Method, device, and electronic device for removing interference points from point cloud

Abstract

The present invention provides a method, device and electronic device for removing interference points in a point cloud. The method comprises: acquiring point cloud data of a target radar frame; searching for interference points of a preset type in the point cloud data; wherein the preset type comprises at least one of a centrifugal point type, a pseudo-highest point type and an interference point group type; and removing all interference points of the preset type found from the point cloud data. The method for removing interference points in a point cloud provided by the present invention can remove interference points at the edge by searching and removing interference points of the centrifugal point type; can remove isolated suspicious high points by searching and removing interference points of the pseudo-highest point type; and can remove interference point groups caused by multipath interference by searching and removing interference points of the interference point group type. Therefore, the accuracy of target object recognition can be improved by the above three methods of searching and removing interference points.

Description

Technical Field

[0001] The present invention relates to the technical field of point cloud data processing, and in particular to a method, device and electronic device for removing interference points in a point cloud. Background Art

[0002] Millimeter-wave radar, which operates in the millimeter-wave band, is widely used in indoor safety monitoring and smart home applications due to its low cost, all-weather operation, and privacy protection. For example, the unordered point cloud generated during millimeter-wave radar signal processing can be subjected to group tracking. Data association tags can then be used to extract valid target timing information, further providing key data for related scalable functions such as size estimation, behavior prediction, and posture recognition.

[0003] During millimeter-wave transmission, they are inevitably affected by multiple scattering and other interference, resulting in interference points. These interference points can affect processing such as size estimation, behavior prediction, and posture recognition, resulting in inaccurate results. Therefore, removing interference points from point clouds has become a pressing technical challenge. Summary of the Invention

[0004] Embodiments of the present invention provide a method, device, and electronic device for removing interference points in a point cloud, so as to solve the problem that interference points exist in a point cloud and a target object cannot be accurately identified.

[0005] In a first aspect, an embodiment of the present invention provides a method for removing interference points in a point cloud, comprising:

[0006] Get the point cloud data of the target radar frame;

[0007] Searching for interference points of a preset type in the point cloud data; wherein the preset type includes at least one of a centrifugal point type, a pseudo-highest point type, and an interference point group type;

[0008] All interference points of preset types found are removed from the point cloud data.

[0009] In one possible implementation,

[0010] Find interference points of preset types in the point cloud data and remove all interference points of preset types found from the point cloud data, including:

[0011] Find interference points of the centrifugal point type in the point cloud data, and remove the interference points of the centrifugal point type found from the point cloud data;

[0012] In the point cloud data from which the interference points of the centrifugal point type are eliminated, interference points of the pseudo-highest point type are searched and the interference points of the pseudo-highest point type found are eliminated;

[0013] In the point cloud data from which interference points of the centrifugal point type and the pseudo-highest point type are eliminated, interference points of the interference point group type are searched and the found interference points of the interference point group type are eliminated.

[0014] In one possible implementation,

[0015] Finds preset types of interference points in point cloud data, including:

[0016] Determine the centroid of point cloud data;

[0017] Calculate the centrifugal distance between each point in the point cloud data and the centroid respectively;

[0018] All points in the point cloud data whose centrifugal distance from the centroid exceeds a preset distance are determined as interference points of the centrifugal point type.

[0019] In one possible implementation,

[0020] The centrifugal distance from the center of mass is:

[0021] Euclidean distance from the center of mass in three-dimensional space;

[0022] or, in three-dimensional space, the Manhattan distance from the centroid;

[0023] Alternatively, the Manhattan distance from the centroid in any dimension.

[0024] In one possible implementation, searching for interference points of a preset type in point cloud data includes:

[0025] Get the coordinate value of each point in the point cloud data in the Z dimension and sort them;

[0026] Determine whether the highest point is a pseudo-highest point: If the difference between the coordinate values ​​of the highest point and the second-highest point is greater than the preset difference, the highest point is determined to be a pseudo-highest point, otherwise the highest point is determined to be a reliable highest point; when the highest point is a pseudo-highest point, the second-highest point is set as the new highest point, and the steps of determining whether the highest point is a pseudo-highest point are repeated until a reliable highest point is found.

[0027] In one possible implementation, searching for interference points of a preset type in point cloud data includes:

[0028] In the case where the preset dimension is an abnormal dimension, the interval consisting of the maximum and minimum values ​​of the point cloud data in the preset dimension is divided into multiple sub-intervals according to preset intervals; wherein the preset dimension is the X dimension or the Y dimension;

[0029] If the number of points in all subintervals is greater than the preset point number threshold, then the points in all subintervals are determined to be interference points without interference point group type; otherwise, the subintervals with the number of points less than the preset point number threshold are determined to be broken subintervals; except for the broken subintervals, the broken subintervals divide multiple subintervals into multiple continuous intervals;

[0030] Determining a non-interference sub-interval from a plurality of continuous intervals according to a non-interference sub-interval judgment condition;

[0031] Points corresponding to all subintervals except the non-interference subinterval in the plurality of subintervals are determined as interference points of the interference point group type.

[0032] In one possible implementation, when the preset dimension is an abnormal dimension, before dividing an interval consisting of the maximum and minimum values ​​of the point cloud data in the preset dimension into a plurality of sub-intervals according to preset intervals, the method further includes:

[0033] Obtain the size of the point cloud data in a preset dimension and the coordinate Z value of the highest point in the Z dimension; if the size in the preset dimension is greater than the preset size threshold corresponding to the Z value, determine that the preset dimension is an abnormal dimension;

[0034] Alternatively, the standard deviation of all coordinates of the point cloud data in a preset dimension is obtained; if the standard deviation is greater than a preset standard deviation threshold, the preset dimension is determined to be an abnormal dimension.

[0035] In one possible implementation, the non-interference subinterval determination condition includes:

[0036] The continuous interval containing the largest number of points is the non-interference subinterval;

[0037] Alternatively, the continuous interval containing the points with the smallest average distance is the non-interference subinterval;

[0038] Alternatively, the continuous interval containing the maximum sum of the signal-to-noise ratios of the points is the non-interference subinterval.

[0039] In a second aspect, an embodiment of the present invention provides a device for removing interference points in a point cloud, comprising:

[0040] Acquisition module, used to obtain point cloud data of target radar frame;

[0041] A search module, configured to search for interference points of a preset type in the point cloud data; wherein the preset type includes at least one of a centrifugal point type, a pseudo-highest point type, and an interference point group type;

[0042] The elimination module is used to eliminate all interference points of preset types found from the point cloud data.

[0043] In a possible implementation, the search module and the elimination module are further used to:

[0044] Find interference points of the centrifugal point type in the point cloud data, and remove the interference points of the centrifugal point type found from the point cloud data;

[0045] In the point cloud data from which the interference points of the centrifugal point type are eliminated, interference points of the pseudo-highest point type are searched and the interference points of the pseudo-highest point type found are eliminated;

[0046] In the point cloud data from which interference points of the centrifugal point type and the pseudo-highest point type are eliminated, interference points of the interference point group type are searched and the found interference points of the interference point group type are eliminated.

[0047] In a possible implementation, the search module is further configured to:

[0048] Determine the centroid of point cloud data;

[0049] Calculate the centrifugal distance between each point in the point cloud data and the centroid respectively;

[0050] All points in the point cloud data whose centrifugal distance from the centroid exceeds a preset distance are determined as interference points of the centrifugal point type.

[0051] In one possible implementation, the centrifugal distance from the center of mass is:

[0052] Euclidean distance from the center of mass in three-dimensional space;

[0053] or, in three-dimensional space, the Manhattan distance from the centroid;

[0054] Alternatively, the Manhattan distance from the centroid in any dimension.

[0055] In a possible implementation, the search module is further configured to:

[0056] Get the coordinate value of each point in the point cloud data in the Z dimension and sort them;

[0057] Determine whether the highest point is a pseudo-highest point: If the difference between the coordinate values ​​of the highest point and the second-highest point is greater than the preset difference, the highest point is determined to be a pseudo-highest point, otherwise the highest point is determined to be a reliable highest point; when the highest point is a pseudo-highest point, the second-highest point is set as the new highest point, and the steps of determining whether the highest point is a pseudo-highest point are repeated until a reliable highest point is found.

[0058] In a possible implementation, the search module is further configured to:

[0059] In the case where the preset dimension is an abnormal dimension, the interval consisting of the maximum and minimum values ​​of the point cloud data in the preset dimension is divided into multiple sub-intervals according to preset intervals; wherein the preset dimension is the X dimension or the Y dimension;

[0060] If the number of points in all subintervals is greater than the preset point number threshold, then the points in all subintervals are determined to be interference points without interference point group type; otherwise, the subintervals with the number of points less than the preset point number threshold are determined to be broken subintervals; except for the broken subintervals, the broken subintervals divide multiple subintervals into multiple continuous intervals;

[0061] Determining a non-interference sub-interval from a plurality of continuous intervals according to a non-interference sub-interval judgment condition;

[0062] Points corresponding to all subintervals except the non-interference subinterval in the plurality of subintervals are determined as interference points of the interference point group type.

[0063] In a possible implementation, the search module is further configured to:

[0064] Obtain the size of the point cloud data in a preset dimension and the coordinate Z value of the highest point in the Z dimension; if the size in the preset dimension is greater than the preset size threshold corresponding to the Z value, determine that the preset dimension is an abnormal dimension;

[0065] Alternatively, the standard deviation of all coordinates of the point cloud data in a preset dimension is obtained; if the standard deviation is greater than a preset standard deviation threshold, the preset dimension is determined to be an abnormal dimension.

[0066] In one possible implementation, the non-interference subinterval determination condition includes:

[0067] The continuous interval containing the largest number of points is the non-interference subinterval;

[0068] Alternatively, the continuous interval containing the points with the smallest average distance is the non-interference subinterval;

[0069] Alternatively, the continuous interval containing the maximum sum of the signal-to-noise ratios of the points is the non-interference subinterval.

[0070] In a third aspect, an embodiment of the present invention provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the method described in the first aspect or any possible implementation of the first aspect are implemented.

[0071] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0072] Embodiments of the present invention provide a method, device, and electronic device for removing interference point clouds. The method first acquires point cloud data of a target radar frame; then, interference points of preset types in the point cloud data are searched for; and finally, all interference points of preset types found are removed from the point cloud data.

[0073] Because point cloud data contains many types of interference points, such as clusters of interference points caused by multipath interference, interference points at the edge, or isolated suspicious maxima, the three preset types of interference points in this invention can more comprehensively cover the characteristics of all interference points in the point cloud. By finding and eliminating centrifugal point interference points, interference points at the edge can be removed; by finding and eliminating pseudo-maxima interference points, isolated suspicious maxima can be removed; and by finding and eliminating interference points of the interference point cluster type, clusters of interference points caused by multipath interference can be removed. Therefore, the accuracy of target object recognition can be improved by using these three methods of finding and eliminating interference points. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0075] Figure 1 is a flowchart of a method for removing interference points from a point cloud provided by an embodiment of the present invention;

[0076] Figure 2 1 is a schematic diagram of the distribution of interference points of the pseudo highest point type provided by an embodiment of the present invention;

[0077] Figure 3 Schematic diagram of the distribution of interference point groups provided by an embodiment of the present invention;

[0078] Figure 4 is a structural diagram of a device for removing interference points in a point cloud provided by an embodiment of the present invention;

[0079] Figure 5 is a schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0080] In the following description, specific details such as particular system structures and techniques are provided for purposes of illustration, not limitation, to facilitate a thorough understanding of the embodiments of the present invention. However, it will be apparent to those skilled in the art that the present invention may be practiced in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present invention with unnecessary detail.

[0081] In order to make the purpose, technical solutions and advantages of the present invention more clear, specific embodiments will be described below with reference to the accompanying drawings.

[0082] As described in the background technology, when interference points, especially interference points at the edges, exist, they will have a great impact on subsequent modules such as target size assessment, behavior prediction, or posture recognition. Therefore, there is an urgent need for a method to remove interference points in the point cloud with a high elimination rate, thereby improving the accuracy of target object recognition.

[0083] In order to solve the problems in the prior art, embodiments of the present invention provide a method, device, and electronic device for removing interference points from a point cloud. The method for removing interference points from a point cloud provided by embodiments of the present invention is first introduced below.

[0084] The executor of the method for removing interference points in a point cloud may be a device for removing interference points in a point cloud. The device for removing interference points in a point cloud may be an electronic device having a processor and a memory, such as a millimeter-wave radar, and the embodiment of the present invention does not specifically limit this.

[0085] See also Figure 1 , which shows a flowchart of the method for removing interference points in a point cloud provided by an embodiment of the present invention, and is described in detail as follows:

[0086] Step S110: Acquire point cloud data of the target radar frame.

[0087] Point cloud data is the result of scanning using scanning devices such as lidar, millimeter-wave radar, stereo cameras, and time-of-flight cameras. These devices automatically measure a large number of points on an object's surface and output point cloud data. This point cloud data contains the three-dimensional position of each point, specifically its x, y, and z coordinates in three-dimensional space.

[0088] When performing signal processing, radar frames are usually used as processing units. The radar frame sends the clustered point cloud data to a device for removing interference points in the point cloud for subsequent interference point processing.

[0089] Step S120: Search for interference points of a preset type in the point cloud data.

[0090] In some embodiments, based on the position distribution of interference points in the point cloud data and the causes of the interference points, the preset types of interference points can be divided into centrifugal point type, pseudo highest point type and interference point group type.

[0091] Specifically, a centrifugal point type interference point is one whose centrifugal distance from the point cloud data's centroid exceeds a preset distance. A pseudo-peak point type interference point is one where the height difference between the highest and second-highest points in the point cloud data exceeds a preset difference. An interference point cluster type interference point is a cluster of point clouds with abnormal dimensions along a preset dimension.

[0092] In some embodiments, after determining the preset type of interference point, it is necessary to search for the corresponding preset type of interference point in the point cloud data and mark the interference point. For the centrifugal point type of interference point, the following method can be used to search, and the specific steps are:

[0093] S1201. Determine the centroid of the point cloud data.

[0094] The centroid of point cloud data refers to the center of all point clouds. The centroid of point cloud data can be expressed as xCenter, yCenter, and zCenter, and the corresponding calculation formula can be as follows:

[0095]

[0096]

[0097]

[0098] Among them, N represents the number of all points in the point cloud data associated with the current target radar frame, x i 、y i and z i is the 3D coordinate of any point in the point cloud data.

[0099] S1202: Calculate the centrifugal distance between each point in the point cloud data and the centroid.

[0100] The coordinates of the centroid of the point cloud data have been determined in the above steps. Then, the centrifugal distance between each point and the centroid can be calculated based on the three-dimensional coordinates of each point in the point cloud data and the three-dimensional coordinates of the centroid.

[0101] Depending on the application scenario, different methods can be selected to calculate the centrifugal distance.

[0102] Optionally, when global eccentric distance is required, the eccentric distance between each point and the center of mass in three-dimensional space can be selected. Specifically, in three-dimensional space, the eccentric distance between each point and the center of mass can be the Euclidean distance between each point and the center of mass, and the specific calculation formula is:

[0103]

[0104] Where dist(i) represents the centrifugal distance from any point in the point cloud data to the centroid, and x(i), y(i), and z(i) are the three-dimensional coordinate values ​​of any point in the point cloud data.

[0105] Optionally, when you need to focus on the global eccentric distance, you can also choose to express the eccentric distance between each point and the centroid in three-dimensional space as the Manhattan distance between each point and the centroid. The specific calculation formula is:

[0106] dist(i)=|x(i)-xCenter|+|y(i)-yCenter|+|z(i)-zCenter|;

[0107] Optionally, when the centrifugal distance in a certain dimension is of primary interest, the Manhattan distance between the point and the centroid in that dimension can be selected as the centrifugal distance. The specific calculation formula is:

[0108] dist(i) = |z(i) - zCenter|;

[0109] Or dist(i) = |x(i) - xCenter|;

[0110] Or dist(i) = |y(i) - yCenter|.

[0111] Specifically, when the millimeter-wave radar is detecting indoor members, generally speaking, since the data in the z dimension is more discrete, for example, the point cloud data of indoor members near a shaking shower curtain requires the Manhattan distance between each point in the z dimension and the center of mass to eliminate interference points.

[0112] S1203 : Determine all points in the point cloud data whose centrifugal distance from the centroid exceeds a preset distance as interference points of the centrifugal point type.

[0113] Depending on the application scenario, users can set different preset distances. This allows users to determine whether a point in the point cloud data is a centrifugal interference point based on the relationship between the centrifugal distance from the centroid and the preset distance. Specifically, all points in the point cloud data whose centrifugal distance from the centroid exceeds the preset distance can be identified as centrifugal interference points. These centrifugal interference points can then be marked for subsequent removal.

[0114] In some embodiments, since the highest point contains a lot of information, based on the assumption that the point cloud distribution associated with a single target is continuous, discontinuous isolated points are considered interference points. For example, when a millimeter-wave radar detects indoor occupants, the fall data contains the most information in the Z dimension, especially the information represented by the highest point. However, due to the poor stability of the data in this dimension, when the highest point in this dimension is used as the main feature of the detected target, the probability of detection error is high, so the discontinuous isolated points in this dimension must be eliminated. Figure 2 The distribution of interference points of the pseudo-maximum type is shown.

[0115] Depending on the application scenario, a preset difference between the highest point and the second highest point is first set. The coordinate values ​​of all points in the point cloud data in the Z dimension are sorted, and the relationship between the difference between the highest point and the second highest point and the preset difference is used to determine whether it is a pseudo-highest point. If the difference between the highest point and the second highest point is not greater than the preset difference, the current highest point is the reliable highest point. If the difference between the highest point and the second highest point is greater than the preset difference, the current highest point is the pseudo-highest point, and the pseudo-highest point is marked as an interference point of the pseudo-highest point type. The second highest point is then determined as the new highest point, and the relationship between the difference between the highest point and the second highest point and the preset difference is repeated for the remaining points that are not pseudo-highest points. This is done until a reliable highest point is found.

[0116] Specifically, for different application scenarios, users can also extract the coordinate information of a certain dimension based on the amount of information contained in each different dimension, thereby eliminating all interference points in that dimension.

[0117] The following describes in detail the process of finding interference points of the pseudo-highest point type in the Z dimension. Taking the point cloud data of a radar frame as an example, first, the coordinate values ​​of all points in the point cloud data in the Z dimension are extracted and stored in the array tepDataForSort.

[0118] Then, all the data in tepDataForSort is sorted. You can sort from largest to smallest. By default, the first digit is the largest value, and the second digit is the second largest value. You can also sort from smallest to largest. Here, we'll only explain sorting from largest to smallest.

[0119] Next, when the difference between the maximum and the second-largest values ​​is greater than the preset difference, the point corresponding to the maximum is determined to be an outlier, a pseudo-maximum point type interference point. The remaining non-pseudo-maximum point types are re-evaluated, with the second-largest value as the new highest point, and the point immediately following the second-largest value as the new second-highest point. The difference between the new maximum and the new second-largest values ​​is again calculated to see if it is greater than the preset difference. If it is still greater than the preset difference, the new maximum is determined to be a pseudo-maximum point type interference point, and the subsequent points are re-evaluated. If the difference between the new maximum and the new second-largest values ​​is not greater than the preset difference, the point corresponding to the new maximum is determined to be the true highest point, and the evaluation stops.

[0120] In this way, the interference point of the pseudo highest point type in the Z dimension can be found.

[0121] Similarly, users can also search for interference points of the pseudo-highest point type in the X dimension or Y dimension according to their needs, which will not be described here.

[0122] In some embodiments, when a millimeter-wave radar is used to detect indoor occupants, the millimeter waves will inevitably be scattered multiple times due to the narrow indoor space. In a narrow space, metal, walls, floors, etc. may all become mirrors of millimeter waves, and the interference points caused by mirror-like reflections are called multipath interference points. These points due to multipath interference will also form groups, but they are different from isolated points and are called interference point groups. When a person is close to the wall, the multipath interference points will be closer to the person's real echo points. In the group tracking module, these interference point groups will also be associated with the target, thereby affecting the recognition results. Figure 3 The interference point group distribution diagram is shown. Interference point groups with abnormal sizes in the preset dimensions of the point cloud data are determined as interference points of the interference point group type.

[0123] The following describes a detailed method for finding interference points of the interference point cluster type. Taking the point cloud data of a radar frame as an example, first, when a preset dimension is an anomaly dimension, the interval consisting of the maximum and minimum values ​​of the point cloud data along the preset dimension is divided into multiple subintervals at preset intervals. Next, if the number of points within all subintervals exceeds a preset point count threshold, all points within the subinterval are determined to be interference points of the interference point cluster type, without the presence of an interference point cluster. Otherwise, subintervals with less than the preset point count threshold are determined to be broken subintervals. Except for the broken subintervals, the broken subintervals divide the multiple subintervals into multiple continuous intervals. Next, non-interference subintervals are determined within the multiple continuous intervals based on the non-interference subinterval determination criteria. Finally, points corresponding to all subintervals within the multiple subintervals, excluding the non-interference subintervals, are determined to be interference points of the interference point cluster type, with anomaly in size.

[0124] Optionally, to determine whether any preset dimension within the XOY plane has a dimensional anomaly, the size of the point cloud data in the preset dimension and the coordinate Z value of the highest point in the Z dimension can be first obtained. Then, if the size in the preset dimension is greater than a preset size threshold corresponding to the Z value, the preset dimension is determined to be an abnormal dimension. The preset dimension is the X dimension or the Y dimension. Different Z values ​​within different ranges correspond to different preset size thresholds. The size in the preset dimension is the difference between the maximum and minimum coordinate values ​​in the preset dimension.

[0125] Optionally, since point cloud data without interfering point clusters is highly concentrated, the point cloud distribution standard deviation can be used to determine whether any preset dimension within the XOY plane has a dimensional anomaly. When the calculated standard deviation is greater than a preset standard deviation threshold, the preset dimension is determined to be an abnormal dimension. Specifically, the preset standard deviation threshold can be in the range of 0.3-0.5.

[0126] The corresponding standard deviation calculation formula is as follows:

[0127]

[0128]

[0129] Where stdValueX is the standard deviation of the distribution in the X dimension, and stdValueY is the standard deviation of the distribution in the Y dimension.

[0130] Optionally, when it is determined that there is an abnormal dimension, it is necessary to extract the values ​​of all points in the direction of the abnormal dimension, and divide the interval consisting of the maximum and minimum values ​​on the abnormal dimension into multiple sub-intervals according to the preset interval. For example: when the abnormal dimension is the X dimension, the range of x on the X dimension is (-0.3, 0.8), and the preset interval is 0.2, then the interval consisting of the maximum and minimum values ​​is divided into multiple sub-intervals, which are (-0.3, -0.1], (-0.1, 0.1], (0.1, 0.3], (0.3, 0.5], (0.5, 0.7], (0.7, 0.8).

[0131] Optionally, if the number of points corresponding to all subintervals in the point cloud data is not less than a preset point count threshold, it is determined that there are no interference points of the interference point group type with abnormal size in the point cloud data of the target radar frame; otherwise, the subinterval with the number of points less than the preset point count threshold is determined as a broken subinterval.

[0132] After determining that there is a broken subinterval within multiple subintervals, it is necessary to judge the points within the continuous interval composed of other subintervals except the broken subinterval to find the interference points. Still taking the above (-0.3, -0.1], (-0.1, 0.1], (0.1, 0.3], (0.3, 0.5], (0.5, 0.7], (0.7, 0.8) as an example, the number of points in each subinterval is counted. For the convenience of description, the subintervals (-0.3, -0.1], (-0.1, 0.1], (0.1, 0.3], (0.3, 0.5], (0.5, 0.7], (0.7, 0.8) are numbered in sequence here, respectively as 1-6 subintervals. If the number of points in subintervals 1-4 and subinterval 6 is greater than or equal to the preset point number threshold, but the number of points in subinterval 5 is less than the preset point number threshold, then it can be determined that subinterval 5 is a broken subinterval, indicating that there are interference points of the interference point group type with abnormal size in the point cloud data, and it is necessary to further search for interference points of the interference point group type with abnormal size.

[0133] Based on the broken subinterval 5, the subintervals other than the broken subinterval 5 can be divided into two continuous intervals, one for 1-4 and one for 6. Then, based on the non-interference subinterval judgment conditions, a non-interference subinterval is determined from the multiple continuous intervals. Of course, depending on the scenario, N broken subintervals may appear, and the subintervals other than the broken subinterval will be divided into N+1 continuous intervals. This will not be repeated here.

[0134] The non-interference subinterval judgment condition may be:

[0135] The continuous interval containing the largest number of points is the non-interference subinterval;

[0136] Alternatively, the continuous interval containing the points with the smallest average distance is the non-interference subinterval;

[0137] Alternatively, the continuous interval containing the maximum sum of the signal-to-noise ratios of the points is the non-interference subinterval.

[0138] As long as any one of the above conditions is met, it can be determined as a non-interference subinterval.

[0139] Finally, after the non-interference subintervals are determined, the points corresponding to all subintervals except the non-interference subintervals in the multiple subintervals can be determined as interference points of the interference point group type with abnormal size.

[0140] Specifically, taking an arbitrary target radar frame as an example, a method for finding interference points of the interval point group type with abnormal size is given below:

[0141] When millimeter-wave radar is used for indoor person detection, interference points are inevitably associated when clustering the person point cloud. Especially when these interference points are distributed at the edges, they can significantly affect the size estimation of the target person. The width of a person on a two-dimensional plane varies at different heights. For example, when a person is standing, the width is narrower. When a person is sitting, the width is slightly wider. However, when a person is lying down, the maximum width is close to their height. Therefore, when processing point cloud data, it is necessary to first determine the height of the highest point in the Z dimension, that is, the person's height, so as to determine the size in other dimensions. Different preset size thresholds are determined according to different height ranges. See Table 1 for details.

[0142] Table 1

[0143] Height range [1.2,2) [0.6,1.2) [0,0.6) Preset size threshold 0.4-0.8 0.5-1.0 1.0-1.5

[0144] First, based on the height of the highest point in the Z dimension, the preset size threshold corresponding to that height can be determined. The size in the X or Y dimension can be determined by the distance between the highest and lowest points in the X or Y dimension, and compared with the preset size threshold to determine whether the size is abnormal. Alternatively, the standard deviation of the distribution in the X or Y dimension can also be used to determine whether the size is abnormal. If the size is normal, there are no interference points. Otherwise, there are interference points of the interval point group type with abnormal size in the point cloud data, which need to be found.

[0145] Then, based on the aforementioned X or Y dimension, we can determine which dimension is the outlier. We extract the maximum and minimum values ​​of all points in the point cloud data along the outlier dimension. We then divide the interval between the maximum and minimum values ​​into multiple subintervals based on a preset interval, and count the points within each subinterval. The preset intervals can vary depending on the height range, as detailed in Table 2.

[0146] Table 2

[0147] Height range [1.2,2) [0.6,1.2) [0,0.6) Preset size threshold 0.4-0.8 0.5-1.0 1.0-1.5 Preset interval 0.1-0.4 0.2-0.5 0.3-0.6

[0148] Then, based on the relationship between the number of points in each subinterval and the preset point threshold, it can be determined whether there is a broken subinterval in all subintervals. When the number of points in the target subinterval is greater than or equal to the preset point threshold, the target subinterval is a non-broken subinterval. When the number of points in the target subinterval is less than the preset point threshold, the target subinterval is a broken subinterval. In addition to the broken subinterval, the subintervals on the left and right sides of the broken subinterval form continuous intervals respectively. Specifically, according to the number of broken subintervals, it can be divided into at least two continuous intervals. For example, if there is 1 broken subinterval, the entire interval except the broken subinterval is divided into two continuous intervals. If there are 2 broken subintervals, the entire interval except the broken subinterval is divided into three continuous intervals. The number of continuous intervals is inferred according to the number of broken subintervals, and will not be repeated here. Among them, the preset point threshold can be set by the user according to different scenarios, and can be set to 0 or any other number. The target subinterval is any interval within all subintervals.

[0149] Finally, according to the above-mentioned non-interference sub-interval judgment conditions, it is determined which of the continuous intervals is the real target, and the points in other intervals are determined to be interference points of the interval point group type with abnormal size and are marked.

[0150] Step S130: remove all found interference points of preset types from the point cloud data.

[0151] By setting three different preset types of interference points, and using different methods to find and mark interference points that meet the three preset types, and finally removing all the found interference points of the preset types from the point cloud, we can obtain non-interference point cloud data composed of non-interference points. This can be used for subsequent processing such as size assessment, behavior prediction, or posture recognition.

[0152] It should be noted that the three preset types of interference point elimination methods in the present invention can be combined according to different application scenarios to achieve a good elimination effect. For example, the three preset types of interference points existing in the point cloud data can be searched first, and then the three preset types of interference points can be eliminated. Alternatively, based on the importance of the interference points to the detection target or the final elimination effect, the preset types of interference points that are found first can be set to be eliminated, and then the remaining preset types of interference points can be searched again for the remaining point cloud data and eliminated, until all preset types of interference points in the point cloud data are eliminated.

[0153] Specifically, taking any target radar frame as an example, a method for removing interference points is given as follows:

[0154] First, centrifugal interference points are searched for in the point cloud data and removed from the point cloud data. Then, pseudo-peak interference points are searched for in the remaining point cloud data and removed. Finally, cluster interference points are searched for in the remaining point cloud data and removed. This allows for a relatively thorough removal of interference points in the target radar frame caused by multiple factors, such as edge interference points, isolated suspicious peaks, and cluster interference points caused by multipath interference. This allows for higher confidence and more reliable stability in subsequent processes such as size assessment, behavior prediction, or gesture recognition. Furthermore, the interference point removal method provided by the present invention requires minimal computational effort, consumes minimal memory, and offers high real-time performance. It should be noted that the present invention is not only applicable to indoor point cloud processing but can also be used in other scenarios where interference points are present, by setting different preset distances, preset difference values, preset point count thresholds, and preset size thresholds, depending on the application scenario.

[0155] In an embodiment of the present invention, point cloud data of a target radar frame is first acquired; then, interference points of a preset type are searched for in the point cloud data, and finally, all interference points of the preset type found are removed from the point cloud data.

[0156] Because point cloud data contains many types of interference points, such as clusters of interference points caused by multipath interference, interference points at the edge, or isolated suspicious maxima, the three preset types of interference points in this invention can more comprehensively cover the characteristics of all interference points in the point cloud. By finding and eliminating centrifugal point interference points, interference points at the edge can be removed; by finding and eliminating pseudo-maxima interference points, isolated suspicious maxima can be removed; and by finding and eliminating interference points of the interference point cluster type, clusters of interference points caused by multipath interference can be removed. Therefore, the accuracy of target object recognition can be improved by using these three methods of finding and eliminating interference points.

[0157] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0158] Based on the method provided in the above embodiment, the present invention also provides a specific implementation of a device applied to the method. Please refer to the following embodiment.

[0159] like Figure 4 As shown, a device 400 for removing interference points in a point cloud is provided, the device comprising:

[0160] An acquisition module 410 is used to acquire point cloud data of a target radar frame;

[0161] A search module 420 is configured to search for interference points of a preset type in the point cloud data; wherein the preset type includes at least one of a centrifugal point type, a pseudo-highest point type, and an interference point group type;

[0162] The elimination module 430 is configured to eliminate all interference points of a preset type found from the point cloud data.

[0163] In a possible implementation, the search module 420 and the elimination module 430 are further configured to:

[0164] Find interference points of the centrifugal point type in the point cloud data, and remove the interference points of the centrifugal point type found from the point cloud data;

[0165] In the point cloud data from which the interference points of the centrifugal point type are eliminated, interference points of the pseudo-highest point type are searched and the interference points of the pseudo-highest point type found are eliminated;

[0166] In the point cloud data from which interference points of the centrifugal point type and the pseudo-highest point type are eliminated, interference points of the interference point group type are searched and the found interference points of the interference point group type are eliminated.

[0167] In a possible implementation, the search module 420 is further configured to:

[0168] Determine the centroid of point cloud data;

[0169] Calculate the centrifugal distance between each point in the point cloud data and the centroid respectively;

[0170] All points in the point cloud data whose centrifugal distance from the centroid exceeds a preset distance are determined as interference points of the centrifugal point type.

[0171] In one possible implementation, the centrifugal distance from the center of mass is:

[0172] Euclidean distance from the center of mass in three-dimensional space;

[0173] or, in three-dimensional space, the Manhattan distance from the centroid;

[0174] Alternatively, the Manhattan distance from the centroid in any dimension.

[0175] In a possible implementation, the search module 420 is further configured to:

[0176] Get the coordinate value of each point in the point cloud data in the Z dimension and sort them;

[0177] Determine whether the highest point is a pseudo-highest point: If the difference between the coordinate values ​​of the highest point and the second-highest point is greater than the preset difference, the highest point is determined to be a pseudo-highest point, otherwise the highest point is determined to be a reliable highest point; when the highest point is a pseudo-highest point, the second-highest point is set as the new highest point, and the steps of determining whether the highest point is a pseudo-highest point are repeated until a reliable highest point is found.

[0178] In a possible implementation, the search module 420 is further configured to:

[0179] In the case where the preset dimension is an abnormal dimension, the interval consisting of the maximum and minimum values ​​of the point cloud data in the preset dimension is divided into multiple sub-intervals according to preset intervals; wherein the preset dimension is the X dimension or the Y dimension;

[0180] If the number of points in all subintervals is greater than the preset point number threshold, then the points in all subintervals are determined to be interference points without interference point group type; otherwise, the subintervals with the number of points less than the preset point number threshold are determined to be broken subintervals; except for the broken subintervals, the broken subintervals divide multiple subintervals into multiple continuous intervals;

[0181] Determining a non-interference sub-interval from a plurality of continuous intervals according to a non-interference sub-interval judgment condition;

[0182] Points corresponding to all subintervals except the non-interference subinterval in the plurality of subintervals are determined as interference points of the interference point group type.

[0183] In a possible implementation, the search module 420 is further configured to:

[0184] Obtain the size of the point cloud data in a preset dimension and the coordinate Z value of the highest point in the Z dimension; if the size in the preset dimension is greater than the preset size threshold corresponding to the Z value, determine that the preset dimension is an abnormal dimension;

[0185] Alternatively, the standard deviation of all dimensional coordinates of the point cloud data in a preset dimension is obtained; if the standard deviation is greater than a preset standard deviation threshold, the preset dimension is determined to be an abnormal dimension.

[0186] In one possible implementation, the non-interference subinterval determination condition includes:

[0187] The continuous interval containing the largest number of points is the non-interference subinterval;

[0188] Alternatively, the continuous interval containing the points with the smallest average distance is the non-interference subinterval;

[0189] Alternatively, the continuous interval containing the maximum sum of the signal-to-noise ratios of the points is the non-interference subinterval.

[0190] Figure 5Schematic diagram of an electronic device provided by an embodiment of the present invention. Figure 5 As shown, the electronic device 5 of this embodiment includes: a processor 50, a memory 51, and a computer program 52 stored in the memory 51 and executable on the processor 50. When the processor 50 executes the computer program 52, the steps in the above-mentioned embodiments of the method for removing interference points from a point cloud are implemented, for example Figure 1 Alternatively, when the processor 50 executes the computer program 52, the functions of the modules in the above-mentioned device embodiments are realized, for example, Figure 4 Functions of modules 410 to 430 are shown.

[0191] Exemplarily, the computer program 52 may be divided into one or more modules, which are stored in the memory 51 and executed by the processor 50 to implement the present invention. The one or more modules may be a series of computer program instruction segments capable of implementing specific functions, which are used to describe the execution process of the computer program 52 in the electronic device 5. For example, the computer program 52 may be divided into Figure 4 Modules 410 to 430 are shown.

[0192] The electronic device 5 may include, but is not limited to, a processor 50 and a memory 51. Those skilled in the art will understand that Figure 5 It is only an example of the electronic device 5 and does not constitute a limitation of the electronic device 5. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the electronic device may also include input and output devices, network access devices, buses, etc.

[0193] The processor 50 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0194] The memory 51 can be an internal storage unit of the electronic device 5, such as a hard disk or memory of the electronic device 5. The memory 51 can also be an external storage device of the electronic device 5, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device 5. Furthermore, the memory 51 can also include both an internal storage unit of the electronic device 5 and an external storage device. The memory 51 is used to store the computer program and other programs and data required by the electronic device. The memory 51 can also be used to temporarily store data that has been output or is about to be output.

[0195] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0196] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0197] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0198] In the embodiments provided by the present invention, it should be understood that the disclosed devices / electronic devices and methods can be implemented in other ways. For example, the device / electronic device embodiments described above are merely schematic. For example, the division of the modules or units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0199] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0200] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0201] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, it can implement the steps of the above-mentioned method embodiments for removing interference points in the point cloud. The computer program includes computer program code, which can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device that can carry the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practices in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practices, computer-readable media does not include electrical carrier signals and telecommunication signals.

[0202] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A method for removing interference points in a point cloud, characterized in that: include: Get the point cloud data of the target radar frame; Searching for interference points of a centrifugal point type in the point cloud data, and removing the interference points of the centrifugal point type found from the point cloud data; wherein the interference points of the centrifugal point type are interference points whose centrifugal distance from the centroid of the point cloud data exceeds a preset distance; In the point cloud data from which the centrifugal point type interference points are eliminated, interference points of the pseudo-highest point type are searched, and the interference points of the pseudo-highest point type that are found are eliminated; wherein the interference points of the pseudo-highest point type are interference points in the point cloud data where the height difference between the highest point and the second highest point is greater than a preset difference; In the point cloud data from which the interference points of the centrifugal point type and the pseudo-highest point type are eliminated, searching for interference points of the interference point group type, and eliminating the interference points of the interference point group type that are found; wherein the interference points of the interference point group type are interference point groups with abnormal sizes in a preset dimension of the point cloud data; The step of searching for interference points of the interference point group type includes: Obtain the size of the point cloud data in a preset dimension and the coordinate Z value of the highest point in the Z dimension; if the size in the preset dimension is greater than a preset size threshold corresponding to the Z value, determine that the preset dimension is an abnormal dimension; or obtain the standard deviation of all coordinates of the point cloud data in the preset dimension; if the standard deviation is greater than a preset standard deviation threshold, determine that the preset dimension is an abnormal dimension; In the case where the preset dimension is an abnormal dimension, dividing an interval consisting of the maximum value and the minimum value of the point cloud data in the preset dimension into a plurality of sub-intervals according to preset intervals; wherein the preset dimension is the X dimension or the Y dimension; If the number of points in all the subintervals is greater than a preset point number threshold, then all the points in the subintervals are determined to be interference points without an interference point group type; otherwise, the subintervals with the number of points less than the preset point number threshold are determined as broken subintervals; except for the broken subintervals, the broken subintervals divide the multiple subintervals into multiple continuous intervals; determining a non-interference sub-interval from the plurality of continuous intervals according to a non-interference sub-interval determination condition; Points corresponding to all subintervals except the non-interference subinterval in the plurality of subintervals are determined as interference points of an interference point group type.

2. The method for removing interference points in a point cloud according to claim 1, wherein: The step of searching for interference points of the centrifugal point type in the point cloud data includes: determining a centroid of the point cloud data; Calculating the centrifugal distance between each point in the point cloud data and the centroid respectively; All points in the point cloud data whose centrifugal distance from the centroid exceeds a preset distance are determined as interference points of the centrifugal point type.

3. The method for removing interference points in a point cloud according to claim 2, wherein: The centrifugal distance from the center of mass is: the Euclidean distance from said centroid in three-dimensional space; or, the Manhattan distance from said centroid in three-dimensional space; Alternatively, the Manhattan distance from the centroid in any dimension.

4. The method for removing interference points in a point cloud according to claim 1, wherein: The step of searching for interference points of the pseudo highest point type includes: Obtaining the coordinate value of each point in the point cloud data in the Z dimension and sorting them; Determine whether the highest point is a pseudo-highest point: if the difference between the coordinate value of the highest point and the coordinate value of the second-highest point is greater than the preset difference, the highest point is determined to be a pseudo-highest point, otherwise the highest point is determined to be a reliable highest point; when the highest point is a pseudo-highest point, the second-highest point is set as the new highest point, and the step of determining whether the highest point is a pseudo-highest point is repeated until a reliable highest point is found.

5. The method for removing interference points in a point cloud according to claim 1, wherein: The non-interference subinterval judgment conditions include: The continuous interval containing the largest number of points is the non-interference subinterval; Alternatively, the continuous interval containing the smallest average distance of the points is the non-interference subinterval; Alternatively, the continuous interval containing the maximum sum of the signal-to-noise ratios of the points is the non-interference subinterval.

6. A device for removing interference points in a point cloud, characterized in that: include: Acquisition module, used to obtain point cloud data of target radar frame; A search module and a removal module, configured to search for interference points of the centrifugal point type in the point cloud data, and remove the interference points of the centrifugal point type found from the point cloud data; In the point cloud data from which the interference points of the centrifugal point type are eliminated, searching for interference points of the pseudo-highest point type, and eliminating the interference points of the pseudo-highest point type that are found; In the point cloud data from which the interference points of the centrifugal point type and the pseudo-highest point type are eliminated, searching for interference points of the interference point group type, and eliminating the interference points of the interference point group type that are found; Among them, the interference point of the centrifugal point type is an interference point whose centrifugal distance from the centroid of the point cloud data exceeds a preset distance; the interference point of the pseudo-highest point type is an interference point whose height difference between the highest point and the second highest point in the point cloud data is greater than a preset difference; the interference point of the interference point group type is an interference point group with abnormal size in the preset dimension of the point cloud data; The search module and the elimination module are specifically used to obtain the size of the point cloud data in a preset dimension and the coordinate Z value of the highest point in the Z dimension; if the size in the preset dimension is greater than a preset size threshold corresponding to the Z value, the preset dimension is determined to be an abnormal dimension; or, obtain the standard deviation of all coordinates of the point cloud data in the preset dimension; if the standard deviation is greater than a preset standard deviation threshold, the preset dimension is determined to be an abnormal dimension; if the preset dimension is an abnormal dimension, the interval formed by the maximum and minimum values ​​of the point cloud data in the preset dimension is divided into multiple sub-intervals according to preset intervals; Wherein, the preset dimension is the X dimension or the Y dimension; if the number of points in all the sub-intervals is greater than the preset point threshold, then all the points in the sub-intervals are determined to be interference points of the type without interference point group, otherwise the sub-interval with the number of points less than the preset point threshold is determined as a broken sub-interval; except for the broken sub-interval, the broken sub-interval divides the multiple sub-intervals into multiple continuous intervals; according to the non-interference sub-interval judgment condition, non-interference sub-intervals are determined in the multiple continuous intervals; points corresponding to all sub-intervals in the multiple sub-intervals except the non-interference sub-interval are determined as interference points of the interference point group type.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

Images