Point cloud denoising method

Abstract

The embodiment of the invention discloses a point cloud denoising method, and the method comprises the steps: obtaining the to-be-denoised point cloud data which is the point cloud data collected in a designated region based on a preset collection track; on the basis of the acquisition track, filtering point cloud data outside the specified area from the to-be-denoised point cloud data; detecting a mirror surface range in the specified area; and in the filtered point cloud data, the point cloud data collected from the interior of the mirror surface range is deleted to obtain the denoised point cloud data, so that the filtering effect of the point cloud data collected in the specified area can be improved.

Description

Technical Field

[0001] The present invention relates to three-dimensional reconstruction technology and point cloud processing technology, and in particular to a point cloud denoising method. Background Art

[0002] With the continuous development of LiDAR detection technology, the depth information of the measured target can be quickly obtained, and then the point cloud data of the measured target can be generated. Affected by the measurement environment and the surface characteristics of the measured object, noise is inevitable. For example, the situation of laser point cloud collection in indoor living environments is very complicated. Reflective objects such as mirrors will interfere with laser collection, causing the original point cloud collected by the laser equipment to be very messy, and invalid point cloud clusters may also be collected outdoors.

[0003] In view of the above situation, in the related art, third-party software is used to perform manual editing to perform point cloud denoising, such as manually cutting out noise points. This manual editing denoising method is cumbersome to operate and it is difficult to ensure that the processed point cloud data is clean and tidy. The denoising method is inefficient and has poor processing effect. Summary of the invention

[0004] In view of the above problems in the prior art, an embodiment of the present disclosure provides a point cloud denoising method.

[0005] A first aspect of an embodiment of the present disclosure provides a point cloud denoising method, comprising:

[0006] Acquire point cloud data to be denoised, where the point cloud data to be denoised is point cloud data collected based on a preset collection trajectory in a specified area;

[0007] Based on the acquisition trajectory, the point cloud data outside the specified area is filtered out from the point cloud data to be denoised;

[0008] Detect the mirror range within the specified area;

[0009] In the point cloud data after filtering, the point cloud data collected from the inside of the mirror range is deleted to obtain the denoised point cloud data.

[0010] As a possible implementation of the first aspect, based on the acquisition trajectory, filtering out point cloud data outside a specified area from the point cloud data to be denoised includes:

[0011] Clustering the point cloud data based on the nearest neighbor point distance to obtain at least one point cloud cluster;

[0012] Obtaining a first distance between at least one point cloud cluster and a collection trajectory;

[0013] Determine the point cloud clusters whose first distance is greater than or equal to a preset first distance threshold as point cloud data outside the designated area;

[0014] From at least one point cloud cluster, point cloud data outside a specified area is filtered out.

[0015] As a possible implementation of the first aspect, based on the acquisition trajectory, filtering out point cloud data outside a specified area from the point cloud data to be denoised includes:

[0016] Project the point cloud data to be denoised onto a horizontal plane to obtain an occupancy grid map corresponding to the point cloud data to be denoised;

[0017] Obtain at least one connected domain in the occupancy grid map;

[0018] The connected domain that the acquisition trajectory has not passed through is determined as point cloud data outside the specified area;

[0019] From at least one connected domain, point cloud data outside a specified area is filtered out.

[0020] As a possible implementation of the first aspect, detecting a mirror range within a specified area includes:

[0021] Detecting a light spot whose intensity value is greater than a preset first intensity threshold from the point cloud data to be denoised;

[0022] Based on the intensity value of the point cloud in the surrounding area of ​​the light spot, the mirror range in the specified area is determined.

[0023] As a possible implementation of the first aspect, in the point cloud data after filtering, the point cloud data collected from the inside of the mirror range is deleted to obtain the denoised point cloud data, which also includes:

[0024] After deleting the point cloud data collected from the inside of the mirror range, the point cloud noise points inside the wall in the specified area are filtered to obtain the denoised point cloud data.

[0025] As a possible implementation manner of the first aspect, filtering point cloud noise points inside a wall in a specified area includes:

[0026] Performing plane segmentation processing on the deleted point cloud data, and obtaining multiple wall planes in the specified area based on the result of the plane segmentation processing;

[0027] Among multiple wall planes, two wall planes that are parallel to each other and whose distance meets a preset condition are determined as point cloud planes on two sides of the same wall;

[0028] The point cloud noise points in the middle of the point cloud planes on both sides of the same wall are removed.

[0029] As a possible implementation of the first aspect, removing point cloud noise points in the middle of point cloud planes on both sides of the same wall includes:

[0030] Project the point cloud data in the middle of the point cloud planes on both sides of the same wall onto a horizontal plane to obtain a projection image;

[0031] The point cloud data corresponding to the area where no ground exists in the projected image is eliminated.

[0032] As a possible implementation manner of the first aspect, before filtering out point cloud data outside a specified area from the point cloud data to be denoised based on the acquisition trajectory, the method further includes:

[0033] Get the shooting distance of the point from the point cloud data to be denoised;

[0034] The points in the point cloud data to be denoised whose shooting distance is greater than or equal to a preset second distance threshold are filtered out.

[0035] As a possible implementation manner of the first aspect, before filtering out point cloud data outside a specified area from the point cloud data to be denoised based on the acquisition trajectory, the method further includes:

[0036] Get the intensity value of the point from the point cloud data to be denoised;

[0037] Points whose intensity values ​​in the point cloud data to be denoised are less than or equal to a preset second intensity threshold are filtered out.

[0038] As a possible implementation manner of the first aspect, before filtering out point cloud data outside a specified area from the point cloud data to be denoised based on the acquisition trajectory, the method further includes:

[0039] Count the average distance of the k nearest neighbors of the points in the cloud data to be denoised;

[0040] Points whose k-nearest-neighbor average distance in the point cloud data to be denoised are filtered out if the average distance is greater than or equal to a preset third distance threshold; wherein the preset third distance threshold is determined based on the standard deviation of the k-nearest-neighbor average distance of all points in the point cloud data to be denoised.

[0041] A second aspect of the embodiments of the present disclosure provides a point cloud denoising device, comprising:

[0042] An acquisition unit is used to acquire point cloud data to be denoised, where the point cloud data to be denoised is point cloud data collected based on a preset collection trajectory in a specified area;

[0043] A first processing unit is used to filter out point cloud data outside a specified area from the point cloud data to be denoised based on the acquisition trajectory;

[0044] A detection unit, used for detecting the range of the mirror surface within a specified area;

[0045] The second processing unit is used to delete the point cloud data collected from the inside of the mirror range in the point cloud data after filtering, so as to obtain the denoised point cloud data.

[0046] As a possible implementation manner of the second aspect, the first processing unit is configured to:

[0047] Clustering the point cloud data based on the nearest neighbor point distance to obtain at least one point cloud cluster;

[0048] Obtaining a first distance between at least one point cloud cluster and a collection trajectory;

[0049] Determine the point cloud clusters whose first distance is greater than or equal to a preset first distance threshold as point cloud data outside the designated area;

[0050] From at least one point cloud cluster, point cloud data outside a specified area is filtered out.

[0051] As a possible implementation manner of the second aspect, the first processing unit is configured to:

[0052] Project the point cloud data to be denoised onto a horizontal plane to obtain an occupancy grid map corresponding to the point cloud data to be denoised;

[0053] Obtain at least one connected domain in the occupancy grid map;

[0054] The connected domain that the acquisition trajectory has not passed through is determined as point cloud data outside the specified area;

[0055] From at least one connected domain, point cloud data outside a specified area is filtered out.

[0056] As a possible implementation manner of the second aspect, the detection unit is used to:

[0057] Detecting a light spot whose intensity value is greater than a preset first intensity threshold from the point cloud data to be denoised;

[0058] Based on the intensity value of the point cloud in the surrounding area of ​​the light spot, the mirror range in the specified area is determined.

[0059] As a possible implementation manner of the second aspect, the apparatus further includes a third processing unit, configured to:

[0060] After deleting the point cloud data collected from the inside of the mirror range, the point cloud noise points inside the wall in the specified area are filtered to obtain the denoised point cloud data.

[0061] As a possible implementation manner of the second aspect, the third processing unit includes:

[0062] A segmentation subunit is used to perform plane segmentation processing on the point cloud data after the deletion processing, and obtain multiple wall planes in the specified area based on the result of the plane segmentation processing;

[0063] A matching subunit is used to determine, among multiple wall planes, two wall planes that are parallel to each other and whose distance meets a preset condition as point cloud planes on two sides of the same wall;

[0064] The culling subunit is used to remove point cloud noise points in the middle of the point cloud planes on both sides of the same wall.

[0065] As a possible implementation manner of the second aspect, the culling subunit is used to:

[0066] Project the point cloud data in the middle of the point cloud planes on both sides of the same wall onto a horizontal plane to obtain a projection image;

[0067] The point cloud data corresponding to the area where no ground exists in the projected image is eliminated.

[0068] As a possible implementation manner of the second aspect, the apparatus further includes a fourth processing unit, configured to:

[0069] Get the shooting distance of the point from the point cloud data to be denoised;

[0070] The points in the point cloud data to be denoised whose shooting distance is greater than or equal to a preset second distance threshold are filtered out.

[0071] As a possible implementation manner of the second aspect, the apparatus further includes a fourth processing unit, configured to:

[0072] Get the intensity value of the point from the point cloud data to be denoised;

[0073] Points whose intensity values ​​in the point cloud data to be denoised are less than or equal to a preset second intensity threshold are filtered out.

[0074] As a possible implementation manner of the second aspect, the apparatus further includes a fourth processing unit, configured to:

[0075] Count the average distance of the k nearest neighbors of the points in the cloud data to be denoised;

[0076] Points whose k-nearest-neighbor average distance in the point cloud data to be denoised are filtered out if the average distance is greater than or equal to a preset third distance threshold; wherein the preset third distance threshold is determined based on the standard deviation of the k-nearest-neighbor average distance of all points in the point cloud data to be denoised.

[0077] A third aspect of the present disclosure provides an electronic device, including:

[0078] a memory for storing a computer program product;

[0079] The processor is used to execute a computer program product stored in the memory, and when the computer program product is executed, the method of any one of the above-mentioned first aspects is implemented.

[0080] A fourth aspect of an embodiment of the present disclosure provides a computer-readable storage medium having computer program instructions stored thereon. When the computer program instructions are executed by a processor, the method of any one of the above-mentioned first aspects is implemented.

[0081] Based on the embodiments of the present disclosure, invalid point cloud data outside the specified area is filtered out based on the collection trajectory, and noise caused by reflective objects such as mirrors is deleted by detecting mirrors, so that the filtering effect of the point cloud data collected in the specified area can be improved.

[0082] The technical solution of the present disclosure is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] The accompanying drawings, which constitute a part of the specification, illustrate embodiments of the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0084] The present disclosure may be more clearly understood from the following detailed description with reference to the accompanying drawings, in which:

[0085] Figure 1 A flowchart of an embodiment of the point cloud denoising method disclosed in the present invention;

[0086] Figure 2 A flowchart of an embodiment of the point cloud denoising method disclosed in the present invention;

[0087] Figure 3 A flowchart of an embodiment of the point cloud denoising method disclosed in the present invention;

[0088] Figure 4 A schematic diagram of point cloud data projection according to an embodiment of the point cloud denoising method disclosed herein;

[0089] Figure 5 A schematic diagram of point cloud data projection according to an embodiment of the point cloud denoising method disclosed herein;

[0090] Figure 6 A schematic diagram of mirror reflection of an embodiment of the point cloud denoising method disclosed in the present invention;

[0091] Figure 7 A flowchart of an embodiment of the point cloud denoising method disclosed in the present invention;

[0092] Figure 8 This is an effect diagram of an embodiment of the point cloud denoising method disclosed in the present invention;

[0093] Fig. 9This is an effect diagram of an embodiment of the point cloud denoising method disclosed in the present invention;

[0094] Fig.10 It is a structural schematic diagram of an embodiment of the point cloud denoising device disclosed in the present invention;

[0095] Fig.11 It is a structural schematic diagram of an embodiment of the point cloud denoising device disclosed in the present invention;

[0096] Fig.12 It is a structural schematic diagram of an embodiment of the point cloud denoising device disclosed in the present invention;

[0097] Fig.13 It is a structural schematic diagram of an embodiment of the point cloud denoising device disclosed in the present invention;

[0098] Fig.14 A block diagram of an electronic device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0099] The following first introduces the existing method, and then introduces the technical solution of the present application in detail.

[0100] In the related art, point cloud denoising is performed by manual editing using third-party software. This manual editing denoising method is cumbersome to operate and it is difficult to ensure that the processed point cloud data is clean and tidy. In addition, in order to address the interference of reflective objects such as mirrors, there is also a processing method of blocking reflective objects to reduce the noise caused by mirrors, but this method requires more cumbersome preparation before collection. In summary, the existing point cloud denoising methods are inefficient and have poor processing effects.

[0101] Based on the technical problems existing in the above-mentioned prior art, an embodiment of the present application provides a point cloud denoising method, which filters out invalid point cloud data outside the specified area based on the collection trajectory, and deletes noise caused by reflective objects such as mirrors by detecting mirrors. This can improve the filtering effect of the point cloud data collected in the specified area and realize automatic denoising, thereby solving the technical problems mentioned in the prior art of low denoising efficiency, cumbersome operation and poor effect.

[0102] Figure 1 FIG. 1 is a flow chart of an embodiment of the point cloud denoising method disclosed in the present invention. Figure 1 As shown, it may specifically include:

[0103] Step S110, obtaining point cloud data to be denoised, where the point cloud data to be denoised is point cloud data collected based on a preset collection trajectory in a specified area.

[0104] In one example, the designated area may include a designated indoor area of ​​a house. Specifically, the collection trajectory may be preset. For example, the collection trajectory in the indoor area of ​​a house may be to move around the wall in each room on a trajectory line L meters away from the wall. Among them, L may be set to a distance of more than 0.5 meters. According to the layout of the house, the point cloud collection device may be controlled in sequence to move indoors according to a preset collection trajectory, and point cloud data may be collected in each room in turn. Among them, the point cloud collection device may be a mobile scanning device that emits lasers, such as a three-dimensional laser scanner, a lidar, and the like. The mobile scanning device continuously emits lasers during movement, and the collection trajectory may be considered as a line connecting the laser emission points, that is, the trajectory of movement of the light source and the emission point.

[0105] In actual application scenarios, point cloud data collected easily generates noise. Noise is an irrelevant or undesirable interference signal or error in point cloud data. Point cloud denoising is to remove noise from the collected point cloud data to be denoised by appropriate processing methods to improve data quality and accuracy. In step S110, the point cloud denoising tool, which is the execution body of the point cloud denoising method disclosed in the present invention, obtains the point cloud data to be denoised from the point cloud acquisition device, and performs denoising on the point cloud data to be denoised in subsequent steps.

[0106] Step S120 , based on the acquisition trajectory, filtering out the point cloud data outside the specified area from the point cloud data to be denoised.

[0107] During the process of indoor shooting by the point cloud acquisition device, the laser may pass through the window to capture outdoor trees, buildings, etc. Therefore, the collected point cloud data to be denoised will include point cloud clusters belonging to outdoor targets. These point cloud clusters are outside the specified area and are invalid point cloud clusters. Usually, these invalid point cloud clusters are not connected to the main body of the house. This feature can be used to filter out invalid point cloud clusters. Since the points in the invalid point cloud clusters are far away from the specified area, the denoised point cloud data to be denoised can be denoised based on the acquisition trajectory. Specifically, the point cloud clusters that are relatively close to the acquisition trajectory can be retained, and the point cloud clusters that are relatively far from the acquisition trajectory can be filtered out, thereby filtering out the invalid point cloud data outside the specified area from the point cloud data to be denoised.

[0108] Step S130, detecting the mirror range in the specified area. The mirror may include the surface area of ​​an object with excellent reflection characteristics such as a mirror and glass. When the laser is incident vertically on the mirror, a high-intensity reflection echo will be generated. The echo characteristics of the mirror reflection can be used to detect the position of the high-intensity light spot in the specified area to find all the mirror ranges in the specified area. Based on the position of the high-intensity light spot, the mirror range in the house is determined.

[0109] Step S140 , in the point cloud data after filtering, point cloud data collected from the inside of the mirror range are deleted to obtain denoised point cloud data.

[0110] In the collected point cloud data to be denoised, all points that pass through the mirror range can be considered as unreal points that should not be collected. In other words, the points collected through the mirror range do not really exist, but are just mirrored points. Therefore, after the mirror range is detected, all mirrored point clouds collected by the light passing through the inside of the range are deleted. In the point cloud data after filtering obtained after step S120, all light reflected back from the mirror is deleted in step S140, and the noise caused by reflective objects such as mirrors is deleted to obtain denoised point cloud data.

[0111] In the above embodiment, the invalid point cloud data outside the specified area is first filtered out based on the acquisition trajectory, and then the point cloud data collected from the inside of the mirror range is deleted to delete the interference noise caused by reflective objects such as the mirror, and finally the denoised point cloud data is obtained. In another embodiment, the interference noise caused by reflective objects such as the mirror can also be deleted first, that is, the point cloud data collected from the inside of the mirror range is first deleted from the point cloud data to be denoised, and then the invalid point cloud data outside the specified area is filtered out based on the acquisition trajectory, and finally the denoised point cloud data is obtained.

[0112] Based on the embodiments of the present disclosure, invalid point cloud data outside the specified area is filtered out based on the collection trajectory, and noise caused by reflective objects such as mirrors is deleted by detecting mirrors, so that the filtering effect of the point cloud data collected in the specified area can be improved.

[0113] like Figure 2 As shown, in one embodiment, Figure 1 In step S120, based on the acquisition trajectory, the point cloud data outside the specified area is filtered out from the point cloud data to be denoised, including:

[0114] Step S210, clustering the point cloud data based on the nearest neighbor point distance to obtain at least one point cloud cluster;

[0115] Step S220, obtaining a first distance between at least one point cloud cluster and the acquisition trajectory;

[0116] Step S230, determining the point cloud clusters whose first distance is greater than or equal to a preset first distance threshold as point cloud data outside the designated area;

[0117] Step S240: filtering out point cloud data outside a designated area from at least one point cloud cluster.

[0118] Among them, the nearest neighbor distance of the point cloud refers to finding the nearest neighbor point to a given point in the point cloud and calculating the Euclidean distance between them. The point cloud clustering algorithm is an algorithm that divides point cloud data into different categories. In step S210, the points in the denoised point cloud data are clustered based on the nearest neighbor distance to obtain multiple different point cloud clusters. In step S220, the first distance between each point cloud cluster in at least one point cloud cluster and the acquisition trajectory is calculated. For example, the shortest distance between the geometric center point of the point cloud cluster and all points on the acquisition trajectory can be calculated as the first distance. In step S230, only the point cloud clusters that are close to the acquisition trajectory are retained. For example, the point cloud acquisition device shoots outdoor trees, buildings, etc. through the window. These objects are far away from the acquisition trajectory and belong to point cloud data outside the specified area, and they should be deleted. Specifically, a preset first distance threshold is set in advance. For point cloud clusters with a first distance greater than or equal to the threshold, their positions are far away from the acquisition trajectory, and they are deleted as point cloud data outside the specified area in step S240.

[0119] In the above implementation, the implementation of the search algorithm using the nearest neighbor distance is relatively simple, does not require assumptions about the distribution form of the data, and is applicable to various types of data. This makes the method highly flexible when processing complex point cloud data. Clustering point cloud data based on the nearest neighbor distance can effectively detect and identify objects in the spatial environment, and then delete point cloud data outside the specified area to achieve a good denoising effect.

[0120] like Figure 3 As shown, in one embodiment, Figure 1 In step S120, based on the acquisition trajectory, the point cloud data outside the specified area is filtered out from the point cloud data to be denoised, including:

[0121] Step S310, projecting the point cloud data to be denoised onto a horizontal plane to obtain an occupancy grid map corresponding to the point cloud data to be denoised;

[0122] Step S320, obtaining at least one connected domain in the occupancy grid map;

[0123] Step S330, determining the connected domain that the acquisition trajectory has not passed through as point cloud data outside the specified area;

[0124] Step S340: filtering out point cloud data outside a designated area from at least one connected domain.

[0125] The point cloud acquisition device captures outdoor trees and buildings through the window. These objects outside the specified area are invalid point cloud data. If the point cloud data is projected onto a horizontal plane, the indoor layout of the house is connected, while the outdoor trees scanned through the window are isolated point cloud clusters that are not connected to the indoor layout. Therefore, the three-dimensional point cloud data can be projected onto a two-dimensional horizontal plane and converted into a two-dimensional connected domain, and then the invalid point cloud data outside the specified area can be removed based on the above characteristics.

[0126] Specifically, in step S310, the point cloud data to be denoised is projected onto a horizontal plane, and the spatial data is converted from a three-dimensional coordinate system to a two-dimensional coordinate system through projection technology, and an occupancy grid map is generated in this process. The occupancy grid map obtained by projection refers to an occupancy grid map, which divides the space into multiple small grids (grids), each grid represents a specific spatial area, and can mark whether these areas are occupied.

[0127] In step S320, the connected domain in the occupancy grid map is calculated. A connected domain generally refers to an image region composed of foreground pixels with the same pixel value and adjacent positions in the image. In the occupancy grid map, the connected domain analysis is achieved by finding and marking adjacent pixels with the same occupancy state. In an example of calculation, the basic algorithms for connected domain analysis include scan line and seed filling algorithms. Among them, the scan line algorithm scans the image line by line, uses a stack to store boundary points, and then expands to the entire connected area. The seed filling algorithm starts from a seed point and gradually fills the surrounding adjacent pixels until there are no more adjacent pixels to fill.

[0128] Figure 4 FIG. 1 shows a schematic diagram of projecting the noise-removing point cloud data of a house onto a horizontal plane. Figure 4 It can be seen that most of the white pixels are connected. The area composed of these connected white pixels belongs to the interior layout of the house, that is, it belongs to the designated area. Outside the designated area, there are some scattered and unconnected white pixels. These white pixels are outside the designated area and are connected domains that the acquisition trajectory has not passed through, which are invalid point cloud data. In step S330, the connected domain that the acquisition trajectory has not passed through is determined as point cloud data outside the designated area. In step S340, the point cloud data outside the designated area is filtered out from at least one connected domain. Figure 5 The following figure shows the effect of filtering out the point cloud data outside the specified area. Figure 4 and Figure 5 ,Comparing the connected domains before and after filtering, it can be seen that the point cloud data becomes clean and tidy after filtering out invalid point cloud clusters.

[0129] The analysis of connected domains is very important in the occupancy grid map, and the environmental information can be better understood and processed based on the analysis results. By identifying the connected domains, the position and shape of each component in the specified area can be more accurately determined, thereby achieving a good filtering effect on the denoised point cloud data.

[0130] In one embodiment, Figure 1 Step S130 in the embodiment of the present invention detects the mirror range in the specified area, including:

[0131] Detecting a light spot whose intensity value is greater than a preset first intensity threshold from the point cloud data to be denoised;

[0132] Based on the intensity value of the point cloud in the surrounding area of ​​the light spot, the mirror range in the specified area is determined.

[0133] The mirror range can include the surface areas of reflective objects such as mirrors and glass. For the mirror range, if the laser is incident vertically on the mirror range, the reflected light in this case is a high-intensity echo. However, if the laser is not incident vertically on the mirror range, that is, in the case of oblique incidence, the intensity of the reflected light is very low. In an example, only in the case of strict vertical incidence, such as when the angle between the incident light and the plane where the mirror is located is 89 degrees, 90 degrees or 91 degrees, a high-intensity return wave can be obtained. If it is found that there is an area with a very high-intensity light spot, and the light intensity in the surrounding area of ​​the light spot is very weak, then in this case it can be determined that the light spot and the surrounding area belong to the mirror range.

[0134] In one embodiment, the mirror frame point cloud can be used to determine the mirror range, and then the point cloud data collected from the inside of the mirror range is deleted. In another embodiment, if it is a frameless mirror attached to a wall or a cabinet, the intensity characteristics of the mirror reflection are used to find the mirror frame point cloud and determine the mirror range. Specifically, a preset first intensity threshold is pre-set, and a light spot with an intensity value greater than the preset first intensity threshold is detected from the point cloud data to be denoised. The light spot is a high-intensity echo generated by the vertical incidence of the laser on the mirror surface. Then, the mirror frame position is calculated based on the intensity value of the point cloud of the surrounding area of ​​the light spot to determine the mirror range in the specified area.

[0135] Specifically, except for the light spot in the mirror range, the intensity of the reflected light in the rest of the area is very low. In the wall area outside the mirror range, the intensity of the reflected light is normal. For example, if a mirror is stuck to the wall, the echo intensity reflected on the wall may be 0.8, the echo intensity of the light spot generated by the vertical reflection of the mirror is 1, and the echo intensity of the point reflection in the mirror area around the light spot may be 0.1 or 0.01, or even zero. However, on the wall outside the mirror range, the intensity of the reflected light is normal. That is to say, there is a light spot with a particularly high intensity in the middle of the mirror range, and the light intensity in other areas of the mirror range is particularly low. Based on the above mirror reflection characteristics, the border position of the mirror area can be clouded.

[0136] like Figure 6 As shown in the figure, the border is calculated by emitting a 360-degree line along the plane from the high-intensity spot position, passing through the area with very low light intensity to find the border point cloud, that is, to find the position of the normal intensity of the wall, and obtain a polygonal circle. The boundary of the polygonal circle is the mirror frame. The intensity of the reflected light outside the polygonal circle becomes normal, that is, a normal wall. See Figure 6 The light spot in the middle is produced by the strongest reflected light. The intensity of reflected light within the radiation range is very low. The boundary of the polygonal circle is the boundary between the mirror and the wall. You can send 360 rays around the light spot along the plane 360 ​​degrees to find the border, that is, to find the position of the normal intensity of the wall, that is, the boundary between the mirror and the wall. Connect the 360 ​​vertices found into a circle, so as to form an enclosed internal area, which is the mirror range.

[0137] In one example, after deleting the point cloud data collected from the inner part of the mirror range, a hole will be obtained on the wall. This hole can be filled by remedial processing, for example, some points can be obtained by using uniformly sampled point cloud data, and then filled on the mirror range to obtain the wall again.

[0138] In addition, the detection accuracy of the mirror surface can be enhanced by sticking auxiliary reflective stickers. The characteristic of reflective stickers is that high-intensity echoes will be obtained when shooting at any angle, which is convenient for the detection of high-intensity light spots. For mirrors, high-intensity light spots can only be generated under very strict vertical incidence. Therefore, when collecting point cloud data, there may be no vertical incidence when scanning the mirror, resulting in missed points and failure to collect the mirror surface. In this case, auxiliary reflective stickers can be affixed so that a high-intensity return wave can be obtained from any angle, which is convenient for detecting high-intensity light spots. The high-intensity light spot can be used to locate the position of the mirror. In this way, it can be ensured that reflective objects such as mirrors will be detected and will not be missed, further improving the accuracy of detection.

[0139] In the disclosed embodiment, mirror detection can accurately detect reflective objects such as mirrors and glass with excellent reflective properties, so that the mirror interference noise caused by the above-mentioned reflective objects can be more accurately filtered out, so that the filtered point cloud data is clean and tidy enough to achieve a good filtering effect.

[0140] In one embodiment, Figure 1 Step S140 in the above method includes deleting the point cloud data collected from the inside of the mirror range in the point cloud data after filtering to obtain the denoised point cloud data, and further includes:

[0141] After deleting the point cloud data collected from the inside of the mirror range, the point cloud noise points inside the wall in the specified area are filtered to obtain the denoised point cloud data.

[0142] The point cloud is a collection of point data on the appearance surface of objects in the collection environment obtained by the point cloud collection device. Theoretically, the inside of the wall cannot be scanned with a point cloud, but there may be various reasons that cause noisy point clouds in the middle of the wall. Therefore, after filtering out the invalid point cloud data outside the specified area based on the collection trajectory in step S120, and deleting the point cloud data collected from the inside of the mirror range in step S140, the point cloud noise inside the wall in the specified area can also be filtered. During the denoising process, the invalid point cloud data is removed in step S120, the mirror interference noise is deleted in step S140, and the noise inside the wall is filtered. The execution order of the above three steps can be interchanged. After filtering the noise inside the wall, the obtained point cloud data is cleaner and neater, which further improves the point cloud denoising effect.

[0143] like Figure 7 As shown, in one implementation, filtering the point cloud noise inside the wall in the specified area includes:

[0144] Step S410, performing plane segmentation processing on the deleted point cloud data, and obtaining a plurality of wall planes in a specified area based on the result of the plane segmentation processing.

[0145] First, the plane segmentation algorithm of the point cloud is used to perform plane segmentation on the deleted point cloud data, and the points in the point cloud data are assigned to clusters belonging to the same plane. Then, the normal, length, and height information of the plane obtained by the plane segmentation are used to obtain multiple wall planes in the specified area. For example, a plane whose normal is along the horizontal plane and whose height and length exceed a certain threshold can be determined as a wall plane. Based on the above method, all wall planes in the specified area are filtered.

[0146] An exemplary point cloud plane segmentation algorithm includes the following steps:

[0147] 1) Preprocess the point cloud data: including data filtering, downsampling and other operations.

[0148] 2) Plane detection: Use appropriate algorithms, such as RANSAC (Random Sample Consensus) or Region Growing, to detect planes in point cloud data. These algorithms find specific plane models by fitting or clustering samples or point sets in the data.

[0149] 3) Plane parameter estimation: After finding the plane, estimate its parameters. Usually, the least square method or the method of minimizing the sum of squared residuals is used to fit the representation and description of the plane for further use and analysis.

[0150] 4) Plane segmentation: Based on the plane model and corresponding parameters, the point cloud data is segmented according to the plane, and the points are assigned to clusters belonging to the same plane. This can be achieved by calculating the distance between the point and the plane or the angle between the normal vector.

[0151] Step S420: Among the multiple wall planes, two wall planes that are parallel to each other and whose distance satisfies a preset condition are determined as point cloud planes on two sides of the same wall.

[0152] Specifically, multiple wall planes are matched with each other to find point cloud planes belonging to two sides of the same wall to form a pair. For example, two planes parallel to each other and with a distance approximately equal to the wall thickness are found and determined as point cloud planes on two sides of the same wall.

[0153] Step S430, removing point cloud noise points in the middle of the point cloud planes on both sides of the same wall.

[0154] In this implementation, the walls in the specified area can be accurately detected through plane segmentation and wall pairing, and then unrealistic point cloud noise in the wall can be removed, making the point cloud data cleaner and neater.

[0155] In one implementation, removing point cloud noise points in the middle of point cloud planes on both sides of the same wall includes:

[0156] Project the point cloud data in the middle of the point cloud planes on both sides of the same wall onto a horizontal plane to obtain a projection image;

[0157] The point cloud data corresponding to the area where no ground exists in the projected image is eliminated.

[0158] When removing point cloud noise in the middle of the wall, the existence of the door opening needs to be considered. The point cloud of the upper edge, edge and lower ground of the door opening is expected to be retained. The point cloud data in the middle of the point cloud plane on both sides of the same wall can be projected onto a horizontal plane to obtain a projected image. Figure 5 The effect diagram of the point cloud data projected on the horizontal plane is shown. There must be ground after the projection of the door opening, but there is no ground when the wall is projected. For a wall without a door, there is no ground area when this wall is projected on the horizontal plane. After the two wall planes are paired, if there are some points between the two walls, these points may be door openings, ground, door frames, and door beams. These points will not be eliminated and need to be retained.

[0159] Based on the above characteristics of point cloud data, it can be determined whether there is ground in the projected area of ​​the two paired wall planes. If there is ground, there is a door opening and it needs to be retained; the point cloud data of the area without ground and not a door opening is considered to be some noise points in the middle of the wall. Specifically, a wall is projected into a long rectangle. If there is no door opening in the projected area, the points in the entire rectangular area are removed as noise points. If there is a door opening, a piece of ground points will be connected in the rectangular area. Only the area without ground is eliminated, so that the door structure can be retained.

[0160] In the embodiment of the present application, by analyzing the projection image of the point cloud data, not only the noise points between the walls can be removed, but also the door structure on the wall can be retained, so that the denoising point cloud data is more realistic and cleaner, achieving a more accurate denoising effect.

[0161] In one embodiment, based on the acquisition trajectory, before filtering out the point cloud data outside the specified area from the point cloud data to be denoised, the method further includes:

[0162] Get the shooting distance of the point from the point cloud data to be denoised;

[0163] The points in the point cloud data to be denoised whose shooting distance is greater than or equal to a preset second distance threshold are filtered out.

[0164] All points in the point cloud data can carry emission point information, and the shooting distance of the point can be obtained based on the emission point information. When one or more laser beams are irradiated onto the surface of an object, the reflected laser will carry information such as direction and distance. The shooting distance of the point in the point cloud data can be determined based on the direction and distance of the source of the reflected light. The shooting distance can be used to perform basic filtering on the point cloud data first. A second distance threshold is set in advance for the shooting scene. Points in the distance may be isolated points, outliers or abnormal points outside the specified area. Points in the point cloud data to be denoised whose shooting distance is greater than or equal to the preset second distance threshold are filtered out. On the basis of the above basic filtering, further subsequent denoising processing can make the final denoising result better.

[0165] In one embodiment, based on the acquisition trajectory, before filtering out the point cloud data outside the specified area from the point cloud data to be denoised, the method further includes:

[0166] Get the intensity value of the point from the point cloud data to be denoised;

[0167] Points whose intensity values ​​in the point cloud data to be denoised are less than or equal to a preset second intensity threshold are filtered out.

[0168] Each point in the point cloud data can have an intensity value, and the intensity value can be used to perform basic filtering on the point cloud data. The second intensity threshold can be set in advance, that is, to determine the effective range of the point cloud intensity value. Except for strictly vertical incidence, in general, point clouds reflected by glass or mirrors often have very low intensity characteristics. In an example, the second intensity threshold can be set to 0.5. Filter out points in the point cloud data to be denoised whose intensity values ​​are less than or equal to the preset second intensity threshold. On the basis of the above basic filtering, further subsequent denoising processing can make the final denoising result better.

[0169] In one embodiment, based on the acquisition trajectory, before filtering out the point cloud data outside the specified area from the point cloud data to be denoised, the method further includes:

[0170] Count the average distance of the k nearest neighbors of the points in the cloud data to be denoised;

[0171] Points whose k-nearest-neighbor average distance in the point cloud data to be denoised are filtered out if the average distance is greater than or equal to a preset third distance threshold; wherein the preset third distance threshold is determined based on the standard deviation of the k-nearest-neighbor average distance of all points in the point cloud data to be denoised.

[0172] Among them, the k-nearest-neighbor average distance of each point in the point cloud refers to the average Euclidean distance between each point in the point cloud and its nearest k neighboring points. Based on the k-nearest-neighbor average distance, the point cloud data can be counted, and the statistical information of the point cloud can be used to perform basic filtering on the point cloud data. Specifically, the k-nearest-neighbor average distance of the points in the point cloud data to be denoised can be calculated, and the average value and standard deviation of the k-nearest-neighbor average distance of all points in the point cloud data to be denoised can be obtained. The third distance threshold can be set in advance, for example, the third distance threshold can be set to n times the above-mentioned standard deviation. Points whose k-nearest-neighbor average distance in the point cloud data to be denoised is greater than or equal to the preset third distance threshold are judged as outliers. These outliers are far away from the adjacent points, and they are eliminated based on this characteristic. On the basis of the above basic filtering, further subsequent denoising processing can make the final denoising result better.

[0173] During the denoising process, invalid point cloud data is removed in step S120, mirror interference noise is removed in step S140, and noise inside the wall is filtered. In addition, the above basic filtering is performed based on the shooting distance, intensity value, and statistical information of the point cloud in the point cloud data. The execution order of the above steps can be interchangeable. In an example, the execution order of the above steps can also be set according to the characteristics of the point cloud noise in a specific scene. For example, if the invalid point cloud data accounts for a relatively large proportion of the point cloud noise, step S120 can be executed first to remove the invalid point cloud data, and then other denoising steps can be executed.

[0174] In another example, basic filtering can be performed first, and then invalid point cloud data can be removed, mirror interference noise can be deleted, and noise inside the wall can be filtered. The specific steps are as follows:

[0175] 1) Direct filtering using basic information of the point cloud, where the basic information of the point cloud may include the emission point information and intensity information of the point in the point cloud.

[0176] 2) Filtering using statistical information of point clouds, where the statistical information of point clouds may include the average value and standard deviation of the average distance of k nearest neighbors of all points in the point cloud data.

[0177] 3) Remove invalid point cloud clusters: Filter out invalid point cloud data outside the specified area based on the acquisition trajectory.

[0178] 4) Delete mirror interference noise: Delete the point cloud data collected from the inside of the mirror range.

[0179] 5) Filter point clouds using structural information such as walls: Filter point cloud noise inside walls in the specified area.

[0180] In summary, the disclosed embodiment proposes a complete set of filtering algorithms for laser point cloud data collected in indoor scenes. On the basis of filtering based on point cloud intensity information and point cloud statistical information, the point cloud data in non-collection areas is further filtered using the collection trajectory, and a method for detecting the mirror range and filtering the mirror point cloud is proposed. Finally, the noise inside the wall is filtered to make the filtered point cloud data clean and tidy. Through the above multi-layer processing of a series of filtering means, a good filtering effect can be achieved for the point cloud data collected in indoor scenes.

[0181] Figure 8 and Fig. 9 The point cloud images before and after denoising are shown respectively based on the point cloud denoising method according to the embodiment of the present disclosure. Figure 8 and Fig. 9 ,Comparing the effect pictures before and after denoising, it can be seen that after denoising, the unreal points that should not be collected are obviously removed, making the point cloud data clean and tidy.

[0182] like Fig.10 As shown, the present disclosure also provides a corresponding embodiment of a point cloud denoising device. For the beneficial effects or technical problems solved by the device, please refer to the description of the method corresponding to each device, or refer to the description in the invention content, which will not be repeated here.

[0183] In an embodiment of the point cloud denoising device, the device comprises:

[0184] The acquisition unit 100 is used to acquire the point cloud data to be denoised, where the point cloud data to be denoised is point cloud data collected based on a preset collection trajectory in a specified area;

[0185] The first processing unit 200 is used to filter out the point cloud data outside the specified area from the point cloud data to be denoised based on the acquisition trajectory;

[0186] A detection unit 300 is used to detect the mirror range within a specified area;

[0187] The second processing unit 400 is used to delete the point cloud data collected from the inside of the mirror range in the point cloud data after filtering, so as to obtain denoised point cloud data.

[0188] In one embodiment, the first processing unit 200 is used to:

[0189] Clustering the point cloud data based on the nearest neighbor point distance to obtain at least one point cloud cluster;

[0190] Obtaining a first distance between at least one point cloud cluster and a collection trajectory;

[0191] Determine the point cloud clusters whose first distance is greater than or equal to a preset first distance threshold as point cloud data outside the designated area;

[0192] From at least one point cloud cluster, point cloud data outside a specified area is filtered out.

[0193] In one embodiment, the first processing unit 200 is used to:

[0194] Project the point cloud data to be denoised onto a horizontal plane to obtain an occupancy grid map corresponding to the point cloud data to be denoised;

[0195] Obtain at least one connected domain in the occupancy grid map;

[0196] The connected domain that the acquisition trajectory has not passed through is determined as point cloud data outside the specified area;

[0197] From at least one connected domain, point cloud data outside a specified area is filtered out.

[0198] In one embodiment, the detection unit 300 is used to:

[0199] Detecting a light spot whose intensity value is greater than a preset first intensity threshold from the point cloud data to be denoised;

[0200] Based on the intensity value of the point cloud in the surrounding area of ​​the light spot, the mirror range in the specified area is determined.

[0201] like Fig.11 As shown, in one embodiment, the device further includes a third processing unit 500, which is used to:

[0202] After deleting the point cloud data collected from the inside of the mirror range, the point cloud noise points inside the wall in the specified area are filtered to obtain the denoised point cloud data.

[0203] like Fig.12 As shown, in one embodiment, the third processing unit 500 includes:

[0204] The segmentation subunit 510 is used to perform plane segmentation processing on the point cloud data after the deletion processing, and obtain multiple wall planes in the specified area based on the result of the plane segmentation processing;

[0205] A matching subunit 520 is used to determine, among multiple wall planes, two wall planes that are parallel to each other and whose distance satisfies a preset condition as point cloud planes on two sides of the same wall;

[0206] The elimination subunit 530 is used to eliminate point cloud noise points in the middle of the point cloud planes on both sides of the same wall.

[0207] In one embodiment, the culling subunit 530 is used to:

[0208] Project the point cloud data in the middle of the point cloud planes on both sides of the same wall onto a horizontal plane to obtain a projection image;

[0209] The point cloud data corresponding to the area where no ground exists in the projected image is eliminated.

[0210] like Fig.13 As shown, in one embodiment, the device further includes a fourth processing unit 600, which is used to:

[0211] Get the shooting distance of the point from the point cloud data to be denoised;

[0212] The points in the point cloud data to be denoised whose shooting distance is greater than or equal to a preset second distance threshold are filtered out.

[0213] In one embodiment, the apparatus further includes a fourth processing unit 600, configured to:

[0214] Get the intensity value of the point from the point cloud data to be denoised;

[0215] Points whose intensity values ​​in the point cloud data to be denoised are less than or equal to a preset second intensity threshold are filtered out.

[0216] In one embodiment, the apparatus further includes a fourth processing unit 600, configured to:

[0217] Count the average distance of the k nearest neighbors of the points in the cloud data to be denoised;

[0218] Points whose k-nearest-neighbor average distance in the point cloud data to be denoised are filtered out if the average distance is greater than or equal to a preset third distance threshold; wherein the preset third distance threshold is determined based on the standard deviation of the k-nearest-neighbor average distance of all points in the point cloud data to be denoised.

[0219] The point cloud denoising device of the embodiment of the present disclosure corresponds to the above-mentioned point cloud denoising method embodiments of the present disclosure in terms of specific implementation and beneficial technical effects. The relevant contents can be referenced to each other and will not be repeated here.

[0220] Below, reference Fig.14 The electronic device according to the embodiment of the present disclosure is described. The electronic device may be any one or both of the first device and the second device, or a stand-alone device independent of them, and the stand-alone device may communicate with the first device and the second device to receive the collected input signals from them.

[0221] Fig.14 A block diagram of an electronic device according to an embodiment of the present disclosure is illustrated.

[0222] like Fig.14 As shown, the electronic device includes one or more processors and memory.

[0223] The processor may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions.

[0224] The memory may store one or more computer program products, and the memory may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program products may be stored on a computer-readable storage medium, and the processor may run the computer program product to implement the point cloud denoising method of each embodiment of the present disclosure and / or other desired functions.

[0225] In one example, the electronic device may further include: an input device and an output device, and these components are interconnected via a bus system and / or other forms of connection mechanisms (not shown).

[0226] In addition, the input device may also include, for example, a keyboard, a mouse, and the like.

[0227] The output device can output various information to the outside, including determined distance information, direction information, etc. The output device can include, for example, a display, a speaker, a printer, a communication network and a remote output device connected thereto, and the like.

[0228] Of course, to simplify, Fig.14 Only some of the components related to the present disclosure in the electronic device are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, according to specific application situations, the electronic device may further include any other appropriate components.

[0229] In addition to the above-mentioned methods and devices, an embodiment of the present disclosure may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps of the point cloud denoising method according to various embodiments of the present disclosure described in the above part of this specification.

[0230] The computer program product may be written in any combination of one or more programming languages ​​to write program code for performing the operations of the disclosed embodiments, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0231] In addition, an embodiment of the present disclosure may also be a computer-readable storage medium on which computer program instructions are stored. When the computer program instructions are executed by a processor, the processor executes the steps of the point cloud denoising method according to various embodiments of the present disclosure described in the above part of this specification.

[0232] Computer readable storage media can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can include, for example, but is not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0233] The basic principles of the present disclosure are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present disclosure are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present disclosure. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, and are not limitations. The above details do not limit the present disclosure to the necessity of adopting the above specific details to be implemented.

[0234] Each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system embodiment, since it basically corresponds to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0235] The block diagrams of the devices, apparatuses, equipment, and systems involved in this disclosure are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagrams. As will be appreciated by those skilled in the art, these devices, apparatuses, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including," "comprising," "having," and the like are open words, referring to "including but not limited to," and can be used interchangeably therewith. The words "or" and "and" used herein refer to the words "and / or," and can be used interchangeably therewith, unless the context clearly indicates otherwise. The word "such as" used herein refers to the phrase "such as but not limited to," and can be used interchangeably therewith.

[0236] The method and apparatus of the present disclosure may be implemented in many ways. For example, the method and apparatus of the present disclosure may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of steps for the method is for illustration only, and the steps of the method of the present disclosure are not limited to the order specifically described above, unless otherwise specifically stated. In addition, in some embodiments, the present disclosure may also be implemented as a program recorded in a recording medium, which includes machine-readable instructions for implementing the method according to the present disclosure. Therefore, the present disclosure also covers a recording medium storing a program for executing the method according to the present disclosure.

[0237] It should also be noted that in the apparatus, device and method of the present disclosure, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present disclosure.

[0238] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present disclosure. Therefore, the present disclosure is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

[0239] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present disclosure to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. A point cloud denoising method, characterized in that: include: Acquire point cloud data to be denoised, wherein the point cloud data to be denoised is point cloud data collected based on a preset collection trajectory in a specified area; Based on the acquisition trajectory, filtering out the point cloud data outside the designated area from the point cloud data to be denoised; Detecting the mirror range within the specified area; In the point cloud data after filtering, the point cloud data collected from the inside of the mirror range is deleted to obtain the denoised point cloud data.

2. The method according to claim 1, characterized in that The filtering out of the point cloud data outside the designated area from the point cloud data to be denoised based on the acquisition trajectory includes: Clustering the point cloud data based on the nearest neighbor point distance to obtain at least one point cloud cluster; Acquire a first distance between the at least one point cloud cluster and the acquisition trajectory; Determine the point cloud clusters whose first distance is greater than or equal to a preset first distance threshold as point cloud data outside the designated area; From the at least one point cloud cluster, point cloud data outside the designated area is filtered out.

3. The method according to claim 1, characterized in that The filtering out of the point cloud data outside the designated area from the point cloud data to be denoised based on the acquisition trajectory includes: Projecting the point cloud data to be denoised onto a horizontal plane to obtain an occupancy grid map corresponding to the point cloud data to be denoised; Acquire at least one connected domain in the occupancy grid map; Determine the connected domain that the acquisition trajectory has not passed through as point cloud data outside the designated area; From the at least one connected domain, point cloud data outside the designated area is filtered out.

4. The method according to claim 1, characterized in that: The detecting the mirror range within the specified area includes: Detecting a light spot whose intensity value is greater than a preset first intensity threshold from the to-be-denoised point cloud data; The mirror range in the designated area is determined based on the intensity value of the point cloud in the surrounding area of ​​the light spot.

5. The method according to any one of claims 1 to 4, characterized in that The point cloud data collected from the inside of the mirror range is deleted from the point cloud data after filtering to obtain denoised point cloud data, and further includes: After deleting the point cloud data collected from the inside of the mirror range, the point cloud noise points inside the wall in the designated area are filtered to obtain denoised point cloud data.

6. The method according to claim 5, characterized in that The filtering of point cloud noise points inside the wall in the designated area includes: Performing plane segmentation processing on the point cloud data after the deletion processing, and obtaining a plurality of wall planes in the designated area based on the result of the plane segmentation processing; Among the multiple wall planes, two wall planes that are parallel to each other and whose distance satisfies a preset condition are determined as point cloud planes on two sides of the same wall; The point cloud noise points in the middle of the point cloud planes on both sides of the same wall are removed.

7. The method according to claim 6, characterized in that The removing of point cloud noise points in the middle of the point cloud planes on both sides of the same wall includes: Projecting the point cloud data in the middle of the point cloud planes on both sides of the same wall onto a horizontal plane to obtain a projection image; The point cloud data corresponding to the area where no ground exists in the projection image is eliminated.

8. The method according to any one of claims 1 to 4, characterized in that Before filtering out the point cloud data outside the designated area from the point cloud data to be denoised based on the acquisition trajectory, the method further includes: Acquire the shooting distance of the point from the point cloud data to be denoised; The points in the point cloud data to be denoised whose shooting distance is greater than or equal to a preset second distance threshold are filtered out.

9. The method according to any one of claims 1 to 4, characterized in that Before filtering out the point cloud data outside the designated area from the point cloud data to be denoised based on the acquisition trajectory, the method further includes: Acquire the intensity value of the point from the point cloud data to be denoised; Points whose intensity values ​​in the point cloud data to be denoised are less than or equal to a preset second intensity threshold are filtered out.

10. The method according to any one of claims 1 to 4, characterized in that Before filtering out the point cloud data outside the designated area from the point cloud data to be denoised based on the acquisition trajectory, the method further includes: Count the average distances of the k nearest neighbors of the midpoints of the point cloud data to be denoised; Points whose k-nearest-neighbor average distance in the point cloud data to be denoised are filtered out if the distance is greater than or equal to a preset third distance threshold; wherein the preset third distance threshold is determined based on the standard deviation of the k-nearest-neighbor average distance of all points in the point cloud data to be denoised.

Images