Radar, data processing method and device for radar, and readable storage medium

Abstract

Radar, data processing method and device for radar, and readable storage medium. Among them, the data processing method for radar includes: 1) obtaining the point cloud of the radar, where the point cloud includes information points obtained by multiple detectors at multiple angles; 2) selecting a point to be determined from the point cloud; 3) using the detector corresponding to the point to be determined as the reference detector, and using the angle corresponding to the point to be determined as the reference angle, finding the detection information corresponding to the adjacent angle of the reference angle and corresponding to the reference detector, and the detection information corresponding to the adjacent detector of the reference detector and corresponding to the reference angle, and using the obtained information points as reference points; 4) judging whether the point to be determined is a noise point based on the correlation between the point to be determined and the reference points. By adopting the above solution, the accuracy of the noise point recognition result can be improved, the misjudgment probability can be reduced, and the data quality of the point cloud is guaranteed.

Description

Technical Field

[0001] Embodiments of this specification relate to the field of radar technology, and in particular, to a radar, a data processing method and device for a radar, and a readable storage medium. Background Art

[0002] Currently, a radar can calculate information points in three-dimensional space by transmitting detection signals to the surrounding environment and receiving echo signals reflected by obstacles, and then obtain a point cloud in three-dimensional space. However, in addition to echo signals, the signals received by the radar may also be interference signals, such as interference signals generated by other electronic devices, interference signals formed by the radar itself, interference signals in the external environment, etc. The interference signals will cause noise points in the point cloud, thereby reducing the accuracy of subsequent processing.

[0003] In some applications, noise points can be identified and filtered by setting a threshold to judge the information points collected by the same detector. For example, if a detector in a radar measures information points PC x-1 、t x and t x+1 respectively at times t X-1 、PC X and PC x+1 , then parameters such as the distances and reflectivities of the three information points PC X-1 、PC X and PC x+1 are obtained, and the correlation degrees between the information points PC X and PC X-1 , and between the information points PC X and PC X+1 are determined according to these parameters. If the correlation degree is lower than a preset threshold, it is determined that the information point PC X is a noise point, and then the information point PC X is filtered.

[0004] However, due to reasons such as small obstacles or low radar resolution, the information points collected by the same detector at different times may be sparsely distributed and weakly correlated. If the threshold is set too high, useful information points (i.e., non-noise information points) with sparse distribution may be misjudged as noise points, and then the useful information points are filtered, resulting in the problem of missing points. In order to try to retain these sparse useful information points, the threshold can only be lowered and the judgment scale can be relaxed, but at the same time, the risk of misjudging noise points as useful information points is increased. Therefore, the accuracy of noise filtering in the existing solutions is relatively low. Summary of the Invention

[0005] In view of this, embodiments of the present specification provide a radar, a data processing method and device for the radar, and a readable storage medium, which can improve the accuracy of the noise point recognition result, reduce the misjudgment probability, and ensure the data quality of the point cloud.

[0006] The present specification provides a data processing method for a radar, including the following steps:

[0007] 1) Obtain the point cloud of the radar, where the point cloud includes: information points obtained by multiple detectors at multiple angles;

[0008] 2) Select a point to be determined from the point cloud;

[0009] 3) Take the detector corresponding to the point to be determined as the reference detector, take the angle corresponding to the point to be determined as the reference angle, find the detection information corresponding to the adjacent angle of the reference angle and corresponding to the reference detector, and the detection information corresponding to the adjacent detector of the reference detector and corresponding to the reference angle, and use the obtained information points as reference points;

[0010] 4) Based on the correlation between the point to be determined and the reference points, determine whether the point to be determined is a noise point.

[0011] Optionally, before the step 4), the following steps are further included:

[0012] A) Find the detection information corresponding to the adjacent detector of the adjacent angle of the reference angle and corresponding to the reference detector, and use the obtained information points as reference points.

[0013] Optionally, the adjacent detector is one or more detectors adjacent to the reference detector in the arrangement position;

[0014] The adjacent angle is one or more angles whose angle difference from the reference angle belongs to a preset range.

[0015] Optionally, for the detection information of the reference detector, the adjacent angle is an angle whose angle difference from the reference angle belongs to a first range;

[0016] For the detection information of the adjacent detector, the adjacent angle is an angle whose angle difference from the reference angle belongs to a second range;

[0017] Wherein, the span of the first range is greater than the span of the second range.

[0018] Optionally, the span difference between the first range and the second range is positively correlated with the field of view angle difference between the detectors.

[0019] Optionally, when the adjacent detectors include a plurality of detectors arranged on one side of the reference detector in the arrangement position, the second range includes a plurality of angular ranges corresponding to the plurality of detectors respectively, and the span of the angular range is negatively correlated with the distance from the adjacent detector to the reference detector in the arrangement position.

[0020] Optionally, step 4) includes:

[0021] 4-1) Determine whether the point to be determined is a noise point according to the point distance between the point to be determined and the reference point.

[0022] Optionally, step 4-1) includes the following steps:

[0023] 4-11) Calculate the distances between each reference point and the point to be determined respectively to obtain a point distance data set;

[0024] 4-12) Determine a point distance threshold;

[0025] 4-13) Determine whether there is point distance data lower than the point distance threshold in the point distance data set. If so, determine that the point to be determined is not a noise point; otherwise, determine that the point to be determined is a noise point.

[0026] Optionally, step 4-12) includes the following steps:

[0027] 4-121) Obtain the obstacle distance corresponding to the point to be determined;

[0028] 4-122) Determine the point distance threshold of the point to be determined based on the corresponding relationship between the obstacle distance and the point distance threshold.

[0029] Optionally, the point distance threshold is positively correlated with the obstacle distance corresponding to the point to be determined, and the point distance threshold is negatively correlated with the resolution of the radar.

[0030] Optionally, when the point loss rate of the radar is higher than the preset point loss rate threshold, increase the point distance threshold.

[0031] Optionally, step 4) includes:

[0032] 4-2) Determine whether the point to be determined is a noise point according to the number of reference points.

[0033] Optionally, step 4-2) includes the following steps:

[0034] 4-21) Determine whether the number of reference points is lower than a preset number threshold. If so, determine that the point to be determined is a noise point; otherwise, determine that the point to be determined is not a noise point.

[0035] Optionally, the point threshold is positively correlated with the resolution of the radar.

[0036] Optionally, when the point loss rate of the radar is higher than a preset point loss rate threshold, the point threshold is increased.

[0037] Optionally, the data processing method for the radar further includes:

[0038] 5) If the point to be determined is determined as a noise point, delete the point to be determined.

[0039] Optionally, the data processing method for the radar further includes:

[0040] 6) Determine whether there is an information point for which noise point judgment has not been performed. If so, select the point to be determined from the information points for which noise point judgment has not been performed, and continue to execute step 3).

[0041] This specification also provides a data processing device for a radar, including:

[0042] A data storage unit, adapted to cache the point cloud of the radar, the point cloud including information points obtained by a plurality of detectors at a plurality of angles;

[0043] A noise point recognition unit, adapted to select a point to be determined from the point cloud; taking the detector corresponding to the point to be determined as a reference detector, taking the angle corresponding to the point to be determined as a reference angle, finding the detection information corresponding to the reference detector at an adjacent angle corresponding to the reference angle and the detection information corresponding to an adjacent detector of the reference detector at the reference angle, and using the obtained information points as reference points; and, based on the correlation between the point to be determined and the reference points, determining whether the point to be determined is a noise point.

[0044] Optionally, the noise point recognition unit is further adapted to find the detection information of an adjacent detector corresponding to an adjacent angle corresponding to the reference angle, and use the obtained information points as reference points.

[0045] Optionally, the adjacent detector is one or more detectors adjacent to the reference detector in the arrangement position;

[0046] The adjacent angle is one or more angles whose angle difference from the reference angle belongs to a preset range.

[0047] Optionally, for the detection information of the reference detector, the adjacent angle is an angle whose angle difference from the reference angle belongs to a first range;

[0048] For the detection information of the adjacent detector, the adjacent angle is an angle whose angle difference from the reference angle belongs to a second range;

[0049] Among them, the span of the first range is greater than the span of the second range.

[0050] Optionally, the relevance between the point to be determined and the reference point includes: the point distance between the point to be determined and the reference point.

[0051] Optionally, the relevance between the point to be determined and the reference point includes: the number of points of the reference point.

[0052] Optionally, the noise point recognition unit is further adapted to delete the point to be determined judged as a noise point.

[0053] Optionally, the noise point recognition unit is further adapted to select the point to be determined from the information points that have not been judged for noise points and perform noise point judgment.

[0054] This specification also provides a radar, including a plurality of transmitters, a plurality of detectors and a data processing device, wherein:

[0055] There is a corresponding relationship between the transmitter and the detector; the transmitter is adapted to emit detection signals at multiple angles; the detector is adapted to collect echo signals at multiple angles;

[0056] The data processing device is adapted to perform data processing according to the detection signal of the transmitter and the echo signal of the corresponding detector to obtain corresponding information points, and execute the method described in any one of the above embodiments.

[0057] Optionally, the radar is a lidar.

[0058] This specification also provides a computer-readable storage medium, on which computer instructions are stored, and when the computer instructions run, they execute the steps of the method described in any one of the above embodiments.

[0059] By using the data processing method for radar according to the embodiments of this specification, by finding the adjacent angles corresponding to the reference angle and the detection information corresponding to the reference detector, and the detection information corresponding to the adjacent detectors of the reference detector and corresponding to the reference angle, reference points that are associated with the point to be determined in dimensions such as space and time sequence are obtained, improving the reliability and diversity of the reference points. Furthermore, when identifying noise points according to the relevance between the point to be determined and the reference point, the accuracy of the identification result can be improved, the misjudgment probability can be reduced, and thus noise points can be effectively identified, ensuring the data quality of the point cloud.

[0060] Furthermore, the correlation between the point to be determined and the reference point set is characterized by the distances between each reference point in the reference point set and the point to be determined, which can enhance the reliability of the correlation, effectively reduce the problem of lost points caused by misjudgment, improve the accuracy of the noise point recognition result, and ensure the data quality of the point cloud.

[0061] Furthermore, the correlation between the point to be determined and the reference point set is characterized by the number of points of the reference point, which can reduce the calculation amount and improve the noise point recognition efficiency on the premise of ensuring the accuracy of the noise point recognition result. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions of the embodiments of this specification, the following will briefly introduce the drawings required for describing the embodiments of this specification or the prior art. Obviously, the following described drawings are only some embodiments of this specification. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0063] Figure 1a It is a partial field of view schematic diagram of a radar in the existing application.

[0064] Figure 1b It is the corresponding Figure 1a Schematic diagram of the arrangement of information points collected by a partial detector of a radar in three-dimensional space.

[0065] Figure 2 It is a flowchart of a data processing method for a radar in the embodiments of this specification.

[0066] Figure 3 It is a flowchart of another data processing method for a radar in the embodiments of this specification.

[0067] Figure 4 It is a distribution schematic diagram of a radar point cloud in the embodiments of this specification.

[0068] Figures 5 to 11 It is Figure 4 Schematic diagrams of multiple reference point selections corresponding thereto.

[0069] Figure 12 It is the corresponding Figure 1a Schematic diagram of the arrangement of information points collected by a partial detector of another radar in three-dimensional space.

[0070] Figure 13 It is a flowchart of a noise point judgment method in the embodiments of this specification.

[0071] Figure 14 It is a flowchart of a method for determining a point distance threshold in the embodiments of this specification.

[0072] Figure 15 It is a schematic diagram showing the correspondence between the dot pitch threshold and the distance to the obstacle in the embodiments of this specification.

[0073] Figure 16 It is another schematic diagram showing the correspondence between the dot pitch threshold and the distance to the obstacle in the embodiments of this specification.

[0074] Figure 17 It is a block diagram of a data processing device for radar in the embodiments of this specification.

[0075] Figure 18 It is a block diagram of a radar in the embodiments of this specification. Detailed implementation manners

[0076] As can be seen from the background art, if there are noise points in the point cloud, the accuracy of subsequent processing will be reduced. To enable those skilled in the art to more clearly understand the problems existing in the prior art, the following will first make a specific description in combination with the accompanying drawings and existing applications.

[0077] In an existing achievable application, a vehicle is equipped with a radar. The radar emits detection signals to the surrounding environment, and the received signals include echo signals and interference signals reflected by obstacles. The point cloud obtained by this radar contains noise points.

[0078] Figure 1a Shows a partial field of view schematic diagram of the radar; Figure 1b Shows the correspondence of a part of the detectors of a radar Figure 1a Schematic diagram of the arrangement of information points collected by the field of view in three-dimensional space.

[0079] The radar may include multiple detectors. For the sake of convenience of description, one detector TC in this radar X is taken as the illustration object.

[0080] With reference to Figure 1a and Figure 1b , during the actual detection process of the radar, the detector TC X at time t 1 ~t 4 respectively receives the detection signals reflected by the external wall WA. Since the wall WA is relatively wide and has a large horizontal angle span in the field of view of the detector TC X , four relatively strongly correlated useful information points can be collected, that is, Figure 1b the useful information points a 1 ~a 4 .

[0081] The detector TC X at time t 5The field of view at a certain moment corresponds to the telegraph pole GA, and the detection signal reflected by the telegraph pole GA is received. Since the telegraph pole is relatively thin, the horizontal angle span in the field of view of the detector TC X is small, and the detector TC X only collects one useful information point in the width perpendicular to the axial direction of the telegraph pole, that is, Figure 1b the useful information point a in 5 . At the next adjacent detection moment of t 5 , the field of view of the detector TC X deviates from the telegraph pole GA.

[0082] The detector TC X receives the detection signals reflected by the external wall WB at the moments from t 6 to t 9 . Since the wall WB is relatively wide, the horizontal angle span in the field of view of the detector TC X is large, and four highly correlated useful information points are obtained, that is, Figure 1b the useful information point a in 6 to a 9 .

[0083] It can be known from Figure 1b that the correlation between the useful information point a 5 and the useful information point a 4 is weak, and the correlation between the useful information point a 5 and the useful information point a 6 is also weak.

[0084] If the solution of the existing technology is adopted, the useful information points a 4 measured at the moment of t 5 , the useful information points a 6 measured at the moment of t 4 , and the useful information points a 5 measured at the moment of t 6 are obtained respectively, and according to the parameters such as the distance and reflectivity of the useful information point a 4 , the useful information point a 5 , and the useful information point a 6 , the correlation between the useful information point a 4 and the useful information point a 5 , as well as the correlation between the useful information point a 5 and the useful information point a 6 will be lower than the threshold, so that the useful information point a 5 is misjudged as noise and filtered out.

[0085] It can be seen from this that when using the existing technology solution for noise judgment, it may cause misjudgment and deletion of useful information points with relatively low correlation (such as Figure 1b the useful information point a in5 )。

[0086] Therefore, the existing solutions cannot achieve high-precision noise filtering.

[0087] To solve the above problems, this specification provides a data processing method for radar. By combining the angle and the detector position, candidate points to be determined and reference points corresponding to multiple detectors are selected from the point cloud, and then based on the correlation between the candidate points to be determined and the reference points, it is determined whether the candidate points to be determined are noise points. Thereby, the accuracy of the noise point recognition result is improved, the misjudgment probability is reduced, and the data quality of the point cloud is guaranteed.

[0088] To enable those skilled in the art to more clearly understand and implement the concept, implementation solution and advantages of the present invention, the following will refer to the accompanying drawings and be described in detail through specific application scenarios.

[0089] Refer to Figure 2 , which is a flowchart of a data processing method for radar provided by an embodiment of this specification. In the embodiment of this specification, as Figure 2 shown, the method may include the following steps:

[0090] S1. Obtain the point cloud of the radar, where the point cloud includes: information points obtained by multiple detectors at multiple angles.

[0091] In a specific implementation, the radar may include multiple transmitters and multiple detectors. The multiple transmitters may be arranged in a specified direction (such as the vertical direction of the radar), and the detection signals emitted by each transmitter have different angles with the specified direction; there is a corresponding relationship between the detectors and the transmitters, which are used to receive signals in the surrounding environment. When a detector receives a signal (such as an echo signal, an interference signal, etc.), an information point can be calculated through the detection signal of the corresponding transmitter and the signal received by the detector.

[0092] During the operation of the radar, it rotates along a specified axis at a certain speed. According to the set sampling frequency, information points are collected once every certain angle of rotation, so as to collect information around the radar during the rotation process to realize the perception of the surrounding environment. Moreover, the information points obtained after the radar rotates one week form a frame of point cloud.

[0093] It should be noted that during the operation of the radar, the detection information of the detector may include the calculated information points and blank results (that is, no signal is received at this angle). For example, if the detector receives a signal before the radar rotates from angle α 1 to angle α 2 , the corresponding information points can be calculated, and the detection information of the detector at angle α 1 is the information points; otherwise, the detection information of the detector at angle α 1The detection information is a blank result.

[0094] In order to distinguish and know the sources of each information point, each of the information points may include detector data and angle data.

[0095] Taking a conventional mechanical rotating radar as an example, multiple detectors are arranged along the vertical direction of the radar. Different detectors are used to receive echo signals at different vertical angles. Therefore, the information points measured by different detectors can have their corresponding vertical angles known according to the detector positions.

[0096] The vertical angle difference between two adjacent detectors is the vertical angular resolution of the radar. Thus, the detection information of the adjacent detector corresponding to the reference detector described in this specification can be understood as: the detection information of the detector whose corresponding vertical angle is separated from the reference detector of the point to be determined by one or a continuous number of vertical angular resolutions.

[0097] The radar can rotate in the 360° horizontal direction. After the radar completes a rotation of a horizontal angle, multiple transmitters and detectors are started for sequential polling to detect the outside world. After all detectors complete the polling, the detection information corresponding to the radar's vertical field of view (FOV) at this horizontal angle is obtained.

[0098] After completing the detection at one horizontal angle, the radar rotates to another horizontal angle for another round of polling detection.

[0099] The horizontal angle difference between two adjacent signal detections by the same detector can be the horizontal angular resolution of the radar. Thus, the detection information corresponding to adjacent angles described in this specification can be understood as: the detection information whose corresponding horizontal angle is separated from the reference angle of the point to be determined by one or a continuous number of horizontal angular resolutions.

[0100] As can be seen from the above, the detector data can be used to characterize the detector corresponding to the information point, and the angle data can be used to characterize the angle by which the detector corresponding to the information point has rotated relative to the reference position.

[0101] Among them, the angle is related to the rotation speed and sampling frequency of the radar. And, the reference position can be the initial position of the radar, and the angle can be the included angle between the starting position and the current position of the corresponding detector.

[0102] S2. Select the point to be determined from the point cloud.

[0103] In specific implementation, the selection method of the point to be determined can be set according to the actual application scenario and requirements. For example, information points can be sequentially selected from the point cloud in a specified order as the points to be determined. Another example is that some information points can be selected from the point cloud as the points to be determined. The embodiments of this specification do not limit this.

[0104] S3, taking the detector corresponding to the point to be determined as the reference detector and the angle corresponding to the point to be determined as the reference angle, find the detection information corresponding to the adjacent angle of the reference angle and corresponding to the reference detector, and the detection information corresponding to the adjacent detector of the reference detector and corresponding to the reference angle, and use the obtained information points as reference points.

[0105] In specific implementation, according to the detector data of the point to be determined, determine the corresponding detector and use this detector as the reference detector. According to the angle data of the point to be determined, determine the corresponding angle and use this angle as the reference angle. Then, based on the reference detector and the reference angle, find the detection information whose detector and angle meet the preset adjacent conditions from the detection information of each detector, and use the information points obtained from the detection information that meets the preset adjacent conditions as reference points, thereby determining the information points in the adjacent spatial domain of the point to be determined.

[0106] Among them, the adjacent detector can be one or more detectors adjacent to the reference detector in the arrangement position; the adjacent angle can be one or more angles whose angular difference from the reference angle is within a preset range.

[0107] In other words, the reference points can include: other information points calculated by the radar when rotating to the reference angle except the point to be determined, and the information points corresponding to the reference detector calculated when the radar rotates to the adjacent angle of the reference angle. Thus, reference points from different sources are obtained, increasing the reliability and diversity of the reference points, and there is a spatial and temporal association with the point to be determined.

[0108] In specific implementation, the preset range can be set according to the sampling frequency, rotation speed of the radar and actual requirements. Among them, the span of the preset range (that is, the difference between the maximum boundary value and the minimum boundary value in the range) is at least greater than the angle rotated by a single sampling, so as to ensure that at least one adjacent angle can be determined according to the range. For example, if the rotation speed of the radar is ω and the sampling frequency is f, the span of the range is at least ω / f.

[0109] In specific implementations, the determination method of adjacent detectors can be set according to the actual application scenario and requirements. According to different determination methods, the adjacent detectors can be the detectors located on a specified side of the reference detector, or the detectors located on both sides of the reference detector. Correspondingly, the determination method of adjacent angles can also be set according to the actual application scenario and requirements. According to different determination methods, the adjacent angle can be the angle detected immediately before the reference angle, and the angular difference between the adjacent angle and the reference angle is represented by a negative value; it can also be the angle detected immediately after the reference angle, and the angular difference between the adjacent angle and the reference angle is represented by a positive value; it can also be the angles corresponding to several detections before and after the reference angle respectively. Furthermore, the corresponding range can be set according to different determination methods. This specification does not make specific restrictions on the determination methods of adjacent detectors and adjacent angles.

[0110] S4. Based on the relevance between the point to be determined and the reference point, determine whether the point to be determined is a noise point.

[0111] Since the reference point includes information points of adjacent angles (horizontal dimension) and adjacent detectors (vertical dimension), obtaining reference points that are associated with the point to be determined in dimensions such as space and time sequence can improve the reliability and diversity of the reference points. Furthermore, when identifying noise points based on the relevance between the point to be determined and the reference point, the accuracy of the identification result can be improved, the misjudgment probability can be reduced, and thus noise points can be effectively identified, ensuring the data quality of the point cloud.

[0112] To enable those skilled in the art to more clearly understand the technical effects of the technical solutions provided in this specification, the following continues to use the above Figure 1a and Figure 1b corresponding realizable applications as examples for specific description.

[0113] According to Figure 1a and Figure 1b and related descriptions, the useful information point a X collected by the detector TC 5 and other information points of the detector TC X (including the useful information points a 1 ~a 4 and the useful information points a 6 ~a 9 ) have weak relevance, resulting in the misjudgment of the existing technical solution that the useful information point a 5 is a noise point.

[0114] Assume that the detectors with a vertical angular resolution difference from the detector TC X are: the detector TC X-1 and the detector TC X+1 . Among them, the detector TCX-1 At time t 5 receive the detection signal reflected by the external utility pole GA, and obtain the useful information point c 1 . The detector TC X+1 At time t 1 ~t 4 receive the detection signals reflected by the external wall WA respectively, and obtain the useful information points d 1 ~d 4 , and obtain the useful information point d 5 at time t 5 .

[0115] Combined with Figure 1a and Figure 1b it can be known that the useful information point c 1 , the useful information point a 1 and the useful information point d 5 correspond to Figure 1b a column of useful information points along the axis of the utility pole GA in

[0116] For the technical solution provided in this specification, if the useful information point a X collected by the detector TC 5 at time t 5 is used as the point to be determined, its reference points can include: the useful information point a 4 , the useful information point a 6 , the useful information point c 1 and the useful information point d 5 .

[0117] Since the correlation between the useful information point a X of the detector TC 5 and other information points of the detector TC X (including the useful information point a 1 ~a 4 and the useful information point a 6 ~a 9 ) is relatively low, therefore, after selecting the useful information point a 5 as the point to be determined, the correlation between the point to be determined a 5 and the reference point a X from the reference detector TC 4 and the reference point a 6 is relatively low.

[0118] However, since the field of view corresponding to the same angle of the detector can be approximately considered the same, the adjacent detectors TC X-1 and the adjacent detector TC X+1 respectively at the reference angle (i.e., the detector TC X when collecting the useful information point a 5The reference point c collected at the rotated angle at that time 1 and the reference point d 5 both have a high correlation with the point a to be determined 5 which improves the overall correlation between the point a to be determined 5 and its reference point, thereby effectively reducing the misjudgment rate of useful information points.

[0119] In a specific implementation, to further improve the accuracy of the noise point recognition result, as Figure 3 shown, it is a flowchart of another data processing method for radar provided by an embodiment of this specification. Compared with Figure 2 the example shown, the difference is that before the step S4), the following steps may further be included:

[0120] SA, find the detection information of the adjacent angle corresponding to the reference angle and the adjacent detector corresponding to the reference detector, and use the obtained information points as reference points.

[0121] Thus, the reference points may further include: the information points corresponding to the adjacent detectors calculated when the radar rotates to the adjacent angle of the reference angle. Thereby, the source of the reference points is further enriched, and the reliability and diversity of the reference points are improved.

[0122] In a specific implementation, for the detection information of the reference detector, the adjacent angle is an angle whose angle difference from the reference angle belongs to a first range; for the detection information of the adjacent detector, the adjacent angle is an angle whose angle difference from the reference angle belongs to a second range. Wherein, the span of the first range may be equal to the span of the second range or may not be equal to the span of the second range.

[0123] In a specific implementation, since the farther the distance between the detectors, the greater the field of view angle difference between the information points collected by the two, and the weaker the correlation between the corresponding information points. Therefore, to effectively reduce the data volume, the span of the first range may be greater than the span of the second range.

[0124] In an optional example, the span difference between the first range and the second range is positively correlated with the field of view angle difference between the detectors.

[0125] In a specific implementation, the adjacent detectors can be the detectors on one side of the reference detector or the detectors on both sides of the reference detector. When the adjacent detectors include multiple detectors located on one side of the reference detector in the arrangement position, that is, when there are multiple adjacent detectors on one side of the reference detector, the second range can include multiple angular ranges corresponding to the multiple detectors respectively. Further, since the farther the distance between the detectors, the greater the field-of-view angle difference between the information points collected by the two, and the weaker the correlation between the corresponding information points, the span of the angular range is negatively correlated with the distance from the adjacent detector to the reference detector in the arrangement position. Thus, the data volume can be effectively reduced.

[0126] To facilitate the understanding and implementation of the process of obtaining the reference point by those skilled in the art, the following will be described through specific embodiments and accompanying drawings.

[0127] It should be noted that, for the convenience of description and intuitive understanding, some of the embodiment drawings in this specification adopt a two-dimensional plane perspective, and the distances between each information point are equal. However, this does not specifically limit the dimension and actual distribution of the information points. In actual applications, the information points can be three-dimensional data, and each information point can be non-uniformly distributed in three-dimensional space, that is, the distances between each information point may not be equal. The embodiments of this specification do not make specific limitations on this.

[0128] In a specific implementation, as Figure 4 shown, it is a schematic diagram of the distribution of radar point clouds. In this example, the radar can include (m + n + 1) vertically arranged detectors, that is, detector DE 1 to detector DE m+n+1 , and the radar rotates horizontally along the vertical axis, and its horizontal angular resolution is θ, that is, the span between two consecutive angles is θ. Among them, the detection information of each detector includes (p + q + 1) information points, corresponding to angles AG 1 to angle AG p+q+1 respectively. The span between any two consecutive angles among angles AG 1 to angle AG p+q+1 is θ. The (p + q + 1) × (m + n + 1) information points form the radar point cloud.

[0129] Select information point A 1 (that is, Figure 4 the hatched information point in the figure) from the point cloud as the point to be determined. Take the detector DE 1 corresponding to the point to be determined A m+1 as the reference detector, and take the angle AG 1 corresponding to the point to be determined A p+1 as the reference angle.

[0130] In some realizable embodiments, in combination with Figure 2 and Figure 4 , among the detection information of detector DE 1 to detector DE m+n+1 , search for the adjacent angles corresponding to the reference angle AG p+1 and the detection information corresponding to the reference detector DE m+1 , the detection information corresponding to the adjacent detector of the reference detector DE m+1 and the adjacent angles corresponding to the reference angle AG p+1 .

[0131] According to different determination methods of adjacent detectors and different determination methods of adjacent angles, detection information in different situations will be obtained, and thus different reference points will be obtained. The following gives examples by listing several situations. It should be noted that these examples do not limit other situations that may exist in the actual application of the embodiments of this specification.

[0132] In some cases, the determination method of the adjacent detector can be: a detector located above the reference detector.

[0133] For the adjacent angle of the reference detector, the determination method can be: the span of the first angle range is not less than the horizontal angular resolution θ and not greater than twice the horizontal angular resolution 2θ. For example, the first angle range can be (-2θ, 0), so that according to the first angle range, the angle corresponding to the previous detection of the reference detector at the reference angle can be obtained.

[0134] For the adjacent angle of the adjacent detector, the determination method can be: the span of the second angle range is less than twice the horizontal angular resolution 2θ. For example, the second angle range can be (-θ, θ). Thus, the adjacent angle of the adjacent detector cannot be obtained according to the second angle range. In other words, there is no need to search for the detection information of the adjacent detector at the adjacent angle.

[0135] According to the above determination method of the adjacent detector and the determination method of the adjacent angle of the reference detector, a schematic diagram of reference point selection corresponding to Figure 5 as shown in Figure 4 can be obtained, where:

[0136] The adjacent angles corresponding to the reference angle AG p+1 and the detection information corresponding to the reference detector DE m+1 may include: the detection information of detector DE m+1 at angle AG p (i.e., the information point B Figure 5 in 11 ).

[0137] corresponding to the reference detector DE m+1 of the adjacent detectors and corresponding to the reference angle AG p+1 The detection information may include: the detector DE m at the angle AG p+1 The detection information (i.e., Figure 5 the information point C in 11 ).

[0138] Then, the information point B 11 and the information point C 11 are used as the reference points for the point A to be determined 1 .

[0139] In some other cases, the determination method for adjacent detectors may be: one detector on each side of the reference detector

[0140] For the adjacent angles of the reference detector, the determination method may be: the span of the first angle range is not less than twice the horizontal angle resolution 2θ and not greater than three times the horizontal angle resolution 3θ. For example, the first angle range may be [-θ, θ], so that according to the first angle range, the angles corresponding to the previous detection and the next detection of the reference detector at the reference angle can be obtained

[0141] For the adjacent angles of the adjacent detectors, the determination method may be: the span of the second angle range is less than twice the horizontal angle resolution 2θ. For example, the second angle range may be (-θ, θ). Thus, the adjacent angles of the adjacent detectors cannot be obtained according to the second angle range. In other words, there is no need to find the detection information of the adjacent detectors at the adjacent angles

[0142] According to the determination method of the adjacent detectors and the determination method of the adjacent angles of the reference detector above, another schematic diagram of the reference point selection corresponding to Figure 6 can be obtained as shown Figure 4 , where

[0143] corresponding to the adjacent angle of the reference angle AG p+1 and corresponding to the detection information of the reference detector DE m+1 The detection information may include: the detector DE m+1 at the angle AG p The detection information (i.e., Figure 6 the information point B in 11 ) and the detection information at the angle AG p+2 The detection information (i.e., Figure 6 the information point B in 21 ).

[0144] corresponding to the adjacent detector of the reference detector DE m+1 and corresponding to the reference angle AGp+1 The detection information may include: detector DE m at angle AG p+1 of the detection information (i.e., Figure 6 information point C in 11 ) and detector DE m+2 at angle AG p+1 of the detection information (i.e., Figure 6 information point C in 21 ).

[0145] Then, information point B 11 , information point B 21 , information point C 11 and information point C 21 are used as reference points for the point A to be determined 1 .

[0146] In some other cases, the determination method for adjacent detectors may be: j detectors on each side of the reference detector. Where j is an integer greater than 1.

[0147] For the adjacent angles of the reference detector, the determination method may be: the span of the first angle range is not less than 2i times the horizontal angle resolution 2i×θ and its span is not greater than 3i times the horizontal angle resolution 3i×θ, where i is an integer greater than 1. For example, the first angle range may be [-i×θ, i×θ], so that according to the first angle range, the i angles corresponding to the first i detections and the last i detections of the reference detector at the reference angle can be obtained.

[0148] For the adjacent angles of the adjacent detectors, the determination method may be: the span of the second angle range is less than twice the horizontal angle resolution 2θ. For example, the second angle range may be (-θ, θ). Thus, the adjacent angles of the adjacent detectors cannot be obtained according to the second angle range. In other words, there is no need to find the detection information of the adjacent detectors at the adjacent angles.

[0149] According to the determination method of the adjacent detectors and the determination method of the adjacent angles of the reference detector, the following detection information can be obtained:

[0150] Corresponding to the adjacent angles of the reference angle AG p+1 and corresponding to the reference detector DE m+1 The detection information may include: detector DE m+1 at angle AG p of the detection information (i.e., Figure 7 information point B in 11 ) to the detection information at angle AG p-i+1 ( Figure 7 not shown in), and detector DE m+1At the angle AG p+2 of the detection information (i.e., Figure 7 the information point B in 21 ) to the detection information at the angle AG p+i+1 ( Figure 7 not shown in).

[0151] The adjacent detector corresponding to the reference detector DE m+1 and the detection information corresponding to the reference angle AG p+1 may include: the detection information of the detector DE m at the angle AG p+1 (i.e., Figure 7 the information point C in 11 ) to the detection information of the detector DE m-j+1 at the angle AG p+1 , and the detection information of the detector DE m+2 at the angle AG p+1 (i.e., Figure 7 the information point C in 21 ) to the detection information of the detector DE m+j+1 at the angle AG p+1 . And the obtained information points are used as reference points.

[0152] For example, when i = 2 and j = 2, as Figure 7 shown, it is another schematic diagram of the selection of the reference point corresponding to Figure 4 . According to the determination method of the adjacent detector and the determination method of the adjacent angle of the reference detector above, the information points B 11 and B 12 , the information points B 21 and B 22 , the information points C 11 and C 12 , and the information points C 21 and the information point C 22 can be obtained and used as the reference points of the point A 1 to be determined.

[0153] In some other implementable embodiments, combining Figure 3 and Figure 4 , in the detection information of the detector DE 1 to the detector DE m+n+1 , find the detection information corresponding to the adjacent angle of the reference angle AG p+1 and corresponding to the reference detector DE m+1 , the detection information corresponding to the adjacent detector of the reference detector DE m+1 and corresponding to the reference angle AG p+1 , and the detection information corresponding to the reference angle AG p+1adjacent angles and corresponding to the reference detector DE m+1 the detection information of adjacent detectors.

[0154] According to different determination methods of adjacent detectors and different determination methods of adjacent angles, detection information in different situations will be obtained, and thus different reference points will be obtained. The following gives examples by listing several situations. It should be noted that these examples do not limit other situations that may exist in the actual application of the embodiments of this specification.

[0155] In some cases, the determination method of adjacent detectors can be: one detector on each side of the reference detector.

[0156] For the adjacent angles of the reference detector, the determination method can be: the span of the first angle range is not less than twice the horizontal angular resolution 2θ and not greater than three times the horizontal angular resolution 3θ. For example, the first angle range can be [-θ, θ], so that according to the first angle range, the angles corresponding to the previous detection and the next detection of the reference detector corresponding to the reference angle can be obtained.

[0157] For the adjacent angles of the adjacent detectors, the determination method can be: the span of the second angle range is not less than twice the horizontal angular resolution 2θ and not greater than three times the horizontal angular resolution 3θ. For example, the second angle range can be [-θ, θ], so that according to the second angle range, the angles corresponding to the previous detection and the next detection of the adjacent detector corresponding to the reference angle can be obtained.

[0158] According to the above determination method of adjacent detectors and the determination method of adjacent angles of the reference detector, another schematic diagram of reference point selection corresponding to Figure 8 can be obtained as shown in Figure 4 , where:

[0159] corresponding to the reference angle AG p+1 adjacent angles and corresponding to the reference detector DE m+1 the detection information may include: detector DE m+1 at angle AG p the detection information (i.e., Figure 9 the information point B in 11 ) and at angle AG p+2 the detection information (i.e., Figure 3 the information point B in 21 ).

[0160] corresponding to the reference detector DE m+1 adjacent detectors and corresponding to the reference angle AG p+1 the detection information may include: detector DE mAt the angle AG p+1 The detection information (i.e., Figure 3 The information point C in 11 ) and the detector DE m+2 At the angle AG p+1 The detection information (i.e., Figure 3 The information point C in 21 )

[0161] Corresponding to the adjacent angles of the reference angle AG p+1 And corresponding to the adjacent detectors of the reference detector DE m+1 The detection information may include: The detector DE m At the angle AG p The detection information (i.e., Figure 8 The information point D in 11 ) and at the angle AG p+2 The detection information (i.e., Figure 8 The information point D in 21 )), the detector DE m+2 At the angle AG p The detection information (i.e., Figure 8 The information point D in 31 ) and at the angle AG p+2 The detection information (i.e., Figure 8 The information point D in 41 ).

[0162] Then, take the information point B 11 , the information point B 21 , the information point C 11 , the information point C 21 , the information point D 11 , the information point D 21 , the information point D 31 And the information point D 41 As the reference points for the point A 1 to be determined.

[0163] In some other cases, the determination method for adjacent detectors can be: One detector on each side of the reference detector.

[0164] For the adjacent angles of the reference detector, the determination method can be: The span of the first angle range is not less than four times the horizontal angular resolution 4θ and its span is not greater than five times the horizontal angular resolution 5θ. For example, the first angle range can be [-2θ, 2θ], so that the angles corresponding to the first two detections and the last two detections of the reference detector corresponding to the reference angle can be obtained according to the first angle range.

[0165] For the adjacent angles of the adjacent detectors, the determination method can be: the span of the second angle range is not less than twice the horizontal angle resolution 2θ and not greater than three times the horizontal angle resolution 3θ. For example, the second angle range can be [-θ, θ], so that the angles corresponding to the previous detection and the subsequent detection of the reference angle corresponding to the adjacent detectors can be obtained according to the second angle range.

[0166] According to the determination method of the adjacent detectors and the determination method of the adjacent angles of the reference detector, the following can be obtained Figure 9 Another schematic diagram of the reference point selection corresponding to Figure 4 is shown, where:

[0167] The adjacent angle corresponding to the reference angle AG p+1 and the detection information corresponding to the reference detector DE m+1 can include: the detection information of the detector DE m+1 at the angle AG p-1 (i.e., the information point B Figure 9 in 12 ), the detection information at the angle AG p (i.e., the information point B Figure 9 in 11 ), the detection information at the angle AG p+2 (i.e., the information point B Figure 9 in 21 ) and the detection information at the angle AG p+3 (i.e., the information point B Figure 9 in 22 ).

[0168] The adjacent detector corresponding to the reference detector DE m+1 and the detection information corresponding to the reference angle AG p+1 can include: the detection information of the detector DE m at the angle AG p+1 (i.e., the information point C Figure 9 in 11 ) and the detection information of the detector DE m+2 at the angle AG p+1 (i.e., the information point C Figure 9 in 21 ).

[0169] The adjacent angle corresponding to the reference angle AG p+1 and the detection information of the adjacent detector corresponding to the reference detector DE m+1 can include: the detection information of the detector DE m at the angle AG p (i.e., the information point D Figure 9 in 11) and the detection information at angle AG p+2 (i.e., Figure 9 the information point D in 21 ), the detector DE m+2 the detection information at angle AG p (i.e., Figure 9 the information point D in 31 ), and the detection information at angle AG p+2 (i.e., Figure 9 the information point D in 41 ).

[0170] Then, the information points B 11 , information point B 12 , information point B 21 , information point B 22 , information point C 11 , information point C 21 , information point D 11 , information point D 21 , information point D 31 and information point D 41 are used as the reference points for the point A to be determined 1 .

[0171] In some other cases, the determination method for adjacent detectors can be: two detectors on each side of the reference detector.

[0172] For the adjacent angles of the reference detector, the determination method can be: the span of the first angle range is not less than four times the horizontal angular resolution 4θ and its span is not greater than five times the horizontal angular resolution 5θ. For example, the first angle range can be [-2θ, 2θ], so that the angles corresponding to the first two detections and the last two detections of the reference detector corresponding to the reference angle can be obtained according to the first angle range.

[0173] For the adjacent detectors located on the upper and lower sides of the reference detector and without other detectors in between, the determination method for their adjacent angles can be: the span of the second angle range is not less than twice the horizontal angular resolution 2θ and its span is not greater than three times the horizontal angular resolution 3θ. For example, the angle range of this adjacent detector can be [-θ, θ], so that the angles corresponding to the first detection and the last detection of the adjacent detector corresponding to the reference angle can be obtained according to the angle range of the adjacent detector.

[0174] For adjacent detectors located on the upper and lower sides of the reference detector with one detector in between, the determination method of the adjacent angle can be: the span of the angle range of the adjacent detector is less than twice the horizontal angular resolution 2θ. For example, the angle range of the adjacent detector can be (-θ, θ). Thus, no adjacent angle can be obtained based on the angle range of the adjacent detector. In other words, there is no need to search for the detection information of the adjacent detector at the adjacent angle.

[0175] According to the determination method of the adjacent detector and the determination method of the adjacent angle of the reference detector described above, the following can be obtained Figure 10 as shown in another schematic diagram of reference point selection corresponding to Figure 4 where:

[0176] The detection information corresponding to the adjacent angle of the reference angle AG p+1 and corresponding to the reference detector DE m+1 can include: the detection information of the detector DE m+1 at the angle AG p-1 (i.e., the information point B Figure 10 in 12 ), the detection information of the detector DE p at the angle AG Figure 10 (i.e., the information point B 11 in p+2 ), the detection information of the detector DE Figure 10 at the angle AG 21 (i.e., the information point B p+3 in Figure 10 ), and the detection information of the detector DE 22 at the angle AG

[0177] The detection information corresponding to the adjacent detector of the reference detector DE m+1 and corresponding to the reference angle AG p+1 can include: the detection information of the detector DE m at the angle AG p+1 (i.e., the information point C Figure 10 in 11 ), the detection information of the detector DE m-1 at the angle AG p+1 (i.e., the information point C Figure 10 in 12 ), the detection information of the detector DE m+2 at the angle AG p+1 (i.e., the information point C Figure 10 in 21 ), the detection information of the detector DE m+3 at the angle AG p+1 (i.e., the information point C Figure 10 in 22 ).

[0178] corresponding to the reference angle AG p+1 for adjacent angles and corresponding to the reference detector DE m+1 the detection information of adjacent detectors may include: detector DE m at the angle AG p the detection information (i.e., Figure 9 information point D in 11 ) and at the angle AG p+2 the detection information (i.e., Figure 9 information point D in 21 ), detector DE m+2 at the angle AG p the detection information (i.e., Figure 9 information point D in 31 ) and at the angle AG p+2 the detection information (i.e., Figure 9 information point D in 41 ).

[0179] Then, take information point B 11 , information point B 12 , information point B 21 , information point B 22 , information point C 11 , information point C 12 , information point C 21 , information point C 22 , information point D 11 , information point D 21 , information point D 31 and information point D 41 as the reference points for the point A to be determined 1 .

[0180] It can be understood that in practical applications, since the detection information of the detector includes information points and blank results, there may be a situation where it is impossible to obtain reference points even if detection information meeting the conditions is found. The embodiments of this specification do not make specific restrictions on this.

[0181] For example, as Figure 11 shown, it is a schematic diagram of the distribution of another radar point cloud. Compared with the example shown in Figure 10 , the difference is that: the detection information of detector DE m+1 at the angle AG p is a blank result. Thus, even if the detection information of detector DE m+1 at the angle AG p meets the adjacent conditions, since its detection information is a blank result, there is no information point available as a reference point.

[0182] It can be understood that the number of adjacent detectors and the number of adjacent angles can be set according to the actual application scenario and requirements. For example, they can be set according to parameters such as the number of detection signals, rotation speed, and resolution of the radar, so as to avoid obtaining too few reference points, resulting in missing points, and avoid obtaining too many reference points, which will lead to a large amount of data, occupying too much computing resources and affecting the data processing efficiency. The embodiments of this specification do not make specific limitations on the number of adjacent detectors and the number of adjacent angles.

[0183] In specific implementations, in some cases, if the number of adjacent detectors is small, there may still be misjudgment situations. For example, if the resolution of the radar is low, the interval between each information point in the point cloud may be relatively large, resulting in relatively weak correlation between information points.

[0184] Specifically, refer to Figure 12 , which shows a schematic diagram of the arrangement of information points collected by some detectors of another radar corresponding to Figure 1a the field of view in three-dimensional space.

[0185] Combined with reference to Figure 1a and Figure 12 , when corresponding to the radar field of view shown in Figure 1a , due to the low resolution of the radar, and the angle span of the utility pole GA in the horizontal direction is small, and the two detectors TC X adjacent to TC X-1 and TC X+1 did not detect the detection signal reflected by the utility pole GA, only received the detection signal reflected by the wall WA, and obtained information points d 1 ~d 4 . At this time, if the determination method of adjacent detectors is: one detector on each side of the reference detector, that is, when setting two adjacent detectors, point a 5 may still be misjudged as a noise point.

[0186] To reduce the misjudgment probability, the number of adjacent detectors can be increased, thereby expanding the source range and diversity of reference points. For example, the determination method of adjacent detectors is: two detectors on each side of the reference detector, that is, when setting four adjacent detectors, when the useful information point a 5 is used as the point to be determined, the source range of its reference points is expanded, the reference points are more diverse, and the overall correlation between the point to be determined and the reference points can be strengthened. As shown in Figure 12 , the useful information point e X-2 of the adjacent detector TC in the radar and the useful information point f 1 of the adjacent detector TC X-2 can be obtained and used as reference points. Since the reference points e 1 and f 1 and f 1has a strong correlation with the point a to be determined 5 so as to reduce the probability that the useful information point a 5 is misjudged as a noise point.

[0187] In a specific implementation, based on the correlation degree between the point to be determined and the reference point, determining whether the point to be determined is a noise point may include: determining whether the point to be determined is a noise point according to the point distance between the point to be determined and the reference point.

[0188] Specifically, as Figure 13 shown, it is a flowchart of a noise point judgment method, which may include the following steps:

[0189] S4-11, respectively calculate the distances between each of the reference points and the point to be determined to obtain a point distance data set.

[0190] As mentioned above, according to the detector corresponding to the information point and the radar rotation data, the three-dimensional angle corresponding to the information point can be obtained. In a specific implementation, according to the time of the detection signal of the transmitter and the time when the corresponding detector receives the signal, using the Time Of Flight (TOF) algorithm, the distance from the obstacle to the radar can be calculated. Therefore, the information point may further include: obstacle distance data to represent the obstacle distance corresponding to the information point.

[0191] S4-12, determine a point distance threshold.

[0192] S4-13, judge whether there is point distance data lower than the point distance threshold in the point distance data set. If so, determine that the point to be determined is not a noise point; otherwise, determine that the point to be determined is a noise point.

[0193] For example, the point distance data set is {dis 1 , dis 2 , dis 3 , dis 4 , dis 5}, where dis k represents the point distance data between the kth reference point and the point to be determined, where k = 1, 2, 3, 4, 5. Based on the corresponding relationship between the obstacle distance and the point distance threshold, the point distance threshold is obtained as r. Thus, judge the size relationship between each point distance data dis k and the point distance threshold r. If there is at least one point distance data dis k lower than the point distance threshold r, then determine that the point to be determined is not a noise point; otherwise, determine that the point to be determined is a noise point.

[0194] Thus, by using the distances between each reference point in the reference point set and the point to be determined to characterize the correlation between the point to be determined and the reference point set, the reliability of the correlation can be enhanced, the problem of lost points caused by misjudgment can be effectively reduced, the accuracy of the noise point recognition result can be improved, and the data quality of the point cloud can be ensured.

[0195] Based on this, the point distance threshold of the point to be determined can be determined according to the obstacle distance corresponding to the point to be determined. Specifically, as Figure 14 shown, it is a flowchart of a method for determining the point distance threshold, which may include the following steps:

[0196] 4-121) Obtain the obstacle distance corresponding to the point to be determined.

[0197] 4-122) Determine the point distance threshold of the point to be determined based on the corresponding relationship between the obstacle distance and the point distance threshold.

[0198] Thus, the point distance threshold can change according to the obstacle distance corresponding to the point to be determined, and the adjustability of the point distance threshold can improve the accuracy of the noise point judgment result.

[0199] In specific implementation, due to the perspective relationship of near large and far small, for objects of the same size, the farther the distance from the radar, the fewer information points the radar can collect, and the weaker the correlation between information points. Based on this, the point distance threshold is positively correlated with the obstacle distance corresponding to the point to be determined, that is, the greater the obstacle distance corresponding to the point to be determined, the greater the point distance threshold. Thus, a more accurate point distance threshold can be obtained to ensure the noise point recognition result.

[0200] For example, as Figure 15 shown, it is a schematic diagram of the corresponding relationship between a point distance threshold and an obstacle distance. In Figure 15 it, the point distance threshold and the obstacle distance are in a linear relationship, which is convenient for quickly determining the point distance threshold.

[0201] In practical applications, considering both the noise filtering effect and the point loss rate, the point distance threshold and the obstacle distance can be in a non-linear relationship. According to the point cloud requirements, in the key information part, a curve segment where the point distance threshold changes rapidly with the obstacle distance is set to reduce the probability of misjudging useful information, thereby reducing the point loss rate and reducing the key information in the lost point cloud; in the non-key information part, a curve segment where the point distance threshold changes slowly with the obstacle distance is set to improve the noise filtering effect.

[0202] For example, as Figure 16 shown, it is another schematic diagram of the corresponding relationship between a point distance threshold and an obstacle distance. In Figure 16Among them, the dot pitch threshold and the obstacle distance have a non-linear relationship. It may include: a curve segment where the dot pitch threshold changes rapidly with the obstacle distance, that is, the first part ① and the third part ③, and a curve segment where the dot pitch threshold changes slowly with the obstacle distance, that is, the second part ②.

[0203] It can be seen from Figure 16 that the key information part corresponds to the obstacles existing at close range and the obstacles existing at long range. Ensuring the key information of these two parts is beneficial for subsequent data processing and analysis; while the non-key information part corresponds to the obstacles existing at a moderate distance. Since the distance from the radar is moderate, the radar can obtain more information at the moderate distance range, and the correlation between information points is relatively strong. Therefore, the noise filtering effect can be improved by reducing the change amplitude of the dot pitch threshold.

[0204] In a specific implementation, the number of lines of the radar (the number of detection signals transmitted in a single acquisition) and the rotation speed of the radar affect the resolution of the radar. Among them, when other parameters remain unchanged, the more the number of lines of the radar, the higher the resolution of the radar; correspondingly, when other parameters remain unchanged, the higher the rotation speed of the radar, the lower the resolution of the radar. Further, the lower the resolution of the radar, the sparser the information points collected by the radar. At this time, if the dot pitch threshold is set too small, misjudgment is likely to occur.

[0205] Based on this, in order to obtain a more accurate dot pitch threshold and improve the accuracy of the noise point recognition result, the dot pitch threshold can be negatively correlated with the resolution of the radar, that is, the lower the resolution of the radar, the larger the dot pitch threshold. In a specific implementation, there may be a point loss probability for each frame of point cloud of the radar, that is, the point loss rate of the radar. If the dot pitch threshold is too small, it will increase the point loss rate of the radar. In order to balance the point loss rate of the radar and the noise point recognition effect, the point loss rate of the radar can be calculated. When the point loss rate of the radar is higher than the preset point loss rate threshold, the dot pitch threshold is increased. Among them, the way to increase the dot pitch threshold can be to modify the corresponding relationship curve between the dot pitch threshold and the obstacle distance, or to obtain the dot pitch threshold according to the corresponding relationship curve between the dot pitch threshold and the obstacle distance, and then superimpose a correction amount to obtain the increased dot pitch threshold.

[0206] It can be understood that the data processing method provided in the embodiments of this specification can be executed first for noise point recognition, and then the radar point loss rate can be calculated; or the radar point loss rate can be calculated first, and then the data processing method provided in the embodiments of this specification can be executed. This specification does not make a specific limitation on the execution order of the two.

[0207] In a specific implementation, based on the correlation between the point to be determined and the reference point, determining whether the point to be determined is a noise point may include: determining whether the point to be determined is a noise point according to the number of points of the reference point. Specifically, count the number of points of the reference point, and determine whether the number of points of the reference point is lower than a preset point number threshold. If so, determine that the point to be determined is a noise point; otherwise, determine that the point to be determined is not a noise point. Wherein, the point number threshold is a positive integer greater than 2.

[0208] Thus, by using the number of points of the reference point to characterize the correlation between the point to be determined and the reference point set, the calculation amount can be reduced, and the noise point recognition efficiency can be improved while ensuring the accuracy of the noise point recognition result.

[0209] As can be seen from the above description, in a specific implementation, the number of lines of the radar and the rotation speed of the radar affect the resolution of the radar. The lower the resolution of the radar, the sparser the information points collected by the radar, that is, the smaller the number of information points. At this time, if the value of the point number threshold is too large, the noise filtering effect will become poor; if the value of the point number threshold is too small, the misjudgment probability will increase, resulting in point loss.

[0210] Based on this, in order to obtain a more accurate point number threshold and improve the accuracy of the noise point recognition result, the point number threshold can be positively correlated with the resolution of the radar, that is, the lower the resolution of the radar, the smaller the point number threshold.

[0211] In a specific implementation, if the point number threshold is too small, the point loss rate of the radar will increase. In order to balance the point loss rate of the radar and the noise point recognition effect, the point loss rate of the radar can be calculated. When the point loss rate of the radar is higher than the preset point loss rate threshold, increase the point number threshold.

[0212] It can be understood that the data processing method provided in the embodiments of this specification can be executed first for noise point recognition, and then the radar point loss rate can be calculated; or the radar point loss rate can be calculated first, and then the data processing method provided in the embodiments of this specification can be executed. This specification does not make a specific limitation on the execution order of the two.

[0213] In a specific implementation, according to the actual application scenario and requirements, at least one of the number of points and the point distance can be used for noise point judgment.

[0214] Further, in order to balance the processing efficiency and the accuracy of the noise point recognition result, the noise point judgment based on the number of points can be performed first, and when the point number correlation is satisfied, the noise point judgment based on the point distance can be performed.

[0215] Since the amount of data for noise judgment based on the number of points is less than that for noise judgment based on the point distance, if in the noise judgment based on the number of points, the point to be judged has been judged as noise, then the noise judgment based on the point distance can be skipped, thus saving computing resources and improving the processing efficiency.

[0216] Since the accuracy of noise judgment based on the number of points is lower than that of noise judgment based on the point distance, if in the noise judgment based on the number of points, the point to be judged is not judged as noise, then the noise judgment based on the point distance can be carried out again, thereby improving the accuracy of the noise recognition result.

[0217] In a specific implementation, as Figure 2 shown, after step S4, the data processing method may further include step S5. If the point to be determined is judged as noise, then the point to be determined is deleted.

[0218] Moreover, the noise judgment operation can also be continued for the information points that have not undergone noise judgment. Specifically, referring to Figure 2 or Figure 3 , the data processing method may further include the following steps:

[0219] S6, determine whether there are information points that have not undergone noise judgment. If so, continue to step S7;

[0220] S7, select the point to be determined from the information points that have not undergone noise judgment, and continue to execute the above step S3.

[0221] In this way, noise recognition operations can be performed on all information points in the point cloud, effectively filtering out noise and improving data quality.

[0222] In a specific implementation, to reduce the amount of calculation and improve the processing efficiency, some information points can be selected from the point cloud for noise recognition according to the actual situation. For example, if there are areas with relatively dense information points and areas with relatively sparse information points in the point cloud, noise recognition can be performed on the areas with relatively dense information points; or, if a detector has obtained information points at multiple adjacent angles, among these information points, corresponding information points can be selected at set intervals as the points to be determined. The embodiments of this specification do not specifically limit the selection method of some information points.

[0223] It should be noted that Figure 2 and Figure 3The processing operations are only given in the case where there are information points for which noise judgment has not been performed. In actual applications, there may also be cases where there are no information points for which noise judgment has not been performed, and corresponding processing operations are performed in such cases. For example, when there are no information points for which noise judgment has not been performed, this process can be ended, and after waiting for the radar to obtain more point clouds, the data processing method described in the embodiments of this specification can be performed again. This specification does not make specific restrictions on this.

[0224] It can be understood that the above describes multiple embodiment solutions provided in this specification. The optional methods introduced in each embodiment solution can be combined and cross-referenced with each other without conflict, thereby extending multiple possible embodiment solutions, all of which can be considered as the embodiment solutions disclosed and made public in this specification.

[0225] This specification also provides a data processing device corresponding to the above data processing method, which will be introduced in detail below with reference to the accompanying drawings through specific embodiments. It should be noted that the data processing device described below can be considered as a functional module required to implement the data processing method provided in this specification; the content of the data processing device described below can be correspondingly referred to the content of the data processing method described above.

[0226] In a specific implementation, as Figure 17 shown, it is a structural block diagram of a data processing device for a radar in an embodiment of this specification. The data processing device M1 may include:

[0227] A data storage unit M11, adapted to cache the point cloud of the radar, where the point cloud includes information points obtained by multiple detectors at multiple angles;

[0228] A noise recognition unit M12, adapted to select a point to be determined from the point cloud; taking the detector corresponding to the point to be determined as a reference detector, taking the angle corresponding to the point to be determined as a reference angle, finding the detection information corresponding to the adjacent angle corresponding to the reference angle and corresponding to the reference detector, and the detection information corresponding to the adjacent detector corresponding to the reference detector and corresponding to the reference angle, and taking the obtained information points as reference points; and, based on the correlation between the point to be determined and the reference points, determining whether the point to be determined is a noise point.

[0229] Adopting the above solution, since the reference points include information points of adjacent angles (lateral dimension) and adjacent detectors (longitudinal dimension), reference points that are associated with the point to be determined in dimensions such as space and time sequence are obtained, which can improve the reliability and diversity of the reference points. Furthermore, when identifying noise points based on the correlation between the point to be determined and the reference points, the accuracy of the identification result can be improved, the misjudgment probability can be reduced, and thus noise points can be effectively identified, ensuring the data quality of the point cloud.

[0230] In a specific implementation, continue to refer to Figure 17 , the noise recognition unit M12 is also adapted to find the detection information of adjacent detectors corresponding to the adjacent angles of the reference angle and corresponding to the reference detector, and use the obtained information points as reference points. For details, please refer to the above relevant content and will not be elaborated here.

[0231] Thus, the reference points may further include: the information points corresponding to the adjacent detectors calculated when the radar rotates to the adjacent angles of the reference angle, thereby further enriching the sources of the reference points and improving the reliability and diversity of the reference points.

[0232] In a specific implementation, the adjacent detectors are one or more detectors adjacent to the reference detector in the arrangement position. The adjacent angles are one or more angles whose angular difference from the reference angle belongs to a preset range. For details, please refer to the above relevant content and will not be elaborated here.

[0233] Thus, reference points from different sources are obtained, increasing the reliability and diversity of the reference points, and there are spatial and temporal correlations between the reference points and the points to be determined.

[0234] In a specific implementation, for the detection information of the reference detector, the adjacent angles are the angles whose angular difference from the reference angle belongs to a first range; for the detection information of the adjacent detectors, the adjacent angles are the angles whose angular difference from the reference angle belongs to a second range.

[0235] Optionally, since the farther the distance between the detectors, the greater the field-of-view angle difference between the information points collected by the two, and the weaker the correlation between the corresponding information points. Therefore, in order to effectively reduce the data volume, the span of the first range can be greater than the span of the second range. Further, the span difference between the first range and the second range is positively correlated with the field-of-view angle difference between the detectors.

[0236] In a specific implementation, when the adjacent detectors include multiple detectors located on one side of the reference detector in the arrangement position, the second range may include multiple angular ranges corresponding to the multiple detectors respectively, and the span of the angular range is negatively correlated with the distance from the adjacent detector to the reference detector in the arrangement position.

[0237] In a specific implementation, the correlation degree between the point to be determined and the reference point includes: the point distance between the point to be determined and the reference point. For details, please refer to the above relevant content and will not be elaborated here.

[0238] Thus, the correlation between the to-be-determined point and the reference point set is characterized by the distances between the reference points in the reference point set and the to-be-determined point, which can enhance the reliability of the correlation, effectively reduce the problem of lost points caused by misjudgment, improve the accuracy of the noise point recognition result, and ensure the data quality of the point cloud.

[0239] In a specific implementation, the correlation between the to-be-determined point and the reference point includes: the number of points of the reference point. For specific reference, please refer to the above relevant content and will not be elaborated here.

[0240] Thus, by characterizing the correlation between the to-be-determined point and the reference point set through the number of points of the reference point, the calculation amount can be reduced, and the noise point recognition efficiency can be improved while ensuring the accuracy of the noise point recognition result.

[0241] In a specific implementation, continue to refer to Figure 17 , the noise point recognition unit M12 is further adapted to delete the to-be-determined points determined as noise points. Further, the noise point recognition unit M12 is further adapted to select the to-be-determined points from the information points that have not been subjected to noise judgment and perform noise judgment.

[0242] The embodiment of this specification also provides a radar, as Figure 18 shown. In the embodiment of this specification, the radar may include a plurality of transmitters L1 to LX, a plurality of detectors D1 to DY, and a data processing device M2, where:

[0243] There is a corresponding relationship between the transmitters L1 to LX and the detectors D1 to DY; the transmitters L1 to LX are adapted to emit detection signals at multiple angles; the detectors D1 to DY are adapted to collect echo signals at multiple angles;

[0244] The data processing device M2 is adapted to perform data processing based on the detection signals of the transmitters L1 to LX and the echo signals of the corresponding detectors D1 to DY to obtain corresponding information points, and execute the method described in any of the above embodiments.

[0245] It should be noted that when the detector collects echo signals, the actually collected echo signals may include true echo signals and interference signals. This specification does not make specific restrictions on this.

[0246] In summary, by finding the adjacent angles corresponding to the reference angle, the detection information corresponding to the reference detector, the adjacent detectors corresponding to the reference detector, and the detection information corresponding to the reference angle, reference points that are associated with the point to be determined in terms of space, time sequence, and other dimensions are obtained, improving the reliability and diversity of the reference points. Furthermore, when identifying noise points based on the correlation between the point to be determined and the reference points, the accuracy of the identification result can be improved, the misjudgment probability can be reduced, and thus noise points can be effectively identified, ensuring the data quality of the point cloud.

[0247] In a specific implementation, the data processing device may include a memory and a processor. The memory may store one or more computer-executable instructions, and the processor may call the one or more computer-executable instructions to execute the steps of the method provided in the embodiments of this specification.

[0248] In a specific implementation, the radar may be a lidar, a millimeter-wave radar, or the like.

[0249] The embodiments of this specification also provide a computer-readable storage medium, on which computer instructions are stored. When the computer instructions run, they can execute the steps of the method in any of the above embodiments of this specification. Among them, the computer-readable storage medium may be various suitable readable storage media such as an optical disc, a mechanical hard disk, a solid-state drive, etc. The instructions stored on the computer-readable storage medium execute the method described in any of the above embodiments, and specific reference may be made to the above embodiments, which will not be elaborated here.

[0250] The computer-readable storage medium may include, for example, any suitable type of memory unit, memory device, memory item, memory medium, storage device, storage item, storage medium, and / or storage unit. For example, a memory, a removable or non-removable medium, an erasable or non-erasable medium, a writable or rewritable medium, a digital or analog medium, a hard disk, a floppy disk, a compact disc read-only memory (CD-ROM), a recordable compact disc (CD-R), a rewritable compact disc (CD-RW), an optical disc, a magnetic medium, a magneto-optical medium, a removable memory card or disk, various types of digital versatile discs (DVDs), magnetic tapes, cassette tapes, etc.

[0251] Computer instructions may include any suitable type of code implemented by using any suitable high-level, low-level, object-oriented, visual, compiled, and / or interpreted programming language. For example, source code, compiled code, interpreted code, executable code, static code, dynamic code, encrypted code, etc.

[0252] It should be noted that the "one embodiment" or "embodiment" referred to in this specification means a specific feature, structure or characteristic that may be included in at least one implementation manner of this specification. And in the description of this specification, terms such as "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with terms such as "first" and "second" may explicitly or implicitly include one or more of such features. Moreover, terms such as "first" and "second" are used to distinguish similar objects and do not necessarily have to be used to describe a specific order or indicate importance. It can be understood that such terms can be interchanged under appropriate circumstances so that the embodiments of this specification described herein can be implemented in an order other than those illustrated or described herein.

[0253] Although the embodiments of this specification are disclosed as above, this specification is not limited thereto. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of this specification. Therefore, the protection scope of this specification should be subject to the scope defined by the claims.

Claims

1. A data processing method for a radar, characterized in that, the data processing method includes the following steps: 1) Obtain the point cloud of the radar, where the point cloud includes: information points obtained by multiple detectors at multiple angles; 2) Select the point to be determined from the point cloud; 3) Take the detector corresponding to the point to be determined as the reference detector, and take the angle corresponding to the point to be determined as the reference angle, find the detection information corresponding to the adjacent angle of the reference angle and corresponding to the reference detector, and the detection information corresponding to the adjacent detector of the reference detector and corresponding to the reference angle, and use the information points included in the obtained detection information as reference points; 4) Based on the correlation between the point to be determined and the reference points, determine whether the point to be determined is a noise point.

2. The data processing method for a radar according to claim 1, characterized in that, before the step 4), the following steps are further included: A) Find the detection information of the adjacent detector corresponding to the adjacent angle of the reference angle and corresponding to the reference detector, and use the information points included in the obtained detection information as reference points.

3. The data processing method for a radar according to claim 1 or 2, characterized in that, the adjacent detector is one or more detectors adjacent to the reference detector in the arrangement position; the adjacent angle is one or more angles whose angle difference from the reference angle belongs to a preset range.

4. The data processing method for a radar according to claim 1 or 2, characterized in that, for the detection information of the reference detector, the adjacent angle is an angle whose angle difference from the reference angle belongs to the first range; for the detection information of the adjacent detector, the adjacent angle is an angle whose angle difference from the reference angle belongs to the second range; wherein, the span of the first range is greater than the span of the second range.

5. The data processing method for a radar according to claim 4, characterized in that, the span difference between the first range and the second range is positively correlated with the field of view angle difference between the reference detector and the adjacent detector.

6. The data processing method for a radar according to claim 4, characterized in that, when the adjacent detector includes multiple detectors arranged on one side of the reference detector in the arrangement position, the second range includes multiple angle ranges corresponding to the multiple detectors respectively, and the span of the angle range is negatively correlated with the distance from the adjacent detector to the reference detector in the arrangement position.

7. The data processing method for a radar according to claim 1 or 2, characterized in that, the step 4) includes: 4-1) Determine whether the point to be determined is a noise point according to the point distance between the point to be determined and the reference points.

8. The data processing method for a radar according to claim 7, characterized in that, the step 4-1) includes the following steps: 4-11) Calculate the distances between each reference point and the point to be determined respectively to obtain a point distance data set; 4-12) Determine the point distance threshold; 4-13) Determine whether there is point distance data lower than the point distance threshold in the point distance dataset. If so, determine that the point to be determined is not a noise point; otherwise, determine that the point to be determined is a noise point.

9. The data processing method for radar according to claim 8, wherein, the step 4-12) includes the following steps: 4-121) Obtain the obstacle distance corresponding to the point to be determined; 4-122) Determine the point distance threshold of the point to be determined based on the corresponding relationship between the obstacle distance and the point distance threshold.

10. The data processing method for radar according to claim 8, wherein, the point distance threshold is positively correlated with the obstacle distance corresponding to the point to be determined, and the point distance threshold is negatively correlated with the resolution of the radar.

11. The data processing method for radar according to claim 8, wherein, when the point loss rate of the radar is higher than the preset point loss rate threshold, increase the point distance threshold.

12. The data processing method for radar according to claim 1 or 2, wherein, the step 4) includes: 4-2) Determine whether the point to be determined is a noise point according to the number of points of the reference point.

13. The data processing method for radar according to claim 12, wherein, the step 4-2) includes the following steps: 4-21) Determine whether the number of points of the reference point is lower than the preset number threshold. If so, determine that the point to be determined is a noise point; otherwise, determine that the point to be determined is not a noise point.

14. The data processing method for radar according to claim 13, wherein, the number threshold is positively correlated with the resolution of the radar.

15. The data processing method for radar according to claim 13, wherein, when the point loss rate of the radar is higher than the preset point loss rate threshold, increase the number threshold.

16. The data processing method for radar according to claim 1 or 2, wherein, further includes: 5) If the point to be determined is determined to be a noise point, delete the point to be determined.

17. The data processing method for radar according to claim 1 or 2, wherein, further includes: 6) Determine whether there is an information point for which noise point judgment has not been performed. If so, select the point to be determined from the information points for which noise point judgment has not been performed, and continue to execute the step 3).

18. A data processing device for radar, wherein, comprises: a data storage unit, adapted to cache the point cloud of the radar, the point cloud including information points obtained by a plurality of detectors at a plurality of angles; a noise point identification unit, adapted to select a point to be determined from the point cloud; Taking the detector corresponding to the point to be determined as a reference detector and the angle corresponding to the point to be determined as a reference angle, find the detection information corresponding to the adjacent angle of the reference angle and corresponding to the reference detector, and the detection information corresponding to the adjacent detector of the reference detector and corresponding to the reference angle, and use the information points included in the obtained detection information as reference points; And, based on the relevance between the point to be determined and the reference point, determine whether the point to be determined is a noise point.

19. The data processing device for radar according to claim 18, wherein, the noise point recognition unit is further adapted to find the detection information of adjacent angles corresponding to the reference angle and adjacent detectors corresponding to the reference detector, and use the information points included in the obtained detection information as reference points.

20. The data processing device for radar according to claim 18 or 19, wherein, the adjacent detectors are one or more detectors adjacent to the reference detector in the arrangement position; the adjacent angles are one or more angles whose angular difference from the reference angle belongs to a preset range.

21. The data processing device for radar according to claim 18 or 19, wherein, for the detection information of the reference detector, the adjacent angles are the angles whose angular difference from the reference angle belongs to the first range; for the detection information of the adjacent detectors, the adjacent angles are the angles whose angular difference from the reference angle belongs to the second range; wherein, the span of the first range is greater than the span of the second range.

22. The data processing device for radar according to claim 18 or 19, wherein, the relevance between the point to be determined and the reference point includes: the point distance between the point to be determined and the reference point.

23. The data processing device for radar according to claim 18 or 19, wherein, the relevance between the point to be determined and the reference point includes: the number of points of the reference point.

24. The data processing device for radar according to claim 18 or 19, wherein, the noise point recognition unit is further adapted to delete the points to be determined judged as noise points.

25. The data processing device for radar according to claim 18 or 19, wherein, the noise point recognition unit is further adapted to select the points to be determined from the information points that have not been judged for noise points and perform noise point judgment.

26. A radar, wherein, comprises a plurality of transmitters, a plurality of detectors and a data processing device, wherein: there is a corresponding relationship between the transmitter and the detector; the transmitter is adapted to emit detection signals at multiple angles; the detector is adapted to collect echo signals at multiple angles; the data processing device is adapted to perform data processing according to the detection signal of the transmitter and the echo signal of the corresponding detector to obtain corresponding information points, and execute the method according to any one of claims 1-17.

27. The radar according to claim 26, wherein, the radar is a lidar.

28. A computer-readable storage medium, on which computer instructions are stored, wherein, when the computer instructions run, they execute the steps of the method according to any one of claims 1 to 17.

Images