals with dense distances. In this case, step 101 can determine only the sparse point cloud intervals. In other application scenarios, there may be point cloud intervals with dense distances between adjacent points, but no point cloud intervals with large distances between adjacent points. In this case, step 101 can determine only the dense point cloud intervals. In still other application scenarios, there may be both point cloud intervals with large and dense distances between adjacent points. In this case, step 101 can determine both sparse and dense point cloud intervals.

[0040] After the dense point cloud interval existing in the point cloud is determined in step 101, step 102 can be performed. In step 102, the points in the dense point cloud interval in the point cloud can be thinned according to the distance between adjacent points in the dense point cloud interval. The thinning process can include the operation of removing part of the points in the dense point cloud interval. In some embodiments, the number of points that need to be thinned and the position of the points that need to be extracted can be determined according to the distance between adjacent points in the dense point cloud interval and the preset distance. By thinning the dense point cloud interval, the complexity of subsequent data processing can be reduced, thereby improving the speed and efficiency of data processing.

[0041] After the sparse point cloud interval existing in the point cloud is determined in step 101, step 103 can be performed. In step 103, the camera image and the point cloud of the same view can be fused to generate a fusion image of the same view, and the sparse point cloud interval can be subjected to density enhancement processing based on the fusion image. The density enhancement processing can include the operation of filling in point data in the sparse point cloud interval. In some embodiments, the same view can include one of a front view, a top view, a bottom view, and a side view. Fusing the camera image and the point cloud of the same view can effectively avoid deformation of the point cloud data, thereby facilitating improvement of the accuracy and precision of the fusion image. In other embodiments, the fusion can be the fusion of image features in the camera image and point cloud features in the point cloud image of the same view to obtain a fusion image containing all features in the camera image and the point cloud.

[0042] For example, when fusing the camera image and the point cloud of the same view, the overlap between each point cloud target in the point cloud and each image target in the camera image can be calculated. If the overlap between a point cloud target in the point cloud and an image target in the camera image is greater than or equal to a preset overlap threshold, the point cloud target in the point cloud and the image target in the camera image can be determined as the same measured target. If the overlap between a point cloud target in the point cloud and each image target in the camera image is less than the preset overlap threshold, the point cloud target in the point cloud can be regarded as a separate measured target. Similarly, if the overlap between an image target in the camera image and each point cloud target in the point cloud is less than the preset overlap threshold, the image target in the camera image can be regarded as a separate measured target. Thus, the camera image and the point cloud of the same view can be more accurately fused.

[0043] In some embodiments, density enhancement processing of sparse point cloud regions based on the fusion map may include: projecting the sparse point cloud regions onto the fusion map, and performing density enhancement processing on the sparse point cloud regions based on image features in the fusion map. In other embodiments, after density enhancement processing in step 103, the processed fusion map can be mapped back to the three-dimensional space of the point cloud to obtain encrypted point cloud data. Density enhancement processing based on the fusion map is beneficial to improving the accuracy and precision of the enhanced point cloud data, thereby making it more conducive to the identification of target objects in point cloud detection and further reducing the false negative rate.

[0044] The above combination Figure 1 A method for processing point cloud data according to embodiments of the present invention has been described exemplarily. Those skilled in the art will understand that the above description is exemplary and not restrictive. For example, steps 102 and 103 may not be performed in parallel as shown in the illustrations, but may be performed sequentially as needed. Furthermore, steps 102 and 103 may not be performed both as shown in the illustrations, but may be performed selectively. In some application scenarios, if a sparse point cloud region is not determined in step 101, step 103 may not be necessary. In other application scenarios, if a dense point cloud region is not determined in step 101, step 102 may not be necessary.

[0045] Figure 2 This schematically illustrates a flowchart of a method for determining sparse point cloud regions and dense point cloud regions according to an embodiment of the present invention. As will be understood from the following description, method 200 is a combination of the methods described above. Figure 1 This describes a specific implementation of step 101, therefore the description of step 101 above can also be applied to the description of method 200 below.

[0046] like Figure 2 As shown, in step 210, the spacing between adjacent points in the point cloud can be determined based on the distance information between points in the point cloud and the radar, as well as the radar's resolution. Here, resolution can include horizontal resolution and / or vertical resolution. Correspondingly, when the resolution includes horizontal resolution, the spacing between adjacent points includes horizontal spacing; when the resolution includes vertical resolution, the spacing between adjacent points includes vertical spacing. The resolution can be set according to the type of radar, its accuracy, application requirements, etc. In some embodiments, resolution can include angular resolution. In other embodiments, horizontal resolution can include the angular resolution of adjacent points in the same row of the point cloud; horizontal spacing can include the spacing between adjacent points in the same row of the point cloud. In still other embodiments, vertical resolution can include the angular resolution of adjacent points in the same column of the point cloud; vertical spacing can include the spacing between adjacent points in the same column of the point cloud.

[0047] Next, in step 220, sparse and dense point cloud regions in the point cloud can be determined based on a preset first ratio of the minimum point cloud interval to the horizontal spacing. The minimum point cloud interval can include the minimum distance between adjacent points in the point cloud. The minimum point cloud interval can be set as needed or based on experience. In some embodiments, when the resolution includes horizontal resolution, the first ratio can be the ratio of the minimum point cloud interval to the horizontal spacing, and the sparse and dense point cloud regions in the horizontal direction (i.e., rows in the point cloud) can be determined based on this first ratio. In other embodiments, when the resolution includes vertical resolution, the first ratio can be the ratio of the minimum point cloud interval to the vertical spacing, and the sparse and dense point cloud regions in the vertical direction (i.e., columns in the point cloud) can be determined based on this first ratio. To facilitate understanding of the method for determining the first ratio, it is assumed that the horizontal resolution of the radar is α and the vertical resolution is β, and the following explanation will be based on the formula.

[0048] ε=γ / (tan(α)*D(i)) (Formula 1)

[0049] σ=γ / (tan(β)*D(i)) (Formula 2)

[0050] In Formula 1 and Formula 2 above, α represents the horizontal resolution, β represents the vertical resolution, D(i) represents the distance between the i-th point in the point cloud and the radar, γ represents the preset minimum point cloud interval, (tan(α)*D(i)) represents the horizontal spacing, (tan(β)*D(i)) represents the vertical spacing, ε represents the first ratio in the horizontal direction of the point cloud, and σ represents the first ratio in the vertical direction of the point cloud.

[0051] like Figure 2 As further illustrated, step 220 may include steps 221 (shown in dashed boxes) and / or 222 (shown in dashed boxes). Specifically, in step 221, a dense point cloud region in the point cloud may be determined in response to a first ratio greater than 1. A first ratio greater than 1 can be understood as the minimum point cloud interval being greater than the spacing, indicating that the spacing between adjacent points is relatively small. In some embodiments, the resolution includes the horizontal resolution, and when the first ratio is greater than 1 (e.g., ε>1 calculated in Formula 1), a dense point cloud region is determined to exist in the horizontal direction of the point cloud, requiring further thinning processing. In other embodiments, the resolution includes the vertical resolution, and when the first ratio is greater than 1 (e.g., σ>1 calculated in Formula 2), a dense point cloud region is determined to exist in the vertical direction of the point cloud, requiring further thinning processing.

[0052] Further, in step 222, a sparse point cloud interval in the point cloud can be determined in response to the first ratio being less than 1. The first ratio being less than 1 can be understood as the minimum point cloud interval being less than the interval, which means that the interval of adjacent points is relatively large. In some embodiments, the resolution comprises a horizontal resolution, and when the first ratio is less than 1 (e.g., ε < 1 calculated in Formula 1), it is determined that there is a sparse point cloud interval in the horizontal direction of the point cloud, which needs further encryption processing. In other embodiments, the resolution comprises a vertical resolution, and when the first ratio is less than 1 (e.g., σ < 1 calculated in Formula 2), it is determined that there is a sparse point cloud interval in the vertical direction of the point cloud, which needs further encryption processing.

[0053] The above description is combined with Figure 2 The method for determining the sparse point cloud interval and the dense point cloud interval in the point cloud according to the embodiments of the present application is described exemplarily, and it can be understood that the sparse point cloud interval and / or the dense point cloud interval formed by a plurality of adjacent point data can be determined according to the method 200 by traversing each point data in the point cloud. After the dense point cloud interval is determined, the points in the dense point cloud interval can be thinned, which will be described in detail below in combination with Figure 3 and Figure 4 .

[0054] Figure 3 A method flowchart for thinning the dense point cloud interval in the horizontal direction of the point cloud according to the embodiments of the present application is schematically shown. It can be understood by those skilled in the art that the method 300 can be a specific form of the step 102 described above in combination with Figure 1 the description of the step 102 can also be applicable to the description of the method 300 below.

[0055] As shown in Figure 3 step 310, in response to the dense point cloud interval comprising points on one row of point clouds or multiple rows of point clouds, the horizontal interval of adjacent points in each row of point clouds can be determined according to the distance information between each row of point clouds in the dense point cloud interval and the radar and the horizontal resolution of the radar. In some embodiments, the dense point cloud interval comprises points on one row of point clouds, which can be a plurality of continuous points or a plurality of groups of spaced continuous dense points, wherein each group of continuous dense points can constitute a dense point cloud sub-interval. In other embodiments, the dense point cloud interval can comprise points on multiple rows of point clouds, and the continuous dense points on each row of point clouds can constitute a dense point cloud sub-interval. In yet other embodiments, determining the horizontal interval of adjacent points in each row of point clouds can comprise determining the horizontal interval of adjacent points in each dense point cloud sub-interval.

[0056] Then, in step 320, a horizontal thinning interval of the dense point cloud interval can be determined according to a second ratio of the preset minimum point cloud interval and the horizontal interval, so as to perform a thinning processing of at least one point for each horizontal thinning interval in the dense point cloud interval. In some embodiments, according to the horizontal thinning interval, when performing the thinning processing for each horizontal thinning interval in the dense point cloud interval, a random extraction, an interval extraction, a sequential extraction, or the like can be adopted to realize the thinning of at least one point. In order to facilitate understanding of the determination method of the horizontal thinning interval, the following will be exemplarily described in combination with formula three.

[0057] δ(i) = γ / (tan(α)*D(i)) (Formula three)

[0058] In formula three, δ(i) represents the horizontal thinning interval of the i-th point in the horizontal direction, γ represents the preset minimum point cloud interval, α represents the horizontal resolution, D(i) represents the distance value between the i-th point in the point cloud and the radar, and (tan(α)*D(i)) represents the horizontal interval.

[0059] In one specific embodiment, if δ(i) = 3 (i.e., the horizontal thinning interval is 3) is calculated according to formula three, it means that one of the three points can be extracted when performing the thinning processing. When performing the specific thinning operation, one of the points in the middle of the three points can be extracted, or some points on the two sides of the three points can be extracted.

[0060] As further shown in Figure 3 When the horizontal thinning intervals of the multiple points in the dense point cloud interval are inconsistent, the implementation of step 321 (shown by a dashed box) can be performed in step 320. Specifically, in step 321, the horizontal thinning interval of each row of point cloud in the dense point cloud interval can be determined according to the average value of the second ratio of each row of point cloud in the dense point cloud interval. For example, in one specific embodiment, a certain row of point cloud in the dense point cloud interval includes 6 points, and the second ratios corresponding to the 6 points are 4, 3, 5, 4, 4.5, and 3.5, respectively. Therefore, the average value of these second ratios is 4, i.e., the horizontal thinning interval of the row of point cloud can be determined as 4. When performing the thinning processing on the row of point cloud, one of every four points can be extracted, and the points less than four can not be extracted. Therefore, one of the six points can be extracted.

[0061] Figure 4 A flowchart of a method for performing a thinning processing on a dense point cloud interval in a vertical direction of a point cloud according to an embodiment of the present application is schematically shown. Through the following description, those skilled in the art can understand that the method 400 can be a specific embodiment of the step 102 described in the foregoing, and therefore the description of the step 102 in the foregoing can also be applicable to the following description of the method 400. Figure 1 ​

[0062] like Figure 4 As shown, in step 410, in response to the dense point cloud interval including points on one or more columns of point clouds, the vertical spacing between adjacent points in each column of point clouds is determined based on the distance information between the points in each column of point clouds within the dense point cloud interval and the radar, as well as the radar's vertical resolution. In some embodiments, the dense point cloud interval includes points on one column of point clouds, which can be a series of consecutive points or multiple groups of consecutive dense points spaced apart, wherein each group of consecutive dense points can form a dense point cloud sub-interval. In other embodiments, the dense point cloud interval can include points on multiple columns of point clouds, and consecutive dense points on each column of point clouds can form a dense point cloud sub-interval. In still other embodiments, determining the vertical spacing between adjacent points in each column of point clouds can include: determining the vertical spacing between adjacent points in each dense point cloud sub-interval.

[0063] Next, in step 420, the vertical thinning interval of the dense point cloud region can be determined according to the preset third ratio of the minimum point cloud interval to the vertical spacing, so that at least one point can be thinned in each vertical thinning interval within the dense point cloud region. In some embodiments, when thinning each vertical thinning interval within the dense point cloud region according to the vertical thinning interval, methods such as random sampling, interval sampling, and sequential sampling can be used to achieve thinning of at least one point. To facilitate understanding of the method for determining the vertical thinning interval, an exemplary description will be provided below in conjunction with Formula 4.

[0064] θ(i)=γ / (tan(β)*D(i)) (Formula 4)

[0065] In Formula 4, θ(i) represents the vertical thinning interval of the i-th point in the vertical direction, γ represents the preset minimum point cloud interval, β represents the vertical resolution, D(i) represents the distance between the i-th point in the point cloud and the radar, and (tan(β)*D(i)) represents the vertical spacing. Thinning is performed based on the vertical thinning interval, combined with the previous text... Figure 3 The method of thinning based on horizontal thinning intervals is similar and will not be repeated here.

[0066] like Figure 4 As further illustrated, in step 420, when the vertical thinning intervals of multiple points within a dense point cloud interval are inconsistent, the implementation of step 421 (shown in the dashed box) can be performed. Specifically, in step 421, the vertical thinning interval of each column of point clouds within the dense point cloud interval can be determined based on the average of the third ratio of each column of point clouds within the dense point cloud interval. Determining the vertical thinning interval based on the average of the third ratio, in conjunction with the preceding text... Figure 3 The method for determining the horizontal thinning interval based on the average of the second ratio in step 321 is the same or similar, and will not be repeated here.

[0067] The above Figure 3 and Figure 4 Embodiments of the implementation of the sparse point cloud region are exemplarily described, and the following will be described in combination with Figure 5 and Figure 6 Embodiments of the implementation of the sparse point cloud region are exemplarily described, and the following will be described in combination with

[0068] Figure 5 The method flowchart for the sparse point cloud region is exemplarily described according to the embodiments of the present application. Through the following description, it can be understood that the method 500 is a specific implementation of the step 103 described in the foregoing, and thus the description of the step 103 in the foregoing can also be applicable to the following description of the method 500. Figure 1

[0069] As shown in Figure 5 In step 510, the camera image and the point cloud of the same view can be fused to generate a fused image of the same view for the sparse point cloud region in the point cloud. In some embodiments, step 510 can include: in step 511 (shown in the dashed box), the camera image is mapped into an image front view or an image top view. In some embodiments, the camera image can be mapped into the image front view or the image top view through a perspective transformation. In other embodiments, the perspective transformation can include: assuming that the pixel value of the camera image before transformation is (x, y), and the pixel value after transformation is (x’, y’), the transformation formula can be:

[0070]

[0071] X = m11 * x + m12 * y + m13

[0072] Y = m21 * x + m22 * y + m23

[0073] Z = m31 * x + m32 * y + m33 (Formula Six)

[0074]

[0075]

[0076] According to formula five and formula six, formula seven and formula eight can be derived to realize the mapping of the camera image. In formula five, formula six, formula seven and formula eight, (x, y) represents the pixel value of the camera image before transformation, (x’, y’) represents the pixel value after transformation, and the others are transformation parameters, which can be set as needed.

[0077] The following returns to Figure 5 ​The description continues, in step 512 (shown in dashed box), the point cloud can be projected into a front view or a top view of the point cloud. In some application scenarios, the front view or top view of the point cloud can accurately represent the characteristics of the target. To facilitate understanding of the implementation method of point cloud projection, the following explanation uses projection into a front view of the point cloud as an example. The method of projection into a top view of the point cloud is similar to that of projection into a front view of the point cloud, and will not be repeated here.

[0078] In one specific embodiment, projecting a point cloud into a front view of the point cloud can be achieved as follows: Set the radar's horizontal viewing angle to M degrees, its horizontal field of view to M1 to M2 degrees, its vertical viewing angle to N degrees, and its vertical field of view to N1 to N2 degrees. Assume the radar's horizontal resolution after density enhancement is α and its vertical resolution is β. Then, the size of the front view of the point cloud is (M / α)*(N / β). The horizontal angle of the j-th point Pij on the i-th line in the point cloud is A(ij), and its vertical angle is B(ij). The position of point Pij projected into the front view of the point cloud is shown in Formula Nine.

[0079]

[0080] Then, the process can proceed to step 513 (shown in the dashed box), where the image front view and the point cloud front view can be merged to generate a merged front view; or the image top view and the point cloud top view can be merged to generate a merged top view. For ease of understanding, the following explanation uses the generation of the merged front view as an example. The method for generating the merged top view is similar to that for generating the merged front view, and will not be repeated here.

[0081] In one specific embodiment, the extrinsic parameter calibration results of the radar and camera can be used to map the image front view onto the point cloud front view to generate a fused front view. In the generated fused front view, each pixel can include four-dimensional information (R, G, B, D), where R, G, and B are the red, green, and blue channel (RGB) color values ​​projected from the image front view onto the fused front view, and D is the distance from the point in the point cloud front view projected onto the fused front view to the radar. In some application scenarios, if a point in the fused front view is covered by the point cloud, then D is the actual acquired distance value; if there are points in the fused front view not covered by the point cloud, then the D value for the points not covered by the point cloud is 0.

[0082] After the fused map is generated in step 510, the process can proceed to step 520. In step 520, density enhancement processing can be performed on the sparse point cloud regions based on the fused map. In some embodiments, the fused map may include a front view or a top view. In other embodiments, step 510 may include steps 521, 522, 523, and 524, which will be described in detail below.

[0083] like Figure 5 As shown, in step 521 (shown by the dashed box), sparse point cloud regions can be projected onto the fused image. Projecting sparse point cloud regions onto the fused image determines their positions within the fused image. In some embodiments, projecting sparse point cloud regions onto the fused image can be achieved through a transmission transformation.

[0084] Next, in step 522 (shown in dashed box), it can be determined whether each point in the sparse point cloud interval has scanned a target object, thus determining whether each point in the sparse point cloud interval is a target point. In some application scenarios, the target object can include fixed or moving objects such as animals, plants, people, buildings, and vehicles. It is understood that points that scan a target object are target points, and points that scan the ground are non-target points. Determining whether each point has scanned a target object can be achieved by judging the distance, angle, etc., between each point and its neighboring points. In another embodiment, determining whether each point in the sparse point cloud interval has scanned a target object can include: for each point to be judged in the sparse point cloud interval, judging whether the point to be judged has scanned a target object based on the angular relationship between the point to be judged and the corresponding point in the previous row. Assuming the point to be judged is the j-th point in the i-th row of the point cloud, then the point in the corresponding position in the previous row is the j-th point in the (i-1)-th row of the point cloud.

[0085] Then, the process can proceed to step 523 (shown in the dashed box), where it can be determined whether the interpolation conditions are met between adjacent points in the sparse point cloud interval of the fused image. In some embodiments, the interpolation conditions may include: all adjacent points in the sparse point cloud interval are target points; or there are non-target points among the adjacent points in the sparse point cloud interval, and the pixel difference between adjacent points in the sparse point cloud interval is less than a first threshold. For sparse point cloud intervals in the horizontal direction, adjacent points can be adjacent points on the same row of the point cloud in the sparse point cloud interval, such as the j-th point on the i-th line and the (j+1)-th point on the i-th line in the point cloud; for sparse point cloud intervals in the vertical direction, adjacent points can be adjacent points on the same column of the point cloud, such as the j-th point on the i-th line and the j-th point on the (i-1)-th line in the point cloud. The first threshold can be set as needed. The first threshold can be a positive number.

[0086] Further, in step 524 (shown by a dashed box), in response to the interpolation condition being met between the adjacent points in the sparse point cloud section, interpolation can be performed between the adjacent points in the sparse point cloud section according to the pixel information and distance information of the adjacent points in the sparse point cloud section. In some embodiments, the adjacent points in the sparse point cloud section are all target points, and the filling of point data can be performed between the adjacent points, and the filling can be performed using linear interpolation of the distance values (i.e., the distance values of each point in the adjacent points from the radar). In other embodiments, there are non-target points among the adjacent points in the sparse point cloud section, and pixel filling can be performed between the adjacent points according to the pixel values of the adjacent points.

[0087] The above description of the method for performing density enhancement processing on the sparse point cloud section according to the embodiments of the present application is exemplary and not limiting, and it can be understood that, Figure 5 The above description of the method for performing density enhancement processing on the sparse point cloud section according to the embodiments of the present application is exemplary and not limiting, and it can be understood that, Figure 5 The above description of the method for performing density enhancement processing on the sparse point cloud section according to the embodiments of the present application is exemplary and not limiting, and it can be understood that, Figure 6 The above description of the method for performing density enhancement processing on the sparse point cloud section according to the embodiments of the present application is exemplary and not limiting, and it can be understood that,

[0088] Figure 6 The above description of the method for performing density enhancement processing on the sparse point cloud section according to the embodiments of the present application is exemplary and not limiting, and it can be understood that, Figure 6 The above description of the method for performing density enhancement processing on the sparse point cloud section according to the embodiments of the present application is exemplary and not limiting, and it can be understood that,

[0089] In a second aspect of the present application, there is provided an apparatus for processing point cloud data, which can comprise: a determining unit configured to determine sparse point cloud intervals and dense point cloud intervals in a point cloud according to the distances between adjacent points in the point cloud; a thinning unit configured to, for the dense point cloud intervals in the point cloud, perform thinning processing on the points in the dense point cloud intervals according to the distances between adjacent points in the dense point cloud intervals; and / or a density enhancement unit configured to, for the sparse point cloud intervals in the point cloud, fuse camera images and the point cloud of the same view angle to generate a fusion image of the same view angle, and perform density enhancement processing on the sparse point cloud intervals based on the fusion image.

[0090] The apparatus according to the embodiments of the present application has been described in the foregoing in detail in connection with the method shown in Figures 1-6 The apparatus according to the embodiments of the present application has been described in the foregoing in connection with the method shown in

[0091] In a third aspect of the present application, there is provided an apparatus for processing point cloud data, comprising: at least one processor; a memory storing program instructions which, when executed by the at least one processor, cause the apparatus to perform the method according to any one of the first aspect of the present application.

[0092] Figure 7 A schematic diagram of an apparatus for processing point cloud data according to an embodiment of the present application is shown schematically. The apparatus 700 can be configured to perform operations such as processing point cloud data, etc., to implement the technical solutions of the present application for processing point cloud data as described in the foregoing in connection with Figures 1-6 A schematic diagram of an apparatus for processing point cloud data according to an embodiment of the present application is shown schematically. The apparatus 700 can be configured to perform operations such as processing point cloud data, etc., to implement the technical solutions of the present application for processing point cloud data as described in the foregoing in connection with

[0093] As shown in Figure 7 The apparatus 700 can comprise a central processing unit (CPU) 701, which can be a general purpose CPU, a dedicated CPU, or other execution unit for information processing and program execution. Further, the apparatus 700 can further comprise a mass storage 702 and a read-only memory (ROM) 703, wherein the mass storage 702 can be configured to store various types of data, including, for example, various programs for processing point cloud data, and the ROM 703 can be configured to store data required for initialization of various functional modules in the apparatus for processing point cloud data of the apparatus 700, drivers for basic input / output of the system, and booting of the operating system.

[0094] Further, the apparatus 700 can further comprise other hardware or components, such as a graphics processing unit (“GPU”) 704 and a field programmable logic array (“FPGA”) 705, etc. It can be understood that although a plurality of hardware or components are shown in the apparatus 700, they are merely exemplary and not limiting, and a person skilled in the art can add or remove corresponding hardware according to actual needs.

[0095] The device 700 of the present application can also include a communication interface 706, so that it can be connected to a local area network / wireless local area network (LAN / WLAN) through the communication interface 706, and then connected to, for example, a control terminal or the Internet ("Internet") through the LAN / WLAN. Alternatively or additionally, the device 700 of the present application can also be directly connected to the Internet or a cellular network based on wireless communication technology through the communication interface 706, such as third generation ("3G"), fourth generation ("4G"), or fifth generation ("5G") based wireless communication technology. In some application scenarios, the device 700 of the present application can also access the servers of external networks and possibly databases as needed in order to obtain various known information, data and modules, etc., and can remotely store various data processed.

[0096] The above-mentioned CPU 701, mass storage 702, read-only memory ("ROM") 703, GPU 704, FPGA 705, and communication interface 706 of the device 700 of the present application can be connected to each other through a bus 707, and interact with peripheral devices through the bus 707. In one embodiment, through the bus 707, the CPU 701 can control other hardware components and their peripherals in the device 700.

[0097] In operation, the processor CPU 701 or the graphics processor GPU 704 of the device 700 of the present application can receive data through the bus 707 and call computer program instructions or codes stored in the memory 702 (such as codes related to processing point cloud data). Specifically, the CPU 701 or the GPU 704 can perform operations such as determining to process point cloud data. Then, the processor CPU 701 or the GPU 704 can perform corresponding density enhancement processing or thinning processing according to the determined sparse point cloud interval and dense point cloud interval. After the CPU 701 or the GPU 704 determines the processed point cloud data by executing the program for processing point cloud data, the processed point cloud data can be uploaded to the network, such as a remote database or an external control terminal, etc. through, for example, the communication interface 706.

[0098] It should also be understood that any module, unit, component, server, computer, terminal or device of the present application that executes instructions can include or otherwise have access to computer-readable media, such as storage media, computer storage media, or data storage devices (removable and / or non-removable) such as magnetic disks, optical disks, or tape. Computer storage media can include volatile and non-volatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data.

[0099] In a fourth aspect of the present application, a computer-readable storage medium storing a program for processing point cloud data is provided, which, when executed by a processor, performs the method according to any one of the first aspect of the present application.

[0100] The computer-readable storage medium can be any suitable magnetic storage medium or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random-Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), and the like, or any other medium that can be used to store the desired information and that can be accessed by an application, module, or both. Any such computer storage media can be part of the device or accessible or connectable thereto. Any applications or modules described in the present application can be implemented using computer-readable / executable instructions that can be stored or otherwise held by such computer-readable media.

[0101] Through the above description of the technical solutions of the present application for processing point cloud data and the multiple embodiments, those skilled in the art can understand that the method of the embodiments of the present application can effectively solve the problem of low target detection accuracy caused by the uneven density of points at different distances in the point cloud by performing targeted density enhancement processing on the sparse point cloud interval in the determined point cloud and by performing targeted thinning processing on the dense point cloud interval. The method of the embodiments of the present application can also effectively reduce the deformation problem of point cloud data by fusing the camera image and the point cloud at the same perspective and performing encryption processing based on the fused image at the same perspective, thereby improving the accuracy of the point cloud data after enhancement processing.

[0102] In some embodiments, the method of the present embodiment can determine the corresponding thinning interval by only the second ratio and / or the third ratio to perform the thinning processing, so that the fast processing manner can be beneficial to improve the real-time performance of the point cloud data processing. In other embodiments, by judging whether the adjacent points in the sparse point cloud interval satisfy the interpolation condition, and performing the interpolation processing according to the pixel information and the distance information, the processing speed and the accuracy of the data processing can be improved, so as to meet the real-time requirement of the point cloud data processing, and the method of the present embodiment can be applied to the real-time perception system.

[0103] Although the above describes the embodiments of the present application, the above is only the embodiment adopted for facilitating the understanding of the present application, and is not used to limit the scope and application scenarios of the present application. Any person skilled in the art of the present application can make any modification and change in the implementation form and details without departing from the spirit and scope of the present application, but the patent protection scope of the present application shall be subject to the scope defined by the appended claims.

Claims

1. A method for processing point cloud data, characterized in that, include: Based on the distance between adjacent points in the point cloud, determine the sparse point cloud region and the dense point cloud region in the point cloud; For dense point cloud regions in the horizontal and / or vertical directions, the points in the dense point cloud regions are thinned out based on the distance between adjacent points in the dense point cloud regions. and / or For sparse point cloud regions in the point cloud, camera images and point clouds from the same viewpoint are fused to generate a fused map from the same viewpoint, and density enhancement processing is performed on the sparse point cloud regions based on the fused map. Based on the distance between adjacent points within the dense point cloud region, the point thinning process within the dense point cloud region includes: In response to the fact that the dense point cloud interval includes points on one row of point cloud or multiple rows of point cloud, the horizontal spacing between adjacent points in each row of point cloud is determined based on the distance information between the points in each row of point cloud and the radar and the horizontal resolution of the radar. as well as Based on the second ratio of the preset minimum point cloud interval to the horizontal spacing, the horizontal thinning interval of the dense point cloud region is determined so that at least one point thinning process is performed on each horizontal thinning interval within the dense point cloud region.

2. The method according to claim 1, characterized in that, Based on the distance between adjacent points in the point cloud, the sparse point cloud regions and dense point cloud regions in the point cloud are determined as follows: Based on the distance information between points in the point cloud and the radar, and the radar's resolution, the spacing between adjacent points in the point cloud is determined; the resolution includes horizontal resolution and / or vertical resolution, and correspondingly, when the resolution includes the horizontal resolution, the spacing between adjacent points includes the horizontal spacing; when the resolution includes the vertical resolution, the spacing between adjacent points includes the vertical spacing. Based on a first ratio of the preset minimum point cloud interval to the distance, the sparse point cloud interval and the dense point cloud interval in the point cloud are determined.

3. The method according to claim 2, characterized in that, Determining the sparse point cloud regions and dense point cloud regions in the point cloud includes: In response to the first ratio being greater than 1, the dense point cloud interval in the point cloud is determined; In response to the first ratio being less than 1, the sparse point cloud interval in the point cloud is determined.

4. The method according to any one of claims 1-3, characterized in that, Based on the distance between adjacent points within the dense point cloud region, the point thinning process within the dense point cloud region includes: In response to the dense point cloud interval comprising points on one or more columns of point clouds, the vertical spacing between adjacent points in each column of point clouds is determined based on the distance information between points in each column of point clouds within the dense point cloud interval and the radar, as well as the radar's vertical resolution; and Based on the third ratio of the preset minimum point cloud interval to the vertical spacing, the vertical thinning interval of the dense point cloud interval is determined so that at least one point is thinned for each vertical thinning interval in the dense point cloud interval.

5. The method according to claim 1, characterized in that, The horizontal thinning interval of the dense point cloud region is determined based on a second ratio of the preset minimum point cloud interval to the horizontal spacing, including: The horizontal thinning interval of each row of points in the dense point cloud interval is determined based on the average of the second ratio of each row of points in the dense point cloud interval.

6. The method according to claim 4, characterized in that, The vertical thinning interval of the dense point cloud region is determined based on a third ratio of the preset minimum point cloud interval to the vertical spacing, including: The vertical thinning interval of each column of point cloud within the dense point cloud interval is determined based on the average of the third ratio of each column of point cloud within the dense point cloud interval.

7. The method according to claim 1, characterized in that, Fusing camera images and point clouds from the same viewpoint to generate a fused map from the same viewpoint includes: Map the camera image to a front view or a top view of the image; Projecting the point cloud into a front view or a top view of the point cloud; and The image front view and the point cloud front view are merged to generate a merged front view; or The top view of the image and the top view of the point cloud are merged to generate a merged top view.

8. The method according to claim 1 or 7, characterized in that, Density enhancement processing of the sparse point cloud region based on the fused graph includes: Projecting the sparse point cloud region onto the fused graph, and determining whether adjacent points within the sparse point cloud region of the fused graph satisfy interpolation conditions; and In response to the interpolation condition being met between adjacent points in the sparse point cloud interval, interpolation is performed between adjacent points in the sparse point cloud interval based on the pixel information and distance information of adjacent points in the sparse point cloud interval.

9. The method according to claim 8, characterized in that, The interpolation conditions include: In the sparse point cloud region, adjacent points are all target points; or There are non-target points among the adjacent points in the sparse point cloud interval, and the pixel difference between adjacent points in the sparse point cloud interval is less than a first threshold.

10. The method according to claim 8 or 9, characterized in that, Before determining whether the interpolation conditions are met between adjacent points in the sparse point cloud interval of the fused map, the method further includes: Determine whether each of the adjacent points in the sparse point cloud interval has scanned the target object, so as to determine whether each of the adjacent points in the sparse point cloud interval is a target point.

11. The method according to claim 10, characterized in that, Determining whether each of the adjacent points in the sparse point cloud region has detected the target object includes: For each point to be judged among the adjacent points in the sparse point cloud interval, it is determined whether the point to be judged has scanned the target object based on the angular relationship between the point to be judged and the corresponding point in the previous row.

12. The method according to claim 11, characterized in that, Determining whether the point to be determined has scanned the target object includes: Connect the point to be judged with the corresponding point in the previous row; In response to the fact that the angle between the line and the ray of the corresponding point in the previous row is less than a second threshold, it is determined that the point to be judged has scanned the target object; If the angle between the line and the ray of the corresponding point in the previous row is greater than the second threshold, it is determined that the target object has not been scanned at the point to be judged.

13. An apparatus for processing point cloud data, characterized in that, include: The determining unit is used to determine sparse point cloud regions and dense point cloud regions in the point cloud based on the distance between adjacent points in the point cloud; A thinning unit is used to thin out points in a dense point cloud region in the horizontal and / or vertical directions, based on the distance between adjacent points in the dense point cloud region. and / or The density enhancement unit is used to fuse camera images and point clouds from the same viewpoint for sparse point cloud regions in the point cloud to generate a fused map from the same viewpoint, and to perform density enhancement processing on the sparse point cloud regions based on the fused map. The process of thinning out points within the dense point cloud region based on the distance between adjacent points within the dense point cloud region includes: In response to the dense point cloud interval comprising points on one or more rows of point clouds, the horizontal spacing between adjacent points in each row of point clouds is determined based on the distance information between points in each row of point clouds within the dense point cloud interval and the radar, as well as the horizontal resolution of the radar; and Based on the second ratio of the preset minimum point cloud interval to the horizontal spacing, the horizontal thinning interval of the dense point cloud region is determined so that at least one point thinning process is performed on each horizontal thinning interval within the dense point cloud region.

14. An apparatus for processing point cloud data, characterized in that, include: At least one processor; A memory storing program instructions that, when executed by the at least one processor, cause the device to perform the method according to any one of claims 1-12.

15. A computer-readable storage medium storing a program for processing point cloud data, which, when run by a processor, performs the method according to any one of claims 1-12.

Images