Pedestrian detection method, device and equipment for airport boundaries and medium

Abstract

The invention provides a pedestrian detection method and device for an airport boundary, equipment and a medium, is used for the technical field of airport boundary detection, and can solve the technical problem of low detection efficiency and accuracy of whether pedestrians exist in the boundary in the prior art. The method comprises the steps of performing target detection on radar data corresponding to an airport boundary to obtain a plurality of point clouds, and performing clustering to obtain a target point cloud cluster; calculating a distance mean value, a distance standard deviation and a target angle difference value corresponding to the target point cloud cluster to obtain a plurality of spatial characteristic indexes, and determining a plurality of speed characteristic indexes according to a speed mean value and a speed variance value corresponding to the target point cloud cluster; determining a time characteristic index according to a target radar detection result corresponding to the target point cloud cluster, and determining a corresponding characteristic value; comparing each characteristic value with a corresponding preset characteristic threshold value to obtain a corresponding comparison result, and determining a detection result corresponding to the airport boundary; therefore, the efficiency and accuracy of boundary pedestrian detection are improved.

Description

Technical Field

[0001] This application relates to the technical field of airport perimeter detection, and more particularly to a method, apparatus, equipment and medium for pedestrian detection at airport perimeters. Background Technology

[0002] Airport perimeter fencing is an important component of the airport security system, primarily used to prevent unauthorized personnel from entering the flight area or sensitive areas. Airport perimeter fencing typically uses a metal mesh structure (such as barbed wire, steel wire mesh, welded wire mesh, etc.), characterized by its long coverage area, open environment, and often accompanied by surrounding facilities such as grasslands, shrubs, and signs. Airport perimeter fencing has an extremely low tolerance for false alarms, requiring a rapid response once an alarm is triggered to quickly remove pedestrians.

[0003] Existing pedestrian detection methods at airport perimeters mostly employ simple speed or energy thresholds. That is, they determine whether a person is a pedestrian by determining the speed of the target or the corresponding radar echo energy. This makes it difficult to effectively distinguish between real pedestrian targets and non-pedestrian targets such as swaying vegetation, resulting in a high false alarm rate and failing to meet the high-level security requirements of airports.

[0004] Therefore, existing pedestrian detection methods for airport perimeters suffer from low detection efficiency and accuracy. Summary of the Invention

[0005] This application provides a method, apparatus, equipment, and medium for pedestrian detection at airport perimeters, in order to solve the problems of low detection efficiency and accuracy in existing pedestrian detection methods at airport perimeters.

[0006] Firstly, this application provides a pedestrian detection method for airport perimeter fencing, the method comprising: Target detection is performed on the radar data corresponding to the airport perimeter to obtain multiple point clouds. Based on multiple preset thresholds, the point clouds are clustered to obtain at least one target point cloud cluster. Calculate the mean distance, standard deviation of distance, and target angle difference corresponding to the target point cloud cluster to obtain multiple spatial feature indicators, and determine multiple velocity feature indicators based on the mean velocity and velocity variance corresponding to the target point cloud cluster; Based on the target radar detection results corresponding to the target point cloud cluster, determine the time characteristic index, and determine the characteristic values ​​corresponding to the spatial characteristic index, velocity characteristic index and time characteristic index respectively; Each feature value and its corresponding preset feature threshold are compared to obtain the comparison results. Based on the comparison results, the detection results corresponding to the airport perimeter are determined.

[0007] In some embodiments of this application, point clouds are clustered according to multiple preset thresholds to obtain at least one target point cloud cluster, including: Based on the distance difference and angle difference between any two point clouds, the distance difference is compared with the corresponding preset threshold, and the angle difference is compared with the corresponding preset threshold to obtain the corresponding comparison result. If the comparison results show that both the distance difference and the angle difference are less than their respective preset thresholds, then the point cloud is marked as a point cloud to be clustered, and all point clouds to be clustered are clustered to obtain the corresponding target point cloud clusters. If the comparison result shows that at least one of the distance difference and angle difference is not less than the corresponding preset threshold, then the point cloud is marked as a non-clustered point cloud.

[0008] In some embodiments of this application, the mean distance, standard deviation of distance, and target angle difference corresponding to the target point cloud cluster are calculated to obtain multiple spatial feature indicators, including: Based on the distance and angle values ​​corresponding to each point cloud, calculate the mean distance, standard deviation of distance, and the target angle difference between the maximum and minimum angle values; Based on the cluster center corresponding to the target point cloud cluster, the corresponding distance variance is calculated, and multiple spatial feature indicators are obtained based on the mean distance, standard deviation of distance, target angle difference, distance variance, and the number of point clouds corresponding to the target point cloud cluster.

[0009] In some embodiments of this application, multiple velocity characteristic indicators are determined based on the mean velocity and variance velocity values ​​corresponding to the target point cloud cluster, including: Based on the velocity values ​​corresponding to each point cloud, calculate the mean velocity and the variance velocity, and based on the echo intensity values ​​corresponding to each point cloud, normalize the echo intensity values ​​to obtain the weight values ​​corresponding to the point clouds. Based on the corresponding weight values, the velocity values ​​of the point cloud are weighted and summed to obtain the target velocity value corresponding to the current target point cloud cluster. Based on the multiple target velocity values ​​corresponding to the target point cloud cluster in the historical time, the difference between adjacent target velocity values ​​is calculated. Based on the positive quantity corresponding to the positive difference, determine the division value between the positive quantity and the total quantity of all differences to obtain the speed change value. Based on the speed mean, speed variance, and speed change value, determine multiple speed characteristic indicators.

[0010] In some embodiments of this application, time characteristic indicators are determined based on the target radar detection results corresponding to the target point cloud cluster, including: Based on all target radar detection results, determine the duration of existence of the target point cloud clusters, and determine whether there is any overlap in the positions of the point cloud clusters corresponding to each target radar detection result; the target radar detection result is the radar detection result that detected the target point cloud clusters. If they exist, determine the duration of overlap, and based on the duration of existence and the duration of overlap, determine the time characteristic indicators; If it does not exist, then determine the time characteristic index based on the duration of its existence.

[0011] In some embodiments of this application, determining the characteristic values ​​corresponding to the spatial characteristic index, velocity characteristic index, and time characteristic index respectively includes: Based on their respective multiple indicator thresholds, the abnormal characteristic indicators corresponding to spatial characteristic indicators, velocity characteristic indicators and time characteristic indicators are determined respectively; the abnormal characteristic indicators are the characteristic indicators that exceed the corresponding indicator thresholds. Based on the division between the number of abnormal indicators and the number of characteristic indicators, the characteristic values ​​corresponding to the spatial characteristic indicators, velocity characteristic indicators, and time characteristic indicators are obtained; the number of abnormal indicators is the number corresponding to the abnormal characteristic indicators, and the number of characteristic indicators is the number of indicators corresponding to the spatial characteristic indicators, velocity characteristic indicators, and time characteristic indicators.

[0012] In some embodiments of this application, determining the detection result corresponding to the airport perimeter based on the comparison results includes: Determine the comparison results; If the comparison results show that the feature values ​​corresponding to the spatial feature index, velocity feature index, and time feature index are all less than the corresponding preset feature thresholds, then the corresponding detection result is determined to be that a pedestrian is approaching the airport perimeter. If the comparison result shows that at least one of the feature values ​​is not less than the corresponding preset feature threshold, then the corresponding detection result is determined to be that no pedestrians approach the airport perimeter.

[0013] Secondly, this application provides a pedestrian detection device for airport perimeter fencing, the device comprising: The detection module is used to detect targets in the radar data corresponding to the airport perimeter, obtain multiple point clouds, and cluster the point clouds according to multiple preset thresholds to obtain at least one target point cloud cluster. The calculation module is used to calculate the mean distance, standard deviation of distance, and target angle difference corresponding to the target point cloud cluster, to obtain multiple spatial feature indicators, and to determine multiple velocity feature indicators based on the mean velocity and velocity variance corresponding to the target point cloud cluster. The determination module is used to determine the time characteristic index based on the target radar detection results corresponding to the target point cloud cluster, and to determine the characteristic values ​​corresponding to the spatial characteristic index, velocity characteristic index and time characteristic index respectively. The comparison module is used to compare each feature value with its corresponding preset feature threshold to obtain the corresponding comparison result, and to determine the detection result corresponding to the airport perimeter based on the comparison result.

[0014] Thirdly, this application provides a computer device, including: a processor, and a memory communicatively connected to the processor; The memory stores the instructions that the computer executes; The processor executes computer execution instructions stored in memory to implement the method of this application.

[0015] Fourthly, this application provides a computer-readable storage medium storing program code, which, when executed by a processor, is used to implement the method of this application.

[0016] Compared with existing technologies, the method in this application performs target detection on airport perimeter radar data and clusters point clouds based on preset thresholds. This integrates discrete radar echo points into at least one target point cloud cluster, effectively aggregating the scattered signals of the same target and avoiding isolated interference from individual echo points. Simultaneously, it preliminarily filters out unrelated noisy point clouds, improving the accuracy of subsequent detection. The method calculates spatial characteristic indicators such as the mean distance, standard deviation of distance, and target angle difference of the target point cloud cluster, as well as velocity characteristic indicators such as the mean velocity and velocity variance. This accurately determines the spatial distribution and motion patterns of the target. The spatial characteristic indicators effectively distinguish between the compact and stable point cloud distribution of pedestrians and the loose and discrete point cloud characteristics of vegetation and metal mesh vibrations. The velocity characteristic indicators clearly distinguish between the stable motion trend of pedestrians and the irregular oscillation characteristics of non-targets. Furthermore, by analyzing the target... Radar detection results determine temporal characteristic indicators and extract feature values ​​from various dimensions, enabling the determination of the target's duration of existence and location migration characteristics. This further distinguishes between the continuous movement of pedestrians and non-target features such as the fixed presence of metal mesh and the short-term appearance of vegetation. By comparing each feature value with the corresponding preset feature thresholds and determining the detection results, a comprehensive judgment of spatial, velocity, and temporal features is achieved. This ensures that real pedestrian targets are accurately identified and alarms are triggered because they meet all feature threshold requirements. At the same time, feature mismatch filtering filters out false targets caused by metal mesh reflections and vibrations, vegetation swaying, etc., significantly reducing the false alarm rate. Furthermore, based on the radar's all-weather operating characteristics and efficient data processing, while ensuring real-time detection and high reliability, it adapts to the security needs of the complex environment of airport perimeters, improving the overall accuracy and operational efficiency of airport perimeter security. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0018] Figure 1 A flowchart illustrating a pedestrian detection method for airport perimeter fencing provided in this application embodiment; Figure 2 A schematic diagram of the framework of a pedestrian detection method for airport perimeter provided in an embodiment of this application; Figure 3 A schematic diagram of the structure of a pedestrian detection device for an airport perimeter provided in this application embodiment; Figure 4 This is a structural block diagram of an apparatus for performing a pedestrian detection method for an airport perimeter according to an embodiment of this application. Detailed Implementation

[0019] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0020] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0021] Figure 1 This is a flowchart illustrating a pedestrian detection method for airport perimeter fencing provided in an embodiment of this application. Figure 1 As shown, this pedestrian detection method for airport perimeter fencing may include the following steps: S110. Target detection is performed on the radar data corresponding to the airport perimeter to obtain multiple point clouds. Based on multiple preset thresholds, the point clouds are clustered to obtain at least one target point cloud cluster.

[0022] Point cloud refers to the discrete target points extracted by millimeter-wave radar when detecting the airport perimeter area. It is obtained by receiving the echo signals reflected by the target, processing them through range-dimensional fast Fourier transform, velocity-dimensional fast Fourier transform, and angle estimation. Each target point carries information such as the target's distance, angle, radial velocity, and echo intensity relative to the radar. It is a digital representation of various reflection sources in the detection area, such as pedestrians, metal mesh, vegetation, and debris.

[0023] The preset threshold is a pre-determined critical value used to determine whether different point clouds can be clustered. It can be determined from the distance and angle between point clouds. If the distance and angle values ​​between any two point clouds are both less than the corresponding preset threshold, it indicates that the two point clouds are adjacent in the distance and angle dimensions and can be clustered.

[0024] A target point cloud cluster is a set of related point clouds formed by clustering discrete point clouds extracted by target detection according to preset distance and angle thresholds. It can be understood that point clouds in the same cluster come from the same reflection source, such as a single pedestrian, a shrub, or a section of metal mesh.

[0025] Based on this, radar is deployed at the airport perimeter to receive the echo data fed back by the radar and to perform target detection on the echo data. For example, the constant false alarm rate (CFAR) algorithm is used to detect targets on the radar data to obtain multiple point clouds. Based on multiple preset thresholds, the point clouds that are related are determined in both distance and angle dimensions and clustered to obtain at least one target point cloud cluster.

[0026] S120. Calculate the mean distance, standard deviation of distance, and target angle difference corresponding to the target point cloud cluster to obtain multiple spatial feature indicators, and determine multiple velocity feature indicators based on the mean velocity and velocity variance corresponding to the target point cloud cluster.

[0027] Among them, the target angle difference is calculated for a single target point cloud cluster by measuring the difference between the maximum and minimum values ​​of all discrete point clouds contained therein in the angular dimension, thereby representing the distribution range of the target point cloud cluster in the azimuth direction. It can be understood as the lateral coverage width of the same target under the radar detection view. That is, the larger the target angle difference, the larger the lateral coverage width of the target.

[0028] Spatial feature indicators are a set of quantitative parameters extracted based on the spatial distribution characteristics of target point cloud clusters. These parameters may include mean distance, standard deviation of distance, target angle difference, number of point clouds, spatial compactness index, etc. These indicators comprehensively characterize the spatial attributes of target point cloud clusters from three dimensions: location concentration, coverage, and structural stability.

[0029] Velocity characteristic indicators are a set of quantitative parameters extracted from the motion state of target point cloud clusters over multiple consecutive scanning cycles. These parameters may include mean velocity, velocity variance, velocity direction consistency index, and velocity positive and negative change frequency. These indicators characterize the motion characteristics of the target from three dimensions: motion trend, stability, and directional regularity. They are the key basis for distinguishing between purposefully moving targets, i.e., pedestrians, and irregularly oscillating targets, such as vibrating metal mesh or swaying vegetation.

[0030] Based on this, multiple spatial feature indicators are obtained by calculating the mean distance, standard deviation of distance, and target angle difference corresponding to the target point cloud cluster. Furthermore, multiple velocity feature indicators are determined based on the mean velocity and velocity variance corresponding to the target point cloud cluster. This allows for subsequent determination of whether the target point cloud cluster is a pedestrian target near the airport perimeter based on its spatial distribution and velocity characteristics.

[0031] S130. Based on the target radar detection results corresponding to the target point cloud cluster, determine the time characteristic index, and determine the characteristic values ​​corresponding to the spatial characteristic index, velocity characteristic index and time characteristic index respectively.

[0032] Among them, the target radar detection result is the radar detection result of the target point cloud cluster.

[0033] Temporal characteristic indicators are a set of quantitative parameters extracted from the two dimensions of temporal persistence and spatial stability based on the target radar detection results. They can include the length of time the target exists continuously, the frequency of the target's appearance and disappearance, the migration trend of the spatial location over time, and the number of frames the target can be continuously detected. These indicators are based on the behavioral patterns of the target in the time dimension, thereby distinguishing between real pedestrians that are constantly moving and non-pedestrian targets that are fixed or appear for a short time.

[0034] Eigenvalues ​​are specific numerical values ​​obtained by quantifying spatial, velocity, and temporal feature indicators. This transforms abstract target features, such as spatial concentration and motion stability, into comparable specific values, avoiding subjective judgment errors and ensuring the consistency of detection logic. In practical applications, eigenvalues ​​can be determined based on the number of target indicators included in each of the spatial, velocity, and temporal feature indicators. In other words, it represents the number of indicators that match the characteristics of pedestrians. The larger the eigenvalue corresponding to an indicator, the more closely the indicator matches the characteristics of pedestrian targets in various dimensions, and the more likely the corresponding target point cloud cluster is a pedestrian.

[0035] Based on this, by continuously detecting radar echo data, target radar detection results containing target point cloud clusters are obtained. Then, based on the duration and trajectory of the target point cloud clusters, time characteristic indicators are determined. Based on the spatial characteristic indicators, velocity characteristic indicators, and time characteristic indicators, corresponding quantitative calculations are performed to obtain the corresponding characteristic values.

[0036] S140. Compare each feature value with its corresponding preset feature threshold to obtain the corresponding comparison result, and determine the detection result corresponding to the airport perimeter based on the comparison result.

[0037] The preset feature threshold is a pre-determined critical value used to determine whether the feature values ​​corresponding to each feature indicator exceed the range. If the feature values ​​corresponding to the spatial feature indicator, velocity feature indicator, and time feature indicator are all greater than the range of the preset feature threshold, it indicates that the target point cloud cluster has a large number of target feature indicators that conform to the characteristics of pedestrians in the three dimensions, and the target point cloud cluster is very likely to be a pedestrian. If the feature value corresponding to any dimension is less than the range corresponding to the preset feature threshold, it indicates that the target point cloud cluster does not meet the characteristic performance of pedestrian targets in that dimension, and the target point cloud cluster is not a pedestrian.

[0038] The detection results are based on spatial features, velocity features, and distribution time features. The final judgment is obtained after multi-dimensional feature determination of the target point cloud cluster. The conclusion includes whether the target is a pedestrian near the airport perimeter, in which case further prompting information needs to be output so that relevant airport personnel can promptly drive the pedestrian away; and whether the conclusion is that the target is not a pedestrian near the airport perimeter, in which case no prompting information needs to be output.

[0039] Based on this, by comparing the value of each feature indicator with the value of its corresponding preset feature threshold, it can be determined whether the feature values ​​are all within the threshold range corresponding to the preset feature threshold, so as to determine the detection result corresponding to the airport perimeter.

[0040] Based on the feasible implementation of S110 described above, this application further provides a method for clustering point clouds according to multiple preset thresholds to obtain at least one target point cloud cluster, including: Based on the distance difference and angle difference between any two point clouds, the distance difference is compared with the corresponding preset threshold, and the angle difference is compared with the corresponding preset threshold to obtain the corresponding comparison result. If the comparison results show that both the distance difference and the angle difference are less than their respective preset thresholds, then the point cloud is marked as a point cloud to be clustered, and all point clouds to be clustered are clustered to obtain the corresponding target point cloud clusters. If the comparison result shows that at least one of the distance difference and angle difference is not less than the corresponding preset threshold, then the point cloud is marked as a non-clustered point cloud.

[0041] Among them, the point cloud to be clustered refers to the point cloud that has significant proximity correlation in spatial location (distance and angle dimension) after determining the distance and angle. It can be determined that it comes from the same reflection source, such as a single pedestrian, a concentrated vegetation, or a section of metal mesh.

[0042] Based on this, point clouds that are related are determined by both distance and angle, and clustering is performed to obtain at least one target point cloud cluster.

[0043] Based on the feasible implementation of S120 described above, this application further provides methods for calculating the mean distance, standard deviation of distance, and target angle difference corresponding to the target point cloud cluster, thereby obtaining multiple spatial feature indicators, including: Based on the distance and angle values ​​corresponding to each point cloud, calculate the mean distance, standard deviation of distance, and the target angle difference between the maximum and minimum angle values; Based on the cluster center corresponding to the target point cloud cluster, the corresponding distance variance is calculated, and multiple spatial feature indicators are obtained based on the mean distance, standard deviation of distance, target angle difference, distance variance, and the number of point clouds corresponding to the target point cloud cluster.

[0044] Among them, the cluster center refers to the average position of all point clouds in a target point cloud cluster in distance and angle space. It represents the "geometric center" or "centroid" of the cluster and is the core benchmark for calculating key indicators such as distance variance and point cloud spatial compactness. The distance variance needs to be calculated from the distance deviation between each point cloud in the cluster and the cluster center. The spatial compactness needs to be determined based on the average deviation of each point cloud from the cluster center. Without a cluster center, it is impossible to quantify the spatial concentration of the point cloud.

[0045] Distance variance is a quantitative indicator that measures the degree of dispersion of all point clouds in the distance dimension within a target point cloud cluster. It can be calculated by first taking the mean distance of the cluster center as a benchmark, calculating the deviation of the distance value of each point cloud in the cluster from the mean distance, and then taking the arithmetic mean of the squares of all deviations. The final result is the distance variance. The larger the value of this indicator, the more dispersed the point cloud is in the distance direction; the smaller the value, the more concentrated the point cloud is in the distance direction.

[0046] Based on this, in practical applications, spatial characteristic parameters such as the number of point clouds, the standard deviation of point clouds in the distance direction, the expansion range of point clouds in the angular direction, and the spatial compactness index of point clouds can be calculated for each candidate target cluster. When the point cloud distribution of the candidate target cluster shows concentrated and stable characteristics, it is determined that it has the spatial characteristics of a pedestrian target; when the point cloud distribution is loose and varies greatly, it is determined that it is a non-pedestrian target.

[0047] Based on the feasible implementation of S120 described above, this application further provides a method for determining multiple velocity characteristic indicators based on the mean velocity and variance velocity values ​​corresponding to the target point cloud cluster, including: Based on the velocity values ​​corresponding to each point cloud, calculate the mean velocity and the variance velocity, and based on the echo intensity values ​​corresponding to each point cloud, normalize the echo intensity values ​​to obtain the weight values ​​corresponding to the point clouds. Based on the corresponding weight values, the velocity values ​​of the point cloud are weighted and summed to obtain the target velocity value corresponding to the current target point cloud cluster. Based on the multiple target velocity values ​​corresponding to the target point cloud cluster in the historical time, the difference between adjacent target velocity values ​​is calculated. Based on the positive quantity corresponding to the positive difference, determine the division value between the positive quantity and the total quantity of all differences to obtain the speed change value. Based on the speed mean, speed variance, and speed change value, determine multiple speed characteristic indicators.

[0048] Among them, the echo intensity value is the quantified value of the signal energy reflected back to the radar receiver after the signal emitted by the millimeter-wave radar encounters the target. Its magnitude is related to the target's reflective cross-section, the distance between the target and the radar, the target's material and surface characteristics. It is one of the core attributes inherent in each point cloud and together with distance, angle and velocity information, it constitutes the complete data features of the point cloud.

[0049] Normalization is a process that maps echo intensity values ​​of different magnitudes to a preset fixed range for all point clouds within the same target point cloud cluster. This eliminates the magnitude differences in echo intensity values ​​caused by factors such as distance and environmental attenuation, making the intensity characteristics of different point clouds comparable. It can be linear normalization or maximum-minimum normalization.

[0050] Historical time refers to a preset length of continuous radar detection time window that traces back from the current scanning cycle. Its duration is strongly correlated with the radar scanning cycle. For example, if the scanning cycle is 0.1 seconds, the historical time can be set to 1-3 seconds, which corresponds to 10-30 consecutive scanning cycles. This time window contains complete velocity data of the target point cloud cluster in multiple consecutive scanning cycles in order to calculate the velocity change trend.

[0051] Positive quantity refers to the statistical value of the number of times the difference between two adjacent target speed values ​​in a sequence of multiple target speed values ​​over a historical period is positive. This value reflects the frequency of positive growth in target speed over a historical period. When positive and negative quantities alternate frequently and have similar proportions, it indicates that the target speed has no clear pattern, such as vegetation swaying in strong winds. This can be directly used as one of the bases for identifying non-pedestrian targets, thus improving the accuracy of speed feature analysis.

[0052] The velocity change value is a proportionality coefficient calculated by dividing the positive quantity by the total number of differences. It quantifies the proportion of positive growth of the target velocity over a historical period and reflects the directionality and regularity of velocity change. The smaller the velocity change value, the more stable the velocity change of the target. Since pedestrian targets usually show a velocity mean deviating from zero and a relatively consistent direction, while targets of metal mesh vibration and vegetation swaying usually show a velocity mean close to zero and a frequent change in direction, the target is more likely to be a pedestrian.

[0053] Based on this, by statistically analyzing the radial velocity of the target in multiple consecutive scanning cycles, the mean velocity, the variance velocity, and the velocity direction consistency index are calculated, thereby determining multiple velocity characteristic indicators.

[0054] Based on the feasible implementation of S130 described above, this application further provides methods for determining time characteristic indicators based on the target radar detection results corresponding to the target point cloud cluster, including: Based on all target radar detection results, determine the duration of existence of the target point cloud clusters, and determine whether there is any overlap in the positions of the point cloud clusters corresponding to each target radar detection result; the target radar detection result is the radar detection result that detected the target point cloud clusters. If they exist, determine the duration of overlap, and based on the duration of existence and the duration of overlap, determine the time characteristic indicators; If it does not exist, then determine the time characteristic index based on the duration of its existence.

[0055] The duration of existence refers to the continuous time from when the target point cloud cluster first appears in the radar detection results to the current scanning cycle (or until it disappears). The calculation uses the radar scanning cycle as the smallest unit of time. For example, if the scanning cycle is 0.1 seconds and the cluster appears continuously for 20 cycles, the duration of existence is 2 seconds.

[0056] Point cloud cluster position refers to the spatial position representation of a single target point cloud cluster in the radar coordinate system, usually with the cluster center as the core coordinate. By observing the changes in the point cloud cluster position within a continuous scanning cycle, it is possible to intuitively reflect whether the target is undergoing continuous displacement. For example, when a pedestrian approaches the boundary, the distance value of the point cloud cluster position gradually decreases, thus distinguishing between continuously moving targets and fixed targets.

[0057] Location overlap refers to the state in which the spatial positions of two or more target point cloud clusters meet the overlap judgment condition in continuous or multiple discrete radar detection results. In practical applications, the judgment condition is usually that the difference in the cluster center distance of the point cloud clusters in different detection results is less than the preset location overlap distance threshold, and the angle difference is less than the preset location overlap angle threshold, which means that the point cloud clusters are considered to overlap in space.

[0058] Overlap duration refers to the cumulative time during which a target point cloud cluster continuously meets the position overlap condition in multiple consecutive radar detection results. That is, the continuous time from the first detection of position overlap to the last time the overlap condition is met.

[0059] Based on this, in practical applications, real targets such as pedestrians will exist continuously in the area outside the boundary for a period of time, while false point cloud clusters formed by environmental noise and discrete clutter mostly appear for a short time, that is, their existence duration is extremely short. By observing the existence duration and the continuous positional changes of the target, we can determine the existence time characteristics of the target, so as to determine whether it is a pedestrian target based on the distribution time of the target.

[0060] Based on the feasible implementation of S130 described above, this application further provides methods for determining the characteristic values ​​corresponding to the spatial characteristic index, velocity characteristic index, and time characteristic index, including: Based on their respective multiple indicator thresholds, the target characteristic indicators corresponding to spatial characteristic indicators, velocity characteristic indicators, and time characteristic indicators are determined respectively; the target characteristic indicators are the characteristic indicators that do not exceed the corresponding indicator thresholds. Based on the division between the number of target indicators and the number of characteristic indicators, we obtain the characteristic values ​​corresponding to the spatial characteristic indicators, velocity characteristic indicators, and time characteristic indicators respectively; the number of target indicators is the number corresponding to the target characteristic indicators, and the number of characteristic indicators is the number of indicators corresponding to the spatial characteristic indicators, velocity characteristic indicators, and time characteristic indicators respectively.

[0061] Among them, the indicator thresholds are pre-set quantitative judgment values ​​for three dimensions: spatial feature indicators, velocity feature indicators, and temporal feature indicators, thereby distinguishing whether the target conforms to or does not conform to the characteristics of a pedestrian. For example, the indicator thresholds for the spatial dimension include the point cloud distance standard deviation threshold, the angle expansion range threshold, the spatial compactness threshold, and the point cloud quantity threshold; the indicator thresholds for the velocity dimension include the velocity mean threshold, the velocity variance threshold, and the velocity direction alternation frequency threshold; the indicator thresholds for the temporal dimension include the minimum continuous existence time threshold, the location migration distance threshold, and the overlap duration ratio threshold. The range corresponding to the indicator thresholds can be understood as the feature range corresponding to a pedestrian. That is, if the feature indicator is within the corresponding indicator threshold range, it indicates that the feature more conforms to the feature performance of a pedestrian target, and the more likely the target point cloud is a pedestrian.

[0062] Based on this, the abstract pedestrian feature requirements can be transformed into quantifiable comparison criteria. By comparing spatial feature indicators, velocity feature indicators, and temporal feature indicators and their respective multiple indicator thresholds, the corresponding target feature indicators can be obtained. Then, based on the division between the number of target indicators and the number of feature indicators, the feature values ​​corresponding to the spatial feature indicators, velocity feature indicators, and temporal feature indicators can be obtained. The larger the feature value corresponding to the feature indicator, the more the feature indicator matches the characteristics exhibited by the pedestrian target, and the more likely the corresponding target point cloud cluster is a pedestrian.

[0063] Based on the feasible implementation of S140 described above, this application further provides a method for determining the detection results corresponding to the airport perimeter based on the comparison results, including: Determine the comparison results; If the comparison results show that the feature values ​​corresponding to the spatial feature index, velocity feature index, and time feature index are all greater than the corresponding preset feature thresholds, then the corresponding detection result is determined to be that a pedestrian is approaching the airport perimeter. If the comparison result shows that at least one of the feature values ​​is not greater than the corresponding preset feature threshold, then the corresponding detection result is determined to be that no pedestrians approach the airport perimeter.

[0064] Based on this, if the feature values ​​corresponding to the spatial feature index, velocity feature index, and temporal feature index are all greater than the corresponding preset feature thresholds, then the target point cloud cluster is highly consistent with the pedestrian target in terms of spatial feature index, velocity feature index, and temporal feature index. Therefore, it can be determined that the target point cloud cluster is a pedestrian, and a pedestrian is approaching the airport perimeter. If at least one of the feature values ​​is not greater than the corresponding preset feature threshold, then the target point cloud cluster is inconsistent with the pedestrian target in terms of that feature index. Therefore, it can be determined that the target point cloud cluster is not a pedestrian, and the corresponding detection result is that no pedestrian is approaching the airport perimeter.

[0065] Please refer to Figure 2 , Figure 2 A schematic diagram illustrating the framework of a pedestrian detection method for airport perimeter provided in this application embodiment; as follows: Figure 2 The diagram illustrates the complete process of pedestrian detection using millimeter-wave radar at airport perimeter. After system startup, the radar is initialized to the parameters of a metal mesh perimeter scene. Echoes are collected and processed using the millimeter-wave radar to obtain distance / velocity / angle information. Candidate points are extracted using CFAR detection and clustered into target clusters. The system first determines whether the target cluster is located within the fixed reflection area of ​​the perimeter. If so, it is classified as a non-pedestrian target and no alarm is triggered. Otherwise, spatial distribution characteristics, motion characteristics, and temporal stability analyses are performed sequentially. Finally, if all three characteristics meet the pedestrian detection criteria, an alarm is triggered; otherwise, it is classified as a non-pedestrian target and no alarm is triggered.

[0066] Based on the above steps, it can be seen that this application performs target detection on airport perimeter radar data and clusters point clouds based on preset thresholds. This integrates discrete radar echo points into at least one target point cloud cluster, effectively aggregating the scattered signals of the same target, avoiding isolated interference from individual echo points, and initially filtering out unrelated noisy point clouds, thus improving the accuracy of subsequent detection. It calculates spatial characteristic indicators such as the mean distance, standard deviation of distance, and target angle difference of the target point cloud cluster, as well as velocity characteristic indicators such as the mean velocity and velocity variance, thereby accurately determining the spatial distribution and motion patterns of the target. The spatial characteristic indicators can effectively distinguish between the compact and stable point cloud distribution of pedestrians and the loose and discrete point cloud characteristics of vegetation and metal mesh vibrations. The velocity characteristic indicators can clearly distinguish between the stable motion trend of pedestrians and the irregular oscillation characteristics of non-targets. Furthermore, through the target... Radar detection results determine temporal characteristic indicators and extract feature values ​​from various dimensions, enabling the determination of the target's duration of existence and location migration characteristics. This further distinguishes between the continuous movement of pedestrians and non-target features such as the fixed presence of metal mesh and the short-term appearance of vegetation. By comparing each feature value with the corresponding preset feature thresholds and determining the detection results, a comprehensive judgment of spatial, velocity, and temporal features is achieved. This ensures that real pedestrian targets are accurately identified and alarms are triggered because they meet all feature threshold requirements. At the same time, feature mismatch filtering filters out false targets caused by metal mesh reflections and vibrations, vegetation swaying, etc., significantly reducing the false alarm rate. Furthermore, based on the radar's all-weather operating characteristics and efficient data processing, while ensuring real-time detection and high reliability, it adapts to the security needs of the complex environment of airport perimeters, improving the overall accuracy and operational efficiency of airport perimeter security.

[0067] Figure 3 This is a schematic diagram of the structure of a pedestrian detection device for an airport perimeter, provided as an embodiment of this application. Figure 3 As shown, this type of pedestrian detection device for airport perimeter fencing includes: a detection module, a calculation module, a determination module, and a comparison module; wherein: The detection module is used to detect targets in the radar data corresponding to the airport perimeter, obtain multiple point clouds, and cluster the point clouds according to multiple preset thresholds to obtain at least one target point cloud cluster. The calculation module is used to calculate the mean distance, standard deviation of distance, and target angle difference corresponding to the target point cloud cluster, to obtain multiple spatial feature indicators, and to determine multiple velocity feature indicators based on the mean velocity and velocity variance corresponding to the target point cloud cluster. The determination module is used to determine the time characteristic index based on the target radar detection results corresponding to the target point cloud cluster, and to determine the characteristic values ​​corresponding to the spatial characteristic index, velocity characteristic index and time characteristic index respectively. The comparison module is used to compare each feature value with its corresponding preset feature threshold to obtain the corresponding comparison result, and to determine the detection result corresponding to the airport perimeter based on the comparison result.

[0068] In this embodiment of the application, the detection module can also be specifically used for: Based on the distance difference and angle difference between any two point clouds, the distance difference is compared with the corresponding preset threshold, and the angle difference is compared with the corresponding preset threshold to obtain the corresponding comparison result. If the comparison results show that both the distance difference and the angle difference are less than their respective preset thresholds, then the point cloud is marked as a point cloud to be clustered, and all point clouds to be clustered are clustered to obtain the corresponding target point cloud clusters. If the comparison result shows that at least one of the distance difference and angle difference is not less than the corresponding preset threshold, then the point cloud is marked as a non-clustered point cloud.

[0069] In this embodiment of the application, the calculation module can also be specifically used for: Based on the distance and angle values ​​corresponding to each point cloud, calculate the mean distance, standard deviation of distance, and the target angle difference between the maximum and minimum angle values; Based on the cluster center corresponding to the target point cloud cluster, the corresponding distance variance is calculated, and multiple spatial feature indicators are obtained based on the mean distance, standard deviation of distance, target angle difference, distance variance, and the number of point clouds corresponding to the target point cloud cluster.

[0070] In this embodiment of the application, the calculation module can also be specifically used for: Based on the velocity values ​​corresponding to each point cloud, calculate the mean velocity and the variance velocity, and based on the echo intensity values ​​corresponding to each point cloud, normalize the echo intensity values ​​to obtain the weight values ​​corresponding to the point clouds. Based on the corresponding weight values, the velocity values ​​of the point cloud are weighted and summed to obtain the target velocity value corresponding to the current target point cloud cluster. Based on the multiple target velocity values ​​corresponding to the target point cloud cluster in the historical time, the difference between adjacent target velocity values ​​is calculated. Based on the positive quantity corresponding to the positive difference, determine the division value between the positive quantity and the total quantity of all differences to obtain the speed change value. Based on the speed mean, speed variance, and speed change value, determine multiple speed characteristic indicators.

[0071] In this embodiment of the application, the determining module can also be specifically used for: Based on all target radar detection results, determine the duration of existence of the target point cloud clusters, and determine whether there is any overlap in the positions of the point cloud clusters corresponding to each target radar detection result; the target radar detection result is the radar detection result that detected the target point cloud clusters. If they exist, determine the duration of overlap, and based on the duration of existence and the duration of overlap, determine the time characteristic indicators; If it does not exist, then determine the time characteristic index based on the duration of its existence.

[0072] In this embodiment of the application, the determining module can also be specifically used for: Based on their respective multiple indicator thresholds, the target characteristic indicators corresponding to spatial characteristic indicators, velocity characteristic indicators, and time characteristic indicators are determined respectively; the target characteristic indicators are the characteristic indicators that do not exceed the corresponding indicator thresholds. Based on the division between the number of target indicators and the number of characteristic indicators, we obtain the characteristic values ​​corresponding to the spatial characteristic indicators, velocity characteristic indicators, and time characteristic indicators respectively; the number of target indicators is the number corresponding to the target characteristic indicators, and the number of characteristic indicators is the number of indicators corresponding to the spatial characteristic indicators, velocity characteristic indicators, and time characteristic indicators respectively.

[0073] In this embodiment of the application, the comparison module can also be specifically used for: Determine the comparison results; If the comparison results show that the feature values ​​corresponding to the spatial feature index, velocity feature index, and time feature index are all greater than the corresponding preset feature thresholds, then the corresponding detection result is determined to be that a pedestrian is approaching the airport perimeter. If the comparison result shows that at least one of the feature values ​​is not greater than the corresponding preset feature threshold, then the corresponding detection result is determined to be that no pedestrians approach the airport perimeter.

[0074] Figure 4 This is a schematic diagram of the structure of an apparatus for performing a pedestrian detection method for an airport perimeter according to an embodiment of this application. Figure 4 As shown, the device includes: The device may include one or more processors with processing cores, one or more computer-readable storage media such as memory, communication components, etc. The processor, memory, and communication components are connected via a bus.

[0075] In the specific implementation process, at least one processor executes computer execution instructions stored in memory, causing at least one processor to execute the pedestrian detection method for airport perimeter as described above.

[0076] The specific implementation process of the processor can be found in the above method embodiments, and its implementation principle and technical effect are similar, so it will not be repeated here.

[0077] Furthermore, the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this application can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0078] The memory may include Random Access Memory (RAM) and may also include Non-volatile Memory (NVM), such as at least one disk storage device.

[0079] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0080] In some embodiments, a computer program product is also provided, including a computer program or instructions that, when executed by a processor, implement the steps in any of the above-described pedestrian detection methods for airport perimeters.

[0081] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0082] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be performed by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor.

[0083] Therefore, embodiments of this application provide a computer-readable storage medium storing a plurality of program codes that can be loaded by a processor to execute the steps in any of the pedestrian detection methods for airport perimeters provided in embodiments of this application.

[0084] The storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0085] According to one aspect of this application, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium.

[0086] Since the instructions stored in the storage medium can execute the steps in any of the pedestrian detection methods for airport perimeters provided in the embodiments of this application, the beneficial effects that any of the pedestrian detection methods for airport perimeters provided in the embodiments of this application can achieve can be realized. For details, please refer to the previous embodiments, which will not be repeated here.

[0087] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the appended claims.

[0088] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope.

Claims

1. A method for pedestrian detection at airport perimeter fencing, characterized in that, The method includes: Target detection is performed on the radar data corresponding to the airport perimeter to obtain multiple point clouds. Based on multiple preset thresholds, the point clouds are clustered to obtain at least one target point cloud cluster. Calculate the mean distance, standard deviation of distance, and target angle difference corresponding to the target point cloud cluster to obtain multiple spatial feature indicators, and determine multiple velocity feature indicators based on the mean velocity and velocity variance corresponding to the target point cloud cluster; Based on the target radar detection results corresponding to the target point cloud cluster, determine the time characteristic index, and determine the characteristic values ​​corresponding to the spatial characteristic index, the velocity characteristic index, and the time characteristic index respectively; Each of the aforementioned feature values ​​and their corresponding preset feature thresholds are compared to obtain the corresponding comparison results, and the detection results corresponding to the airport perimeter are determined based on the comparison results.

2. The method according to claim 1, characterized in that, The step of clustering the point cloud according to multiple preset thresholds to obtain at least one target point cloud cluster includes: Based on the distance difference and angle difference between any of the point clouds, the distance difference is compared with the corresponding preset threshold, and the angle difference is compared with the corresponding preset threshold to obtain the corresponding comparison result; If the comparison result is that both the distance difference and the angle difference are less than their respective preset thresholds, then the point cloud is marked as a point cloud to be clustered, and all the point clouds to be clustered are clustered to obtain the corresponding target point cloud cluster. If the comparison result is that at least one of the distance difference and the angle difference is not less than the corresponding preset threshold, then the point cloud is marked as a non-clustered point cloud.

3. The method according to claim 1, characterized in that, The calculation of the mean distance, standard deviation of distance, and target angle difference corresponding to the target point cloud cluster yields multiple spatial feature indicators, including: Based on the distance and angle values ​​corresponding to each point cloud, calculate the mean distance, the standard deviation of the distance, and the target angle difference between the maximum and minimum angle values; Based on the cluster center corresponding to the target point cloud cluster, the corresponding distance variance is calculated, and multiple spatial feature indicators are obtained based on the mean distance, standard deviation of distance, target angle difference, distance variance, and the number of point clouds corresponding to the target point cloud cluster.

4. The method according to claim 1, characterized in that, The step involves determining multiple velocity characteristic indicators based on the mean and variance of the velocity corresponding to the target point cloud cluster, including: Based on the velocity values ​​corresponding to each point cloud, the mean velocity and the variance velocity are calculated, and based on the echo intensity values ​​corresponding to each point cloud, the echo intensity values ​​are normalized to obtain the weight values ​​corresponding to the point clouds. Based on the corresponding weight values, the velocity values ​​of the point cloud are weighted and summed to obtain the target velocity value corresponding to the current target point cloud cluster. Based on multiple target velocity values ​​corresponding to the target point cloud cluster in the historical time period, the difference between adjacent target velocity values ​​is calculated. Based on the positive quantity corresponding to the positive difference among the differences, the division value between the positive quantity and the total quantity of all the differences is determined to obtain the speed change value. Based on the speed mean, the speed variance value and the speed change value, multiple speed characteristic indicators are determined.

5. The method according to claim 1, characterized in that, The step of determining time characteristic indicators based on the target radar detection results corresponding to the target point cloud cluster includes: Based on all the target radar detection results, the existence duration of the target point cloud cluster is determined, and based on the position of the point cloud cluster corresponding to each target radar detection result, it is determined whether the target point cloud clusters have overlapping positions; the target radar detection result is the radar detection result that detects the target point cloud cluster. If they exist, the overlap duration is determined, and the time characteristic index is determined based on the existence duration and the overlap duration. If it does not exist, then the time characteristic index is determined based on the duration of its existence.

6. The method according to claim 1, characterized in that, The step of determining the feature values ​​corresponding to the spatial feature index, the velocity feature index, and the time feature index includes: Based on their respective multiple indicator thresholds, target feature indicators are determined for the spatial feature indicator, the velocity feature indicator, and the time feature indicator, respectively; the target feature indicator is a feature indicator that does not exceed the corresponding indicator threshold. The feature values ​​corresponding to the spatial feature index, the velocity feature index, and the time feature index are obtained by dividing the number of target indicators by the number of feature indicators; the number of target indicators is the number of target feature indicators, and the number of feature indicators is the number of indicators corresponding to the spatial feature index, the velocity feature index, and the time feature index.

7. The method according to claim 1, characterized in that, The step of determining the detection result corresponding to the airport perimeter based on the comparison result includes: Determine the comparison results; If the comparison result shows that the feature values ​​corresponding to the spatial feature index, the velocity feature index, and the time feature index are all greater than the corresponding preset feature threshold, then the corresponding detection result is determined to be that a pedestrian is approaching the airport perimeter. If the comparison result is that at least one of the feature values ​​is not greater than the corresponding preset feature threshold, then the corresponding detection result is determined to be that no pedestrians approach the airport perimeter.

8. A pedestrian detection device for airport perimeter fencing, characterized in that, The device includes: The detection module is used to perform target detection on radar data corresponding to the airport perimeter, obtain multiple point clouds, and cluster the point clouds according to multiple preset thresholds to obtain at least one target point cloud cluster. The calculation module is used to calculate the mean distance, standard deviation of distance, and target angle difference corresponding to the target point cloud cluster to obtain multiple spatial feature indicators, and to determine multiple velocity feature indicators based on the mean velocity and velocity variance corresponding to the target point cloud cluster. The determination module is used to determine the time feature index based on the target radar detection results corresponding to the target point cloud cluster, and to determine the feature values ​​corresponding to the spatial feature index, the velocity feature index and the time feature index respectively; The comparison module is used to compare each of the aforementioned feature values ​​and their respective preset feature thresholds to obtain the corresponding comparison results, and to determine the detection results corresponding to the airport perimeter based on the comparison results.

9. A computer device, characterized in that, include: One or more processors; Memory; One or more programs, wherein the one or more programs are stored in memory and configured to be executed by one or more processors, the one or more programs being configured to perform the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores program code that can be called by a processor to perform the method as described in any one of claims 1 to 7.

Images