Marine target detection method, device and equipment and storage medium

Abstract

The invention discloses a sea target detection method, device and equipment and a storage medium, and the method comprises the steps: carrying out the coarse filtering of an initial coordinate point set through employing a self-adaptive filtering algorithm, obtaining a first screening coordinate point set, and dynamically obtaining a wake flow region according to the first screening coordinate point set; performing fine filtering on the first screening coordinate point set based on the wake flow region to obtain a second screening coordinate point set; and clustering the second screening coordinate point set by adopting an adaptive clustering algorithm based on physical parameters of the laser radar to obtain a detection target. The comprehensiveness of laser radar detection information is improved through fusion of point cloud data, coarse filtering and fine filtering are performed on an initial coordinate point set obtained after fusion, and filtered results are aggregated based on physical parameters of the laser radar to complete detection of each target obstacle. Therefore, in a complex sea surface environment, the laser radar can be used to carry out efficient and real-time detection on the water surface target obstacle.

Description

Technical Field

[0001] This invention relates to the field of unmanned surface vessel control technology, and in particular to a method, apparatus, equipment and storage medium for detecting maritime targets. Background Technology

[0002] As a multifunctional intelligent platform, unmanned surface vessels (USVs) have been widely used in the marine field in recent years. Due to their lightweight design, economy, long endurance, high maneuverability, fast navigation, and adaptability to harsh environments, USVs have shown great application potential in fields such as marine environmental monitoring, seabed topography mapping, and resource development.

[0003] In the autonomous path planning of unmanned surface vessels (USVs), obstacle detection is one of the core tasks. Traditional vessels typically rely on optoelectronic devices, such as visible light and infrared sensors, and marine radar for environmental perception. Optoelectronic devices can provide high refresh rate and high resolution image information, while marine radar is known for its long detection range and all-weather navigation capabilities. However, both types of sensors have significant limitations in complex waters: radar has a low update frequency and insufficient resolution in close-range, high-speed target scenarios, and is susceptible to sea clutter interference in near-shore areas; optoelectronic devices, due to their reliance on external light sources, suffer from instability when weather and lighting conditions change, and cannot accurately detect the distance and true size of targets, making it difficult to meet the requirements for precise detection and positioning.

[0004] As an active sensor, lidar combines the advantages of optical imaging and radar, providing target distance, orientation information, and three-dimensional contour data. LiDAR's active laser ranging technology is unaffected by lighting or weather, has a fast scanning speed, and is suitable for the detection and identification of short- to medium-range targets. LiDAR's three-dimensional information acquisition capability makes it a key tool for unmanned surface vessels (USVs) to perform target detection, size measurement, distance measurement, and speed measurement, and is of great value for obstacle avoidance, target positioning, and tracking.

[0005] However, lidar also faces challenges in practical applications. LiDAR point cloud data may be sparse under complex sea conditions, limiting its ability to detect small targets, and it is susceptible to sea clutter interference. Unmanned surface vessels (USVs) typically operate in complex environments such as nearshore areas, tidal flats, or inland waterways, where the background is variable and interference is strong. Therefore, how to achieve efficient target detection and recognition based on lidar under complex sea conditions and strong interference has become a critical technical problem that needs to be solved in USV environmental perception technology. Summary of the Invention

[0006] This invention provides a method, apparatus, equipment, and storage medium for detecting maritime targets, enabling accurate obstacle detection of unmanned surface vessels based on lidar.

[0007] According to the first aspect of this invention, a method for detecting maritime targets is provided, comprising: scanning the surrounding environment of an unmanned surface vessel with a lidar to obtain point cloud data, and fusing and transforming the point cloud data on a grid map to obtain an initial set of coordinate points;

[0008] An adaptive filtering algorithm is used to coarsely filter the initial coordinate point set to obtain a first filtered coordinate point set, and the wake region is dynamically obtained based on the first filtered coordinate point set.

[0009] Based on the wake region, the first set of filtered coordinate points is finely filtered to obtain the second set of filtered coordinate points;

[0010] Based on the physical parameters of the lidar, an adaptive clustering algorithm is used to cluster the second set of selected coordinate points to obtain the detection target.

[0011] According to another aspect of the present invention, a target detection device is provided, the device comprising: a fusion conversion module, configured to scan the surrounding environment of an unmanned surface vessel using a lidar to obtain point cloud data, and to fuse and convert the point cloud data on a grid map to obtain an initial set of coordinate points;

[0012] The coarse filtering module is used to perform coarse filtering on the initial coordinate point set using an adaptive filtering algorithm to obtain a first filtered coordinate point set, and to dynamically obtain the wake region based on the first filtered coordinate point set.

[0013] The fine filtering module is used to perform fine filtering on the first set of filtered coordinate points based on the wake region to obtain a second set of filtered coordinate points;

[0014] The clustering module is used to cluster the second set of selected coordinate points based on the physical parameters of the lidar using an adaptive clustering algorithm to obtain the detection target.

[0015] According to another aspect of the present invention, a terminal device is provided, the terminal device comprising: one or more processors;

[0016] Storage device for storing one or more programs.

[0017] When the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any embodiment of the present invention.

[0018] According to another aspect of the present invention, a storage medium for computer-executable instructions is provided, on which a computer program is stored, which, when executed by a processor, implements the method described in any of the embodiments of the present invention.

[0019] The technical solution of this invention improves the comprehensiveness of lidar detection information by fusing point cloud data. By coarsely filtering and finely filtering the initial coordinate point set obtained after fusion, significant clutter and wake clutter in the target movement process are filtered out. The filtered results are then aggregated based on the physical parameters of the lidar to complete the detection of various target obstacles. Thus, lidar can be used to efficiently and in real time detect water surface target obstacles in complex marine environments, providing environmental perception support for the autonomous navigation of unmanned surface vessels.

[0020] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart of a maritime target detection method provided in Embodiment 1 of the present invention;

[0023] Figure 2 This is a structural schematic diagram of an unmanned surface vessel provided according to Embodiment 1 of the present invention;

[0024] Figure 3 This is a schematic diagram of a detection target provided according to an embodiment of the present invention;

[0025] Figure 4 This is a flowchart of a maritime target detection method provided in Embodiment 2 of the present invention;

[0026] Figure 5 This is a schematic diagram of the structure of a maritime target detection device according to Embodiment 3 of the present invention;

[0027] Figure 6 This invention provides a structural block diagram of a terminal device. Detailed Implementation

[0028] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0029] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, apparatus, product, or terminal device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or terminal devices.

[0030] Example 1

[0031] Figure 1 This is a flowchart illustrating a data synchronization method provided in an embodiment of the present invention. This embodiment is applicable to data synchronization of cloud databases. The method can be executed by a data synchronization device, which can be implemented in hardware and / or software, and can be integrated into a terminal device. Figure 1 As shown, the method includes:

[0032] Step S101: Scan the surrounding environment of the unmanned surface vessel with lidar to obtain point cloud data, and fuse and transform the point cloud data on the grid map to obtain an initial coordinate point set.

[0033] Optionally, the point cloud data is fused and transformed on the grid map to obtain an initial set of coordinate points, including: performing a two-dimensional transformation on the pre-created grid map to obtain coordinate points; fusing the transformed multi-frame coordinate points on the grid map to obtain fused coordinate points, and constructing an initial set of coordinate points based on the fused coordinate points, wherein each coordinate point in the initial set of coordinate points is marked with height information.

[0034] in, Figure 2The diagram shows a structural schematic of an unmanned surface vessel (USV). This embodiment uses a small USV as a platform for laser radar experiments. The platform employs a single-hull external engine power system, providing good maneuverability. A laser radar is installed in the stern area of ​​the USV; the specific installation location of the laser radar is not limited in this embodiment. The laser radar can accommodate multiple lasers and has a 360-degree horizontal scanning field of view and a 24-degree vertical scanning field of view, providing nearly one million measurement data points per second. Therefore, in this embodiment, the laser radar installed on the USV will perform real-time scanning of the surrounding environment to obtain point cloud data.

[0035] Specifically, this implementation uses the lidar as the coordinate origin, the forward direction of the carrier (i.e., the unmanned surface vessel) as the positive X-axis, and the right-hand side perpendicular to the forward direction as the positive Y-axis. Based on the lidar's range, the X-axis and Y-axis cutoff distances are set to A and B, respectively, with a grid size of m1. A grid map is then created based on A, B, and m1. After acquiring the point cloud data, it is processed on the grid map. Specifically, the three-dimensional point cloud data is converted into two-dimensional coordinate points to achieve rapid processing of the lidar point cloud data. For example, for any point p in the point cloud data... i (x) i y i z i After a two-dimensional transformation, the coordinates of the point on the XY plane are (g x ,g yFurthermore, this embodiment also obtains height information based on the 3D coordinate information of the 3D point cloud data and marks the height information as an attribute at each coordinate point. Of course, this embodiment is only an example and does not limit the specific process of 2D conversion. Since the point cloud data detected by lidar is relatively sparse in complex sea conditions, the corresponding coordinate points after 2D conversion are also sparse, limiting the ability to detect maritime targets. Therefore, this embodiment fuses the converted multi-frame coordinate points on a grid map to obtain fused coordinate points. Considering the speed of the unmanned surface vessel and the speed of dynamic obstacles, in order to ensure the safety of the unmanned surface vessel's autonomous navigation, the unmanned surface vessel needs to have high real-time performance in detecting maritime obstacles. Therefore, three frames of coordinate points can be fused, which ensures both the accuracy of obstacle detection and good real-time performance. The process of fusing three frames of coordinate points involves fusing a third frame of coordinate points based on the fusion of two frames. This means fusing the results of three LiDAR scans for the same target point, thus avoiding the information loss caused by a single LiDAR scan. In this embodiment, an initial set of coordinate points is constructed based on the fused coordinate points, and each fused coordinate point in the initial set is marked with height information, which is the average height of the three fused coordinate points. This embodiment only uses the fusion of three frames as an example and does not limit the specific number of frames to be fused. The specific number of frames can be set according to the accuracy requirements of maritime target detection and the computing power of the available resources.

[0036] Step S102: The initial coordinate point set is coarsely filtered using an adaptive filtering algorithm to obtain the first set of filtered coordinate points, and the wake region is dynamically obtained based on the first set of filtered coordinate points.

[0037] Optionally, an adaptive filtering algorithm is used to coarsely filter the initial coordinate point set to obtain a first-selected coordinate point set, including: extracting coordinate points from the initial coordinate point set based on height information to obtain an input coordinate point set; using a specified number of coordinate points with the smallest height information in the input coordinate point set as initial seed points; using the Random Sample Consensus (RANSAC) algorithm to estimate a plane model based on the initial seed points to obtain a filtering surface; and filtering the coordinate points located on the filtering surface to obtain the first-selected coordinate point set.

[0038] Specifically, in this embodiment, after obtaining the initial coordinate point set, the Random Sample Consensus (RANSAC) algorithm is used for coarse filtering. Since each coordinate point in the initial coordinate point set is marked with height information, and since the wake itself is broken and horizontal, and the obstacle point cloud reflection is stable and has height, this embodiment will extract coordinate points with horizontal and unknown height information, and use the extracted coordinate points as the input coordinate point set of the RANSAC algorithm, thereby avoiding the processing of invalid data. In addition, in this embodiment, the initial seed point is dynamically selected according to the height distribution of coordinate points in the local area. For example, the coordinate points in the input coordinate point set are sorted in descending order according to the height information, and the first three coordinate points in the sequence are selected as the initial seed points. In this embodiment, the Random Sample Consensus (RANSAC) algorithm is used to perform planar model estimation based on the initial seed points to obtain the filter surface, and the coordinate points located on the filter surface are filtered to obtain the first set of filtered coordinate points. Specifically, the following formula (1) can be used to perform planar fitting to obtain the filter surface:

[0039]

[0040] in, For the estimated plane normal vector, This is the intercept offset. Let be any point on the plane. The plane normal vector can be solved by calculating the covariance matrix of the coordinate points in the local region, as shown in the following formula (2):

[0041]

[0042] Where COV is the covariance matrix of coordinate points within the local region. The mean of the seed points. To input the number of coordinate points in the coordinate point set, The coordinate point number, For the first The covariance matrix of the input coordinate points characterizes the distribution characteristics of the input coordinate point set. By solving the matrix to determine the three eigenvalues ​​and their corresponding eigenvectors, the three principal directions of the spatial distribution of the input coordinate point set can be obtained, and the eigenvector corresponding to the smallest eigenvalue can be used as the plane normal vector.

[0043] Optionally, the wake region is dynamically obtained based on the first set of selected coordinate points, including: detecting the working parameters associated with the first set of selected coordinate points through sensors, wherein the working parameters include speed, ship length, turning angle, maneuverability turning index and maneuverability following index; determining the dynamic wake length, dynamic wake width and dynamic wake curvature radius based on the working parameters; and dynamically determining the wake region based on the dynamic wake length, dynamic wake width and dynamic wake curvature radius.

[0044] Specifically, in this embodiment, after obtaining the first set of selected coordinate points through coarse filtering using a filter surface, although obvious noise and noise in the wake are roughly filtered out, in order to ensure the accuracy of noise removal, this embodiment dynamically obtains the wake region based on the first set of selected coordinate points, so as to facilitate subsequent fine filtering based on the wake region. In this embodiment, the following formula (3) is used to obtain the dynamic wake length, dynamic wake width, and dynamic wake curvature radius:

[0045]

[0046] in, For dynamic wake length, Where S is the dynamic wake width and S is the dynamic wake curvature radius. , and This is the proportionality coefficient. For the ship's speed relative to the ground, As captain, For steering angle, The turning performance index is the maneuverability index. This refers to the maneuverability following index. In this implementation, the above will be... , , , and As a working parameter, the ship's length Specifically, the distance between the foremost and the last coordinate point within the region enclosed by the first set of selected coordinate points is used. The remaining four operating parameters can be detected by other sensors installed on the unmanned surface vessel, such as speed sensors. This embodiment is merely illustrative and does not limit the specific method for determining the operating parameters. With the dynamic wake length, dynamic wake width, and dynamic wake curvature radius known, the range of the wake region can be largely determined.

[0047] Step S103: Based on the wake region, perform fine filtering on the first set of filtered coordinate points to obtain the second set of filtered coordinate points.

[0048] Optionally, a second set of selected coordinate points is obtained by finely filtering the first set of selected coordinate points based on the wake region, including: determining a first set of candidate coordinate points located within the wake region and a second set of candidate coordinate points located outside the wake region; determining a height difference threshold associated with the wake region and deleting coordinate points in the first set of candidate coordinate points whose height information is less than the height difference threshold to obtain a third set of candidate coordinate points; and combining the second set of candidate coordinate points and the third set of candidate coordinate points to obtain the second set of selected coordinate points.

[0049] Optionally, determining the height difference threshold associated with the wake region includes: obtaining a basic target height difference threshold, wherein the basic target height difference threshold dynamically changes with the water surface wave height; obtaining the velocity associated with the first set of selected coordinate points, and determining the height difference threshold based on the velocity and the basic target height difference threshold.

[0050] Specifically, after determining the wake region, this embodiment can perform fine filtering on the first set of selected coordinate points obtained from coarse filtering based on the wake region. Fine filtering primarily targets and precisely removes the remaining wake clutter in the wake region to prevent ships from misjudging the wake as an obstacle during navigation. In this embodiment, a first set of candidate coordinate points located within the wake region and a second set of candidate coordinate points located outside the wake region are determined from the first set of selected coordinate points. Since all coordinate points outside the wake region are considered valid information related to the target, the entire second set of candidate coordinate points is retained. While the first set of candidate coordinate points within the wake region mostly consists of wake clutter, it also contains a small amount of valid information related to the target. Therefore, this embodiment requires filtering out this valid information from the first set of candidate coordinate points.

[0051] In this embodiment, when filtering the first candidate coordinate point set, the following formula (4) is used to obtain the height difference threshold associated with the wake region:

[0052]

[0053] in, As the height difference threshold, and This is the proportionality coefficient. For the ship's speed relative to the ground, The basic target height difference threshold, and The wake wave height is dynamically adjusted according to the water surface wave height. Since the wake wave height above the horizontal plane is usually not very large, after dynamically determining the height difference threshold based on the water surface wave conditions, coordinate points in the first candidate coordinate point set whose height information is less than the height difference threshold are deleted to obtain the third candidate coordinate point set. The third candidate coordinate point set contains a small number of valid coordinate points related to the detected target, such as structural parts like the stern. After obtaining the third candidate coordinate point set by filtering the wake region, the second and third candidate coordinate point sets are combined to obtain the second filtered coordinate point set. Here, "combination" simply refers to the process of merging two point sets into one point set. Therefore, in this embodiment, when finely filtering the wake region, the wake region is determined by combining actual working parameters such as speed and steering angle, thereby ensuring the accuracy of the wake region determination. In addition, by dynamically adjusting the height difference threshold associated with the wake region, the actual wake wave is effectively distinguished from real target obstacles, thereby ensuring accurate filtering of invalid clutter.

[0054] Step S104: Based on the physical parameters of the lidar, an adaptive clustering algorithm is used to cluster the second set of selected coordinate points to obtain the detection target.

[0055] Optionally, an adaptive clustering algorithm is used to cluster the second set of selected coordinate points based on the physical parameters of the lidar to obtain the detection target. This includes: obtaining the dynamic target distance based on the second set of selected coordinate points; using the density-based noise application spatial clustering algorithm DBSCAN to determine the dynamic neighborhood radius based on the lidar's physical parameters and the dynamic target distance; calculating the distance correlation density of the second set of selected coordinate points based on the lidar's maximum detection range and atmospheric attenuation coefficient; dynamically calculating the minimum number of points threshold based on the dynamic neighborhood radius and distance correlation density; clustering the second set of selected coordinate points based on the minimum number of points threshold to obtain clusters; and determining the detection target based on the clusters.

[0056] Specifically, this embodiment addresses the core issue of coordinate point density decaying with distance by proposing an adaptive clustering algorithm constrained by the physical parameters of the lidar, such as the Density-Based Spatial Clustering of Applications with Noise (DBSCAN). After obtaining the second set of selected coordinate points through two steps of coarse filtering and fine filtering, the dynamic target distance D is obtained based on the second set of selected coordinate points, and the following formula (5) is used based on the ranging principle to determine the dynamic neighborhood radius based on the physical parameters of the lidar and the dynamic target distance:

[0057]

[0058] in, For the dynamic neighborhood radius, For dynamic target distance, The divergence angle of the lidar. The standard deviation of ranging noise, For an acceptable false positive probability, and the above... The standard deviation can be calculated based on the second set of selected coordinate points, and This can be based on the preset detection accuracy. Furthermore, this implementation incorporates real-time meteorological data into the DBSCAN algorithm, such as the atmospheric attenuation coefficient m, to dynamically update the distance correlation density. The following formula (6) shows the calculation formula for the distance correlation density:

[0059]

[0060] in, For distance-related density, Where m is the maximum detection range of the lidar, and m is the atmospheric attenuation coefficient. For dynamic target distance, The total number of coordinate points in the second set of selected coordinate points is denoted by m, which is updated in real time based on meteorological conditions. Given that the dynamic neighborhood radius and distance correlation density are known, the minimum number of points threshold can be calculated using the following formula (7). :

[0061]

[0062] Specifically, in this embodiment, after obtaining the minimum number of points threshold, the second set of selected coordinate points can be clustered based on the minimum number of points threshold to obtain clusters. For example, when the minimum number of points threshold is determined to be 30, the second set of selected coordinate points can be clustered to obtain multiple clusters, and each cluster contains 30 coordinate points. Each obtained cluster is then used as the detected obstacle target. Figure 3 The diagram shows a target detection method using a lidar system.

[0063] The technical solution of this invention improves the comprehensiveness of lidar detection information by fusing point cloud data. By coarsely filtering and finely filtering the initial coordinate point set obtained after fusion, significant clutter and wake clutter in the target movement process are filtered out. The filtered results are then aggregated based on the physical parameters of the lidar to complete the detection of various target obstacles. Thus, lidar can be used to efficiently and in real time detect water surface target obstacles in complex marine environments, providing environmental perception support for the autonomous navigation of unmanned surface vessels.

[0064] Example 2

[0065] Figure 4 This is a flowchart of a maritime target detection method provided by an embodiment of the present invention. Based on the above embodiments, this embodiment further includes, after clustering the second set of selected coordinate points using an adaptive clustering algorithm based on the physical parameters of the lidar to obtain the detection target, the method further includes: when it is determined that the weather conditions meet preset requirements, capturing image information of the environment surrounding the unmanned surface vessel, and correcting the detection target based on the image information. For example... Figure 4 As shown, the method includes:

[0066] Step S201: Scan the surrounding environment of the unmanned surface vessel with lidar to obtain point cloud data, and fuse and transform the point cloud data on the grid map to obtain an initial coordinate point set.

[0067] Optionally, the point cloud data is fused and transformed on the grid map to obtain an initial set of coordinate points, including: performing a two-dimensional transformation on the pre-created grid map to obtain coordinate points; fusing the transformed multi-frame coordinate points on the grid map to obtain fused coordinate points, and constructing an initial set of coordinate points based on the fused coordinate points, wherein each coordinate point in the initial set of coordinate points is marked with height information.

[0068] Step S202: The initial coordinate point set is coarsely filtered using an adaptive filtering algorithm to obtain the first set of filtered coordinate points, and the wake region is dynamically obtained based on the first set of filtered coordinate points.

[0069] Optionally, an adaptive filtering algorithm is used to coarsely filter the initial coordinate point set to obtain a first-selected coordinate point set, including: extracting coordinate points from the initial coordinate point set based on height information to obtain an input coordinate point set; using a specified number of coordinate points with the smallest height information in the input coordinate point set as initial seed points; using the Random Sample Consensus (RANSAC) algorithm to estimate a plane model based on the initial seed points to obtain a filtering surface; and filtering the coordinate points located on the filtering surface to obtain the first-selected coordinate point set.

[0070] Optionally, the wake region is dynamically obtained based on the first set of selected coordinate points, including: detecting the working parameters associated with the first set of selected coordinate points through sensors, wherein the working parameters include speed, ship length, turning angle, maneuverability turning index and maneuverability following index; determining the dynamic wake length, dynamic wake width and dynamic wake curvature radius based on the working parameters; and dynamically determining the wake region based on the dynamic wake length, dynamic wake width and dynamic wake curvature radius.

[0071] Step S203: Based on the wake region, perform fine filtering on the first set of filtered coordinate points to obtain the second set of filtered coordinate points.

[0072] Optionally, a second set of selected coordinate points is obtained by finely filtering the first set of selected coordinate points based on the wake region, including: determining a first set of candidate coordinate points located within the wake region and a second set of candidate coordinate points located outside the wake region; determining a height difference threshold associated with the wake region and deleting coordinate points in the first set of candidate coordinate points whose height information is less than the height difference threshold to obtain a third set of candidate coordinate points; and combining the second set of candidate coordinate points and the third set of candidate coordinate points to obtain the second set of selected coordinate points.

[0073] Optionally, determining the height difference threshold associated with the wake region includes: obtaining a basic target height difference threshold, wherein the basic target height difference threshold dynamically changes with the water surface wave height; obtaining the velocity associated with the first set of selected coordinate points, and determining the height difference threshold based on the velocity and the basic target height difference threshold.

[0074] Step S204: Based on the physical parameters of the lidar, an adaptive clustering algorithm is used to cluster the second set of selected coordinate points to obtain the detection target.

[0075] Optionally, an adaptive clustering algorithm is used to cluster the second set of selected coordinate points based on the physical parameters of the lidar to obtain the detection target. This includes: obtaining the dynamic target distance based on the second set of selected coordinate points; using the density-based noise application spatial clustering algorithm DBSCAN to determine the dynamic neighborhood radius based on the lidar's physical parameters and the dynamic target distance; calculating the distance correlation density of the second set of selected coordinate points based on the lidar's maximum detection range and atmospheric attenuation coefficient; dynamically calculating the minimum number of points threshold based on the dynamic neighborhood radius and distance correlation density; clustering the second set of selected coordinate points based on the minimum number of points threshold to obtain clusters; and determining the detection target based on the clusters.

[0076] Step S205: When it is determined that the weather conditions meet the preset requirements, image information of the surrounding environment of the unmanned surface vessel is captured, and the detection target is corrected based on the image information.

[0077] Specifically, in this embodiment, after detecting a target using lidar, the unmanned surface vessel (USV) can automatically avoid obstacles based on the detected target. However, to ensure the safety of automatic navigation, the detected target is also corrected. In good weather conditions, such as when visibility exceeds 50 meters, images of the USV's surrounding environment can be captured using a camera. Since these images are valuable for reference, this embodiment uses them to verify whether there are any missed or false detections. If a missed detection is detected, the target is added based on the image information; if a false detection is detected, it is deleted, thus ensuring the accuracy of the detected targets displayed on the grid map. In this embodiment, under good weather conditions, the detection targets detected by the lidar are corrected using highly reliable images, thereby ensuring the accuracy of the final displayed detected targets.

[0078] It should be noted that in this embodiment, during the process of correcting the detected target using image information, the proportion of missed detections and false detections of the detected target is also obtained. When the proportion of missed detections and false detections is too large, it indicates that the lidar equipment may have malfunctioned or the software module may have malfunctioned. At this time, an alarm message will be generated and displayed on the human-machine interface, so as to promptly prompt the staff to inspect and maintain the lidar or software module to ensure the accuracy of maritime target detection.

[0079] The technical solution of this invention improves the comprehensiveness of lidar detection information by fusing point cloud data. By coarsely filtering and finely filtering the initial coordinate point set obtained after fusion, significant clutter and wake clutter in the target movement process are filtered out. The filtered results are then aggregated based on the physical parameters of the lidar to complete the detection of various target obstacles. Thus, lidar can be used to efficiently and in real time detect water surface target obstacles in complex marine environments, providing environmental perception support for the autonomous navigation of unmanned surface vessels.

[0080] Example 3

[0081] Figure 5 This is a schematic diagram of a marine target detection device provided in an embodiment of the present invention. Figure 5 As shown, the device includes: a fusion conversion module 310, a coarse filtration module 320, a fine filtration module 330, and a clustering module 340.

[0082] Among them, the fusion conversion module 310 is used to scan the surrounding environment of the unmanned surface vessel with lidar to obtain point cloud data, and to fuse and convert the point cloud data on the grid map to obtain an initial coordinate point set.

[0083] The coarse filtering module 320 is used to perform coarse filtering on the initial coordinate point set using an adaptive filtering algorithm to obtain the first filtered coordinate point set, and to dynamically obtain the wake region based on the first filtered coordinate point set.

[0084] The fine filtering module 330 is used to perform fine filtering on the first set of filtered coordinate points based on the wake region to obtain the second set of filtered coordinate points;

[0085] Clustering module 340 is used to cluster the second set of selected coordinate points based on the physical parameters of the lidar using an adaptive clustering algorithm to obtain the detection target.

[0086] Optionally, the fusion conversion module 310 is used to perform two-dimensional conversion on the point cloud data on a pre-created grid map to obtain coordinate points;

[0087] The converted multi-frame coordinate points are fused on the grid map to obtain fused coordinate points, and an initial coordinate point set is constructed based on the fused coordinate points. Each coordinate point in the initial coordinate point set is marked with height information.

[0088] Optionally, the coarse filtering module 320 includes a coarse filtering subunit, used to extract coordinate points from the initial coordinate point set based on the height information to obtain the input coordinate point set;

[0089] Use a specified number of coordinate points with the smallest height information from the input coordinate point set as the initial seed points;

[0090] The Random Sampling Consensus (RANSAC) algorithm is used to estimate the plane model based on the initial seed points to obtain the filter surface. The coordinate points located on the filter surface are then filtered to obtain the first set of filtered coordinate points.

[0091] Optionally, the coarse filtering module 320 includes a wake region acquisition subunit for detecting operating parameters associated with the first set of filtered coordinate points via sensors, wherein the operating parameters include speed, ship length, steering angle, maneuverability turning index, and maneuverability following index.

[0092] The dynamic wake length, dynamic wake width, and dynamic wake curvature radius are determined based on the operating parameters.

[0093] The wake region is dynamically determined based on the dynamic wake length, dynamic wake width, and dynamic wake curvature radius.

[0094] Optionally, the fine filtering module 330 is used to determine a first set of candidate coordinate points located within the wake region and a second set of candidate coordinate points located outside the wake region from the first set of filtered coordinate points.

[0095] Determine the height difference threshold associated with the wake region, and delete the coordinate points in the first candidate coordinate point set whose height information is less than the height difference threshold to obtain the third candidate coordinate point set;

[0096] The second set of selected coordinate points is obtained by combining the second and third candidate sets of coordinate points.

[0097] Optionally, the fine filtration module 330 is also used to obtain a basic target height difference threshold, wherein the basic target height difference threshold dynamically changes with the water surface wave height;

[0098] Obtain the velocity associated with the first set of filtered coordinate points, and determine the height difference threshold based on the velocity and the basic target height difference threshold.

[0099] Optionally, clustering module 340 is used to obtain the dynamic target distance based on the second set of selected coordinate points, and uses the density-based noise application spatial clustering algorithm DBSCAN to determine the dynamic neighborhood radius based on the physical parameters of the lidar and the dynamic target distance.

[0100] The distance correlation density of the second set of selected coordinate points is calculated based on the maximum detection range of the lidar and the atmospheric attenuation coefficient.

[0101] The minimum number of points threshold is obtained by dynamically calculating based on the dynamic neighborhood radius and distance-related density.

[0102] Clustering is performed on the second set of selected coordinate points based on the minimum number of points threshold to obtain clusters, and the detection target is determined based on the clusters.

[0103] The maritime target detection device provided in this embodiment of the invention can execute a maritime target detection method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0104] Example 4

[0105] Figure 6 A schematic diagram of a terminal device 10 that can be used to implement embodiments of the present invention is shown. The terminal device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The terminal device can also represent various forms of mobile devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0106] The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the invention described and / or claimed herein.

[0107] like Figure 6 As shown, the terminal device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer programs stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the terminal device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0108] Multiple components in terminal device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows terminal device 10 to exchange information / data with other terminal devices through computer networks such as the Internet and / or various telecommunications networks.

[0109] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as maritime target detection methods.

[0110] In some embodiments, the maritime target detection method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on terminal device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the maritime target detection method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the maritime target detection method by any other suitable means (e.g., by means of firmware).

[0111] Various embodiments of the apparatuses and techniques described above herein can be implemented in digital electronic circuit devices, integrated circuit devices, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), device-on-a-chip (SoCs), complex programmable logic terminal devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable device including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage device, at least one input device, and at least one output device, and transmitting data and instructions to the storage device, the at least one input device, and the at least one output device.

[0112] Computer programs used to implement the maritime target detection method of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to the processor of a general-purpose computer, a special-purpose computer, or other operational non-stop data migration device, such that when executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer programs can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0113] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution apparatus, device, or terminal device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage terminal devices, magnetic storage terminal devices, or any suitable combination thereof.

[0114] To provide interaction with a user, the apparatus and techniques described herein can be implemented on a terminal device having: a display device (e.g., a touchscreen) for displaying information to the user; and buttons through which the user can provide input to the terminal device. Other types of apparatus can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or haptic feedback); and input from the user can be received in any form (including voice input, speech input, or haptic input).

[0115] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0116] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method of detecting a marine target, characterized in that, The method comprises: The unmanned ship peripheral environment is scanned by laser radar to obtain point cloud data, and the point cloud data is fused and converted on a grid map to obtain an initial coordinate point set; An adaptive filtering algorithm is used to coarsely filter the initial coordinate point set to obtain a first screening coordinate point set, and a wake region is dynamically obtained according to the first screening coordinate point set; The first screening coordinate point set is finely filtered based on the wake region to obtain a second screening coordinate point set; An adaptive clustering algorithm is used to cluster the second screening coordinate point set based on the physical parameters of the laser radar to obtain a detection target.

2. The method of claim 1, wherein, The point cloud data is converted on a pre-created grid map to obtain coordinate points; The converted multiple frames of coordinate points are fused on the grid map to obtain fused coordinate points, and the initial coordinate point set is constructed according to the fused coordinate points, wherein each coordinate point in the initial coordinate point set is marked with height information. The adaptive filtering algorithm is used to coarsely filter the initial coordinate point set to obtain a first screening coordinate point set, which comprises:

3. The method of claim 2, wherein, According to the height information, coordinate points are extracted from the initial coordinate point set to obtain an input coordinate point set; The specified number of coordinate points with the smallest height information in the input coordinate point set are taken as initial seed points; A random sample consensus (RANSAC) algorithm is used to estimate a plane model according to the initial seed points to obtain a filtering surface, and the coordinate points located on the filtering surface are filtered to obtain the first screening coordinate point set. The wake region is dynamically obtained according to the first screening coordinate point set, which comprises:

4. The method of claim 1, wherein, The working parameters associated with the first screening coordinate point set are detected by a sensor, wherein the working parameters include speed, ship length, turning angle, maneuvering turning index and maneuvering following index; The dynamic wake length, dynamic wake width and dynamic wake radius of curvature are determined according to the working parameters; The wake region is dynamically determined according to the dynamic wake length, dynamic wake width and dynamic wake radius of curvature. The first screening coordinate point set is finely filtered based on the wake region to obtain a second screening coordinate point set, which comprises:

5. The method of claim 1, wherein, The first candidate coordinate point set located in the wake region and the second candidate coordinate point set located outside the wake region in the first screening coordinate point set are determined; The height difference threshold associated with the wake region is determined, and the coordinate points with height information less than the height difference threshold in the first candidate coordinate point set are deleted to obtain a third candidate coordinate point set; The second screening coordinate point set is obtained by combining the second candidate coordinate point set and the third candidate coordinate point set. The height difference threshold associated with the wake region is determined, which comprises:

6. The method of claim 5, wherein, A basic target height difference threshold is obtained, wherein the basic target height difference threshold dynamically changes with the water surface wave height; The speed associated with the first screening coordinate point set is obtained, and the height difference threshold is determined according to the speed and the basic target height difference threshold. ​ 7. The method of claim 1, wherein, The adaptive clustering algorithm based on the physical parameters of the laser radar is used to cluster the second screening coordinate point set to obtain a detection target, including: A dynamic target distance is obtained according to the second screening coordinate point set, a dynamic neighborhood radius is determined according to the physical parameters of the laser radar and the dynamic target distance by using a density-based noise application spatial clustering algorithm DBSCAN; The distance-related density of the second screening coordinate point set is calculated according to the maximum detection distance of the laser radar and the atmospheric attenuation coefficient; The minimum point number threshold is obtained by dynamically calculating according to the dynamic neighborhood radius and the distance-related density; The second screening coordinate point set is clustered based on the minimum point number threshold to obtain a clustering cluster, and a detection target is determined according to the clustering cluster.

8. A target detection apparatus characterized by comprising: The device comprises: A fusion conversion module is configured to scan the surrounding environment of the unmanned ship by the laser radar to obtain point cloud data, and to fuse and convert the point cloud data on a grid map to obtain an initial coordinate point set; A rough filtering module is configured to perform rough filtering on the initial coordinate point set by using an adaptive filtering algorithm to obtain a first screening coordinate point set, and to dynamically obtain a wake region according to the first screening coordinate point set; An accurate filtering module is configured to perform accurate filtering on the first screening coordinate point set based on the wake region to obtain a second screening coordinate point set; A clustering module is configured to use an adaptive clustering algorithm based on the physical parameters of the laser radar to cluster the second screening coordinate point set to obtain a detection target.

9. A terminal device, comprising: The terminal device comprises: One or more processors; A storage device configured to store one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method of any one of claims 1-7.

10. A storage medium of computer executable instructions, on which a computer program is stored, characterized in that, The program is executed by the processor to implement the method of any one of claims 1-7.

Images