Method for detecting integrity of laser radar point cloud in service area

Abstract

The invention discloses a laser radar point cloud integrity detection method in a service area. The method comprises the following steps: obtaining continuous multi-frame radar point cloud data and carrying out preprocessing; according to the maximum sensing radius of the radar, a plurality of cylindrical detection areas with the radar as the center and with the radiuses increasing in sequence are constructed; projecting the point cloud data falling in each detection area to the surface of the corresponding detection area along the emission direction to obtain a projection point set corresponding to each detection area; and calculating the dispersion degree of each projection point set on the surface of the corresponding detection area so as to carry out point cloud integrity judgment. The method has the beneficial effects that the point cloud is projected to the surfaces of the detection areas with different radiuses, so that regularized reconstruction of disordered and sparse three-dimensional point cloud data is realized. According to the invention, quantitative statistics can be carried out on the number of point clouds in different distance ranges, and the reason for the point cloud missing problem can be judged in an auxiliary manner, so that the precision and robustness of defect identification are improved.

Description

Technical Field

[0001] This application relates to the field of intelligent transportation technology, and in particular to a method for detecting the integrity of lidar point clouds in a service area. Background Art

[0002] Highway service areas, as key nodes in the modern highway transportation network, play important roles in vehicle diversion, short-term parking, energy supply, and emergency management. With the continuous growth of traffic volume, the service area scenario has become increasingly complex. Behaviors such as high vehicle density, frequent entry and exit, and irregular parking have put forward higher requirements for the real-time perception and refined management of service areas.

[0003] Fixed-deployed lidar (LiDAR), due to its advantages such as all-weather, high-precision, and anti-interference, has gradually become an important sensor for vehicle monitoring and environmental modeling in service area scenarios. Compared with traditional vision solutions, lidar can provide dense and high-precision three-dimensional point cloud data, facilitating target detection and behavior tracking without relying on lighting conditions.

[0004] However, in the actual deployment and application process, it is limited by the following multiple factors: limitations of the radar's own technical parameters (such as vertical field of view, rotation frequency), occlusion during the point cloud acquisition process (such as large vehicles blocking small vehicles), possible delays or packet losses during data transmission, and incomplete point cloud fusion between multiple devices.

[0005] Based on the above limited factors, it may lead to situations such as spatial discontinuity, regional sparsity, or even structural missing in the point cloud data obtained from a single lidar at a certain moment. This incomplete point cloud structure seriously affects the effect of subsequent intelligent perception tasks, such as vehicle recognition errors, trajectory breaks, and behavior misjudgments, which will ultimately weaken the operation efficiency and safety guarantee ability of the service area in the direction of digitization and intelligence. Summary of the Invention

[0006] One of the purposes of this application is to provide a method for detecting the integrity of lidar point clouds in a service area that can solve at least one defect in the above background art.

[0007] To achieve at least one of the above purposes, the technical solution adopted in this application is: a method for detecting the integrity of lidar point clouds in a service area, including the following steps: S100: Obtain continuous multi-frame radar point cloud data and perform preprocessing; S200: According to the maximum sensing radius of the radar, construct multiple cylindrical detection areas centered on the radar with gradually increasing radii; S300: Project the point cloud data falling within each detection area along the emission direction onto the surface of the corresponding detection area to obtain a set of projection points corresponding to each detection area; S400: Calculate the degree of dispersion of each set of projection points on the surface of the corresponding detection area to determine the integrity of the point cloud.

[0008] Among them, the degree of dispersion of the surface of the detection area is preferably represented by the continuous point density in the circumferential direction of the detection area, and the integrity of the point cloud is judged according to the ratio of the calculated actual point density to the corresponding theoretical point density.

[0009] Preferably, in step S200, the model expression of the detection area in the coordinate system is: ; where x0 and y0 respectively represent the horizontal and vertical coordinates of the radar center point; r represents the distance from the surface of the detection area to the radar center point; θ represents the angle between the radar installation attitude and the horizontal direction; x, y, and z respectively represent the three-dimensional point coordinates on the surface of the detection area.

[0010] Preferably, in step S200, the interval distance between adjacent detection areas is equal.

[0011] Preferably, the number of detection areas is set to five, and the distances of each detection area from the radar center point are 20%, 40%, 60%, 80%, and 100% of the maximum sensing radius of the radar respectively.

[0012] Preferably, step S300 includes the following specific steps: S310: Calculate the distance from each point in all the point cloud data to the radar center point, and judge the detection area corresponding to each point in the point cloud data according to the calculated distance; S320: Construct a direction vector pointing from the radar center point to each point and parameterize it; S330: Based on the perpendicular intersection of the direction vector and the corresponding detection area, substitute the parameterized direction vector into the expression of the detection area for solution to obtain the coordinate position of each point in all the point cloud data projected onto the surface of the corresponding detection area; S340: Summarize the projected point coordinates on the surface of each detection area to obtain the corresponding set of projection points.

[0013] Preferably, step S400 includes the following specific process: S410: Project each detection area along the axis to form a corresponding plurality of projection circles, and simultaneously map the projection points on the surface of the detection area to the corresponding projection circles; S420: Divide each projection circle into a plurality of equal-area sectors along the circumferential direction to obtain the total number of points corresponding to each detection area and the number of points corresponding to each sector; S430: Calculate the theoretical point density of each detection area according to the total number of points, and calculate the actual point density corresponding to each sector according to the number of points corresponding to the sector; S440: Determine the point cloud integrity based on the ratio of the actual point density of each sector to the corresponding theoretical point density.

[0014] Preferably, in step S400, the ratio of the actual point density to the corresponding theoretical point density is used as the first completeness of the sector; the multiple sectors corresponding to the detection area are divided into multiple groups, each group includes at least one sector; the average completeness of each group of sectors in multiple consecutive frames is calculated; if the calculated average completeness is less than the set first threshold, and the number of corresponding points of the sectors in multiple consecutive frames is less than the minimum value of the theoretical number of points, it is determined that there is a high missing area in the group of sectors.

[0015] Preferably, when judging the integrity of the point cloud in step S400, it is also necessary to judge the point cloud distribution in each sector. The specific judgment process is: divide the multiple sectors corresponding to the detection area into multiple test areas, each test area includes multiple continuous sectors, and there is sector overlap between adjacent test areas; calculate the spacing between adjacent points in the test area and the corresponding standard deviation; if the calculated standard deviation is greater than the set second threshold, or the maximum point spacing is greater than the set third threshold, it is determined that the corresponding sector has a fault.

[0016] Preferably, step S400 also includes the following process: S450: Calculate the theoretical number of points in each sector according to the angular resolution and the number of horizontal beams of the radar, and use the ratio of the actual number of points in the sector to the theoretical number of points as the second integrity; S460: Calculating the second completeness of the sectors corresponding to the same phase angle in each detection area in order from small to large area radius; S470: If the second completeness of a sector corresponding to a certain phase angle does not continue to increase, it indicates that an abnormality exists in the sector corresponding to the phase angle.

[0017] Preferably, the specific judgment process for the abnormal situation in step S470 is: calculate the difference between the distances between two consecutive points in the abnormal sector; if the calculated difference is less than the set fourth threshold, it means that the point cloud of the sector is continuous and complete; if the calculated difference is greater than the set fourth threshold, and the number of points in the sector is less than the minimum value of the theoretical number of points, the sector is determined to be a missing or abnormal area; perform difference judgment on the missing or abnormal area for multiple consecutive frames, if the same missing or abnormal area always exists, it is determined that the data transmission or radar sensor in the area is abnormal.

[0018] Compared with the prior art, the beneficial effects of this application are: By projecting the point cloud onto the surfaces of detection regions with different radii, the regularization reconstruction of disordered and sparse three-dimensional point cloud data is achieved. It can not only quantitatively count the number of point clouds within different distance ranges, but also assist in determining whether the point cloud missing problem is caused by field of view occlusion, or due to abnormalities in the radar sensor itself or the data transmission process, thereby improving the accuracy and robustness of defect recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a schematic diagram of the overall working process of the present application.

[0020] Figure 2 This is a schematic diagram of the structure of the detection region constructed in the present application.

[0021] Figure 3 This is a schematic diagram of the structure of multiple detection regions projected along the axis in the present application.

[0022] Figure 4 This is a schematic diagram of the structure of the detection region divided into sectors in the present application.

[0023] Figure 5 This is a schematic diagram of the structure of the point cloud projected onto the sectors in the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] Next, in combination with the specific embodiments, the present application will be further described. It should be noted that in the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms should not be understood as necessarily referring to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In addition, those skilled in the art can combine and combine the different embodiments or examples described in this specification.

[0025] In the description of the present application, it should be noted that for orientation terms, if there are terms such as "center", "lateral", "longitudinal", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", etc., indicating the orientation and position relationship is based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and should not be construed as limiting the specific protection scope of the present application.

[0026] It should be noted that the terms "first", "second", etc. in the description and claims of this application are used to distinguish similar objects and do not necessarily describe a specific order or sequence.

[0027] In this application, unless otherwise clearly defined and limited, terms such as "installed", "connected", "linked", "fixed", etc. should be understood in a broad sense. For example, it can be a connection, a detachable connection, or integrated; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two components or the interaction relationship between two components. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0028] In this application, unless otherwise clearly defined and limited, the first feature being "above" or "below" the second feature may include the first and second features being in direct contact, or may include the first and second features not being in direct contact but in contact through other features therebetween. Moreover, the first feature being "above", "over" and "on" the second feature includes the first feature being directly above and obliquely above the second feature, or merely indicating that the first feature has a higher horizontal height than the second feature. The first feature being "below", "under" and "beneath" the second feature includes the first feature being directly below and obliquely below the second feature, or merely indicating that the first feature has a lower horizontal height than the second feature.

[0029] The terms "comprising" and "having" in the description and claims of this application, and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0030] One preferred embodiment of this application, as Figure 1 and Figure 2 shown, a method for detecting the integrity of lidar point clouds in a service area, comprising the following steps: S100: Obtain continuous multi-frame radar point cloud data and perform preprocessing.

[0031] S200: According to the maximum sensing radius of the radar, construct a plurality of cylindrical detection areas centered on the radar with sequentially increasing radii.

[0032] S300: Project the point cloud data falling within each detection area along the emission direction onto the surface of the corresponding detection area to obtain a projection point set corresponding to each detection area.

[0033] S400: Calculate the dispersion degree of each projected point set on the surface of the corresponding detection area to judge the integrity of the point cloud.

[0034] Among them, the dispersion degree of the surface of the detection area is preferably represented by the continuous point density in the circumferential direction of the detection area, and the integrity of the point cloud is judged according to the ratio of the calculated actual point density to the corresponding theoretical point density.

[0035] It can be understood that the point cloud data collected by the radar is disordered in space, and it is difficult or impossible to directly judge the integrity. The technical solution of this application can realize the regular reconstruction of disordered and sparse three-dimensional point cloud data by projecting the point cloud onto the surface of cylindrical detection areas with different radii. It can not only quantitatively count the number of point clouds in different distance ranges, but also assist in judging the specific principle of the point cloud missing problem, thereby improving the accuracy and robustness of defect recognition. For the convenience of understanding the technical solution of this application, each step will be described in detail below.

[0036] In this embodiment, for step S100, the radar point cloud data of continuous T frames can be obtained, and the obtained point cloud data can be expressed as: , t = 1, 2, ……, T; where represents the i-th point in the point cloud data, represents the spatial coordinate corresponding to the i-th point.

[0037] It can be understood that after obtaining the required point cloud data, in order to ensure the accuracy of subsequent integrity evaluation, the point cloud data needs to be preprocessed. The preprocessing of the point cloud data includes necessary denoising, time series synchronization and registration operations to eliminate the influence of accidental interference and data fluctuations. Moreover, multiple frames of collected data can more comprehensively reflect the stable output ability and coverage range of the radar, providing a basis for subsequent integrity evaluation.

[0038] In this embodiment, according to the working principle of the lidar, the point cloud scanned by the radar shows a form in which the point cloud is denser when the distance from the center point of the radar is smaller and sparser when the distance is larger. As Figure 2 and Figure 3 shown, in order to reduce the point distortion caused by the radar sensing range, according to the theoretical maximum sensing radius R of the radar, several sensing ratios are selected to construct corresponding cylindrical regions, that is, detection regions, for hierarchical evaluation of the spatial coverage of the point cloud.

[0039] Specifically, each detection area is centered on the radar projection point O, with a radius r = kR and a height equal to the installation height H of the radar; where k represents the sensing ratio of each detection area. Then the expression of the constructed detection area projected on the ground is: (x - x0)2 +(y - y0) 2 = r 2 ; where x0 and y0 respectively represent the abscissa and ordinate of the radar center point on the ground, and x and y represent the abscissa and ordinate of the point where the detection area is projected onto the ground.

[0040] It can be understood that the above expression of the detection area projection is for the scenario where the axis of the detection area is aligned along the z-axis; however, in the actual use process, the installation scenario of the radar is not necessarily horizontal, and it may be an uneven ground scenario or the radar is installed obliquely; then, in order to accurately represent the detection area formed by the radar in any posture, the model expression of the detection area in the earth coordinate system can be constructed in step S200 as: ; where r represents the distance from the surface of the detection area to the radar center point, that is, the radius of the detection area; θ represents the angle between the radar installation posture and the horizontal direction, and the value range of θ is [0, 2π); x, y, and z respectively represent the three-dimensional point coordinates on the surface of the detection area.

[0041] It should be noted that when constructing the detection area, in order to ensure the integrity analysis of the point cloud of each detection area, the interval distance between adjacent detection areas can be set to be equal in step S200. For example, the distance between the surfaces of adjacent detection areas is 0.1R or 0.2R, etc., and specifically can be set according to the actual needs of those skilled in the art; in this embodiment, 0.2R is taken as an example, that is, as Figure 3 shown, the number of detection areas is set to five, and the distances from the radar center point to each detection area are 20%, 40%, 60%, 80%, and 100% of the maximum sensing radius R of the radar respectively; for the convenience of subsequent description, these five detection areas can be defined as detection area a, detection area b, detection area c, detection area d, and detection area e respectively.

[0042] In this embodiment, step S300 includes the following specific steps: S310: Calculate the distance from each point in all the point cloud data to the radar center point, and determine the detection area corresponding to each point in the point cloud data according to the calculated distance.

[0043] S320: Construct the direction vector from the radar center point to each point and parameterize it.

[0044] S330: Based on the perpendicular intersection of the direction vector and the corresponding detection area, substitute the parameterized direction vector into the expression of the detection area to solve, and obtain the coordinate position of each point in all the point cloud data projected onto the surface of the corresponding detection area.

[0045] S340: Aggregate the projected point coordinates on the surface of each detection area to obtain the corresponding set of projected points.

[0046] It can be understood that the main purpose of step S300 is to project all the point cloud data located within different detection areas onto the surfaces of the corresponding detection areas. That is, calculate the distance D from each point p(x p , y p , z p ) in the point cloud data to the radar center point O(x0, y0, z0). According to the calculated distance D, it can be determined which detection area the point is in. For example, if the distance D from a certain point to the radar center point is D = 0.5R, it can be obtained that the point is located within detection areas a and b. More specifically, the points within detection area a also fall within detection areas b, c, d, and e, the points within detection area b also fall within detection areas c, d, and e, the points within detection area c also fall within detection areas d and e, and the points within detection area d also fall within detection area e.

[0047] After completing the regional division of the point cloud data, in order to project the spatial points onto the surfaces of the detection areas, the ray method can be used for projection; specifically, a direction vector pointing from the radar center point to point p can be constructed . After that, the perpendicular intersection point of the ray and the detection area can be calculated. For the convenience of calculation, the direction vector used to represent the ray can be parameterized as: .

[0048] Among them, R λ represents the parametric form of the ray, and λ represents the parameter.

[0049] Substitute R λ into the model expression of the detection area, and a quadratic equation of one variable about the parameter λ can be obtained: .

[0050] Solve the above equation to obtain the positive real root closest to λ. According to the obtained positive real root, the intersection position of the ray and the detection area can be calculated, so as to determine the coordinate position of the intersection point in the detection area model, and further obtain the set of projected points of all points within each radius range to the surface of the corresponding detection area for subsequent density and continuity analysis.

[0051] In this embodiment, as Figure 4 shown, step S400 includes the following specific processes: S410: Project each detection area along the axis to form corresponding multiple projection circles, and simultaneously map the projected points on the surface of the detection area to the corresponding projection circles.

[0052] S420: Divide each projected circle into multiple sectors with equal areas along the circumferential direction to obtain the total number of points corresponding to each detection area and the number of points corresponding to each sector.

[0053] S430: Calculate the theoretical point density of each detection area based on the total number of points, and at the same time calculate the actual point density corresponding to each sector based on the number of points corresponding to the sector.

[0054] S440: Judge the integrity of the point cloud based on the ratio of the actual point density to the corresponding theoretical point density of each sector.

[0055] It can be understood that the integrity of radar point cloud data can mainly be regarded as the uniformity of the distribution of the point cloud in the radar radiation direction, or the degree of discreteness of the distribution; then, by partitioning the detection area in the circumferential direction and calculating the point cloud density of each area respectively, the integrity of the radar point cloud data can be obtained. Since the detection area is cylindrical as a whole, the calculation of the spatial point density can be simplified to represent it as the planar point density of the spatial points mapped to the axial projected circle of the detection area; of course, it can also be simplified to represent it as the line point density of the spatial points mapped to the axial projected circle of the detection area; specifically, it can be selected according to the actual needs of those skilled in the art. In this embodiment, the planar point density is preferably used to judge the integrity of the point cloud. By dividing the sensing space of the radar into multiple fan-shaped areas, the overall point cloud data is structured. Combining with the subsequent continuity analysis method, the integrity of the point cloud in each sector can be quickly and effectively evaluated, and the precise positioning of the point cloud missing area can be achieved. This method has the advantages of high calculation efficiency and clear spatial positioning, providing a reliable basis for subsequent tasks such as point cloud completion and anomaly detection.

[0056] Specifically, it can be set that the axial projected circle of the detection area is divided into M sectors, then the area A of each sector i =πr 2 / M. For the number of points N corresponding to each sector i which can be obtained through the projected point set, the actual point density ρ corresponding to each sector is i =N i / A i . For the theoretical point density ρ ref , the total number of points N corresponding to the entire detection area can be obtained through the projected point set. Then, the required theoretical point density ρ 2 can be obtained by the ratio of the total number of points N to the projected circle area πr ref of the detection area. Finally, the ratio of the actual point density to the theoretical point density can be used as an evaluation index of integrity, which can be represented by the first integrity C i , that is, C i =ρ i / ρ refThe value of M can be selected by those skilled in the art according to actual needs. For example Figure 4 As shown, the value of M is 36, that is, the central angle corresponding to a single sector is 10°.

[0057] It should be noted that in order to ensure the accuracy of the judgment of the integrity of sector point clouds, the average integrity of a single sector over multiple consecutive frames can be calculated, that is, the average value of the first integrity calculated for each frame is obtained. After obtaining the average integrity of each sector in the detected area division, the calculated average integrity can be compared with the set first threshold C thr The specific value of the first threshold C thr can be selected according to actual needs. For example, it can be taken as 0.8 - 0.9. Then when the value of the average integrity is greater than the first threshold C thr and the number of points corresponding to the sector in multiple consecutive frames is less than the minimum value N of the theoretical number of points min , it can be determined that the sector is a high - missing area; otherwise, it can be initially judged that the point cloud integrity of the sector is normal.

[0058] It should be noted that during the actual measurement of radar point cloud data, due to some refraction and reflection errors, the points may have slight jitters. That is to say, even if the scene does not change at all, the radar data of two consecutive frames is very likely to be different. Therefore, only calculating the average integrity of a single sector over multiple consecutive frames may lead to inconsistent calculation processes before and after due to the existence of point jitters, thereby affecting the judgment of point cloud integrity.

[0059] In this embodiment, the influence of point jitters is suppressed by calculating the average integrity jointly using multiple sectors; specifically, the multiple sectors corresponding to the detected area are divided into multiple groups, and each group includes at least two sectors; the average integrity of each group of sectors within multiple consecutive frames is calculated; if the calculated average integrity is less than the set first threshold and the number of points corresponding to the sectors in multiple consecutive frames is less than the minimum value of the theoretical number of points, it is determined that there is a high - missing area in this group of sectors.

[0060] It should be noted that the specific number of each group of sectors can be selected according to actual needs. For example, three sectors or four sectors can be selected for joint calculation. It should be noted that the more the number of sectors in each group, the better the suppression effect on the point jitter during the multi-frame measurement, but the worse the judgment effect on the high-missing area. Therefore, the number of sectors in each group in this embodiment should not be set too much, and it is preferably to use three sectors for joint calculation. Generally speaking, for three consecutive sectors ABC, if the average integrity of a single sector is calculated for sector B, then during the measurement of multiple consecutive frames, some points may jump to sector A or sector C, that is, the points at both ends of sector B will jump to adjacent sectors. If the three sectors ABC are used for joint calculation, then the jitter of the points in sector B must be within sector A or sector C, and only the points at the ends of sector A and sector C in the joint area will jump to other sectors. That is, the number of points that jump out of the range in the multi-sector joint is the same as that in the single sector, but the total number of points used for the average integrity calculation increases by several times, which effectively suppresses the influence of the jumping points on the average integrity calculation result.

[0061] It should be noted that the first integrity is only a representation of the overall point cloud distribution or dispersion degree of the detection area, and the unit of its analysis is the sector; and there may be a large difference between adjacent points in the sector, which cannot be reflected by the first integrity. Therefore, after the integrity of the point cloud is preliminarily judged by the first integrity, it is also necessary to judge the point cloud spacing in each sector.

[0062] In this embodiment, as Figure 5 shown, the adjacent point spacing d and the corresponding standard deviation σ D of the point cloud in each sector can be calculated. For the calculation of the adjacent point spacing d, the phase angle between adjacent points can be calculated through the position coordinates of the corresponding points in the sector, and then the spacing d between adjacent points can be calculated according to the arc length calculation formula. For the standard deviation σ D it can be calculated by the following formula: .

[0063] Among them, n represents the number of adjacent points in the sector, d j represents the jth adjacent point spacing, represents the average spacing of all points in the sector.

[0064] After the calculation of the adjacent point spacing d j and the corresponding standard deviation σ D is completed, the obtained adjacent point spacing d j and the corresponding standard deviation σ D can be respectively compared with the set second threshold d max and the third threshold δ thrIf there is one or more adjacent points in the sector with a distance greater than the second threshold d max , or the entire standard deviation is greater than the third threshold δ thr , then combined with the preliminary judgment result of the first completeness, it can be determined that there is a fault in this sector.

[0065] It should be known that for the second threshold d max and the third threshold δ thr The specific value of needs to be determined according to the radar sensor's field of view and resolution. By introducing the distance change between adjacent points as the core criterion for completeness, this method is more suitable for dealing with spatial sparseness problems caused by occlusion, reflection, etc. compared to traditional point density indicators. On the unified projection surface, the geometric relationship between points is clearer, which helps to quantitatively analyze the continuity of the point cloud structure, thereby improving the stability and spatial adaptability of the completeness evaluation.

[0066] It is understandable that the calculation of the above adjacent point spacing and standard deviation is for a single sector, and there may be abnormal endpoint spacing for adjacent sectors. For example, there are continuous points a1, a2, a3, a4 and a5 in a certain sector, and continuous points b1, b2, b3, b4 and b5 in another adjacent sector. In the former sector, there are adjacent point spacings a1-a2, a2-a3, a3-a4 and a4-a5, in the latter sector, there are adjacent point spacings b1-b2, b2-b3, b3-b4 and b4-b5, and there is an endpoint spacing a5-b1 between the two sectors; for the above point spacings, there may be a situation where the point spacings a1-a2, a2-a3, a3-a4, a4-a5, b1-b2, b2-b3, b3-b4 and b4-b5 are normal, and the point spacing a5-b1 is abnormal. In this case, if only the distance between adjacent points in a single sector is analyzed, it will not be discovered, which will lead to inaccurate judgment of the integrity of the radar point cloud.

[0067] Therefore, in the technical solution of the present application, the multiple sectors corresponding to the detection area can be divided into multiple areas to be tested, each area to be tested includes multiple continuous sectors, and there are overlapped sectors between adjacent areas to be tested. That is, the technical solution of the present application adopts a multi-sector sliding window strategy for the calculation of the distance between adjacent points; in layman's terms, a sliding window can be set for multiple sectors of a single detection area, and the sliding window contains at least two sectors; then the distance between adjacent points and the standard deviation can be calculated for the points corresponding to all sectors in the sliding window; each time the sliding window calculation is completed, the sliding window can be moved by one sector until the calculation of the distance between adjacent points and the standard deviation of all sectors in the detection area is completed.

[0068] It should be noted that the above point cloud integrity judgment is only for a single detection area, that is, for multiple local parts within the radiation range of the radar point cloud. To further improve the accuracy of the point cloud integrity judgment, the point cloud integrity can be further judged by combining all detection areas, which will be introduced in detail below.

[0069] In this embodiment, step S400 further includes the following process: S450: Calculate the theoretical number of points N in each sector according to the angular resolution Δθ of the radar and the number of horizontal beams L theory , and use the ratio of the actual number of points N i of the sector to the theoretical number of points N theory as the second integrity Q i .

[0070] Specifically, N theory = θ sector · L / Δθ, Q i = N i / N theory ; where θ sector represents the central angle corresponding to the sector.

[0071] S460: Calculate the second integrity of the sectors corresponding to the same phase angle in each detection area in ascending order of the regional radius.

[0072] S470: If the second integrity of the sector corresponding to a certain phase angle does not increase continuously, it indicates that there is an abnormality in the sector corresponding to this phase angle.

[0073] It can be understood that through the second integrity evaluation index, the ratio of the actual number of points obtained by the radar in a single area to the total number of points that should be obtained can be obtained. Theoretically, this ratio should increase with the increase of the sector radius r; by comprehensively using the second integrity evaluation index and the previous density evaluation index, the model accuracy of calculating the integrity can be effectively enhanced.

[0074] It should be known that if the second integrity of a certain phase angle sector does not reach the set threshold, such as 80%, within a small radius range, such as 0.2R or 0.4R, and there is no obvious increase in this value subsequently, there are two possibilities for this area. Possibility one: This area is a relatively empty space, so the points at a relatively far distance are not reflected back to the radar and the data is not collected; Possibility two: There is a problem with the radar or data transmission in this area. For these two possibilities, specific judgments can be made through the aforementioned adjacent point spacing judgment method.

[0075] Specifically, the specific judgment process for the abnormal situation in step S470 is: the difference d i+1 - d iPerform calculations. If the calculated difference d i+1 -d i is less than the set fourth threshold ε, it indicates that the sector point cloud is continuous and complete, corresponding to possibility one. If the calculated difference d i+1 -d i is greater than the set fourth threshold ε, and the number of points N in the sector i is less than the minimum value N of the theoretical number of points min , it is determined that this sector is a missing or abnormal area; for the detected missing or abnormal area, it is necessary to check multiple frames of data for multiple periods of time; if the same missing or abnormal area exists in multiple frames of data for multiple periods of time, it is determined that there is a problem with the data transmission or radar sensor in this area, corresponding to possibility two.

[0076] The above describes the basic principle, main features and advantages of the present application. Those skilled in the art should understand that the present application is not limited by the above embodiments. What is described in the above embodiments and the specification is only the principle of the present application. Without departing from the spirit and scope of the present application, the present application will have various changes and improvements, and these changes and improvements all fall within the scope of the present application claimed. The scope of protection required by the present application is defined by the appended claims and their equivalents.

Claims

1. A method for detecting the integrity of lidar point clouds in a service area, characterized in that, It includes the following steps: S100: Obtain consecutive multiple frames of radar point cloud data and perform preprocessing; S200: According to the maximum sensing radius of the radar, construct multiple cylindrical detection areas centered on the radar with gradually increasing radii; S300: Project the point cloud data falling within each detection area along the emission direction onto the surface of the corresponding detection area to obtain the projected point set corresponding to each detection area; S400: Calculate the dispersion degree of each projected point set on the surface of the corresponding detection area to judge the integrity of the point cloud; Among them, the dispersion degree of the detection area surface is suitable to be represented by the continuous point density in the circumferential direction of the detection area, and the integrity of the point cloud is judged according to the ratio of the calculated actual point density to the corresponding theoretical point density.

2. The method for detecting the integrity of lidar point clouds in the service area according to claim 1, wherein In step S200, the model expression of the detection area in the coordinate system is: ； where x0 and y0 respectively represent the horizontal and vertical coordinates of the radar center point; r represents the distance from the detection area surface to the radar center point; θ represents the angle between the radar installation attitude and the horizontal direction; x, y, and z respectively represent the three-dimensional point coordinates on the detection area surface.

3. The method for detecting the integrity of lidar point clouds in the service area according to claim 1, wherein, In step S200, the interval distance between adjacent detection areas is equal.

4. The method for detecting the integrity of lidar point clouds in the service area according to claim 3, characterized in that, The number of detection areas is set to five, and the distances of each detection area from the radar center point are 20%, 40%, 60%, 80%, and 100% of the radar maximum sensing radius respectively.

5. The method for detecting the integrity of lidar point clouds in the service area according to any one of claims 2-4, characterized in that, Step S300 includes the following specific steps: S310: Calculate the distance from each point in all the point cloud data to the radar center point, and judge the detection area corresponding to each point in the point cloud data according to the calculated distance; S320: Construct the direction vector from the radar center point to each point and parameterize it; S330: Based on the perpendicular intersection of the direction vector and the corresponding detection area, substitute the parameterized direction vector into the expression of the detection area for solution to obtain the coordinate position of each point in all the point cloud data projected onto the surface of the corresponding detection area; S340: Summarize the projected point coordinates on the surface of each detection area to obtain the corresponding projected point set.

6. The method for detecting the integrity of lidar point clouds in the service area according to claim 5, wherein, Step S400 includes the following specific processes: S410: Project each detection area along the axis to form corresponding multiple projection circles, and simultaneously map the projected points on the detection area surface to the corresponding projection circles; S420: Divide each projection circle into multiple equal-area sectors in the circumferential direction to obtain the total number of points corresponding to each detection area and the number of points corresponding to each sector; S430: Calculate the theoretical point density of each detection area according to the total number of points, and calculate the actual point density corresponding to each sector according to the number of points corresponding to the sector; S440: Judge the integrity of the point cloud based on the ratio of the actual point density to the corresponding theoretical point density of each sector.

7. The method for detecting the integrity of lidar point clouds in the service area according to claim 6, wherein, In step S400, the ratio of the actual point density to the corresponding theoretical point density is used as the first integrity degree of the sector; Divide the multiple sectors corresponding to the detection area into multiple groups, and each group includes at least one sector; Calculate the average integrity degree of each group of sectors within consecutive multiple frames; If the calculated average integrity degree is less than the set first threshold, and the number of points corresponding to the sectors in consecutive multiple frames is less than the minimum value of the theoretical number of points, it is determined that there is a high missing area in this group of sectors.

8. The method for detecting the integrity of lidar point clouds in the service area according to claim 6, wherein, When judging the point cloud integrity in step S400, it is also necessary to judge the point cloud distribution in each sector. The specific judgment process is as follows: Divide the multiple sectors corresponding to the detection area into multiple areas to be measured. Each area to be measured includes multiple consecutive sectors, and there is sector overlap between adjacent areas to be measured; Calculate the distance between adjacent points and the corresponding standard deviation within the area to be measured; If the calculated standard deviation is greater than the set second threshold, or the maximum point distance is greater than the set third threshold, it is determined that there is a fault in the corresponding sector.

9. The method for detecting the integrity of lidar point clouds in the service area according to claim 8, wherein, Step S400 also includes the following process: S450: Calculate the theoretical number of points in each sector according to the angular resolution of the radar and the number of horizontal beams, and use the ratio of the actual number of points to the theoretical number of points in the sector as the second integrity; S460: Calculate the second integrity of the sectors corresponding to the same phase angle in each detection area in ascending order of the area radius; S470: If the second integrity of the sector corresponding to a certain phase angle does not continue to increase, it indicates that there is an abnormality in the sector corresponding to this phase angle.

10. The method for detecting the integrity of lidar point clouds in the service area according to claim 9, characterized in that, The specific judgment process for the abnormal situation in step S470 is as follows: Calculate the difference between the distances of two consecutive points in the abnormal sector; If the calculated difference is less than the set fourth threshold, it indicates that the point cloud in this sector is continuous and complete; If the calculated difference is greater than the set fourth threshold and the number of points in the sector is less than the minimum value of the theoretical number of points, it is determined that this sector is a missing or abnormal area; Perform difference judgment on the missing or abnormal area for multiple consecutive frames. If the same missing or abnormal area always exists, it is determined that there is an abnormality in the data transmission or radar sensor in this area.

Images