CN115877404A - Point cloud data preprocessing method and device and water surface target laser radar - Google Patents
Point cloud data preprocessing method and device and water surface target laser radar Download PDFInfo
- Publication number
- CN115877404A CN115877404A CN202211298880.6A CN202211298880A CN115877404A CN 115877404 A CN115877404 A CN 115877404A CN 202211298880 A CN202211298880 A CN 202211298880A CN 115877404 A CN115877404 A CN 115877404A
- Authority
- CN
- China
- Prior art keywords
- point cloud
- point
- target
- cloud data
- water surface
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/30—Assessment of water resources
Landscapes
- Optical Radar Systems And Details Thereof (AREA)
Abstract
A point cloud data preprocessing method and device and a water surface target laser radar relate to the field of intelligent unmanned intelligent ships. Aiming at the problems that the existing simple European style clustering method is insufficient in ship tail wave treatment, can cause misjudgment on the position and the volume of a target ship and influence the judgment of an unmanned ship on the real attribute of the target, the invention provides the following technical scheme: the point cloud data preprocessing method is used for removing noise points of water surface point cloud data, and comprises the following steps: step 1: selecting a point in a target water surface space, and obtaining a point cloud according to the point; step 2: obtaining a laser point set with a distance meeting a preset threshold according to the point cloud; and step 3: judging whether the number of laser points in the laser point set meets a preset requirement, if so, keeping the set as a target point cloud cluster, and performing step 4; if not, returning to repeat the step 1 to the step 3; and 4, step 4: and acquiring target water surface information according to the target point cloud cluster. The method is suitable for being applied to a point cloud processing technology and applied to a shipborne water surface laser radar sensor.
Description
Technical Field
Relate to unmanned wisdom ship field of intelligence, concretely relates to shipborne laser radar.
Background
The unmanned surface vessel is an unmanned surface vessel, is a small unmanned ship sailing on the water surface in an autonomous or remote control mode, can be carried by a large ship to a preset place and then released to execute tasks, can also sail to offshore targets autonomously, is mainly used for executing dangerous tasks and tasks which are not suitable for being executed by the manned ship, and can execute various war and non-war military tasks once an advanced control system, a sensor system, a communication system and a weapon system are equipped;
in the process of driving the unmanned ship on the water surface, some obstacles on the water surface need to be subjected to obstacle avoidance treatment, so that the damage of the unmanned ship caused by collision between the unmanned ship and the obstacles is avoided;
aiming at the problem of obstacle avoidance of unmanned surface vehicles, scholars at home and abroad have already carried out some researches. Before obstacle avoidance, a method for detecting obstacles is of great importance, and the current mainstream obstacle detection technologies are mainly divided into the following four types: the system comprises an ultrasonic obstacle detection technology, a stereoscopic vision obstacle detection technology, a multi-sensing fusion obstacle detection technology and a laser radar obstacle detection technology. The ultrasonic obstacle detection technology mainly utilizes the echo characteristics of sound waves to calculate the position of an obstacle, and due to the fact that the Doppler effect exists in the sound waves, if the driving speed reaches a certain degree, data are abnormal. The stereoscopic vision obstacle detection technology mainly expands a planar two-dimensional vision technology to a three-dimensional space through mathematical principle calculation in image processing to obtain more target object information, but the cost and the complexity of information processing are increased. The multi-sensing fusion obstacle detection technology mainly carries out comprehensive processing, optimization and combination on two or more same-class or heterogeneous sensor information so as to obtain more environmental characteristic expressions, but the difficulty lies in that a fusion algorithm is difficult to select a proper fusion algorithm which can be matched with a plurality of same or different sensing devices in practice. The laser radar obstacle detection technology mainly utilizes the time difference between a transmitting laser line and a reflecting laser line to obtain the angle and the distance value of an obstacle point through calculation, and the detection system has the advantages of high precision, strong resolution, good real-time performance and the like.
Laser radar sensors are widely applied to the field of unmanned boats, and point cloud processing technology is becoming more mature. However, in the actual working environment of the unmanned surface vehicle, there are various interference factors such as ship tail waves, waterweeds and fishing nets, and mutual interference among various sensors such as interaction between the marine radar and the laser radar. Firstly, the interference factors can cause scattered noise points to exist in raw data of the laser radar, and if the raw data is not preprocessed, the noise points can bring great influence on multi-sensor information fusion and subsequent target detection and tracking. Secondly, the simple European-style clustering method is insufficient in ship tail wave treatment, so that misjudgment of the position and the volume of the target ship can be caused, and the judgment of the real attribute of the target by the unmanned ship is influenced.
Disclosure of Invention
Aiming at the problems that the existing simple European style clustering method is insufficient in ship tail wave treatment, can cause misjudgment on the position and the volume of a target ship and influence the judgment of an unmanned ship on the real attribute of the target, the invention provides the following technical scheme:
the point cloud data preprocessing method is used for removing noise points of water surface point cloud data, and comprises the following steps:
step 1: selecting a point in a target water surface space, and obtaining a point cloud according to the point;
step 2: obtaining a laser point set with a distance meeting a preset threshold according to the point cloud;
and step 3: judging whether the number of laser points in the laser point set meets a preset requirement, if so, keeping the set as a target point cloud cluster, and performing step 4; if not, returning to repeat the step 1 to the step 3;
and 4, step 4: and acquiring the target water surface information according to the target point cloud cluster.
Further, there is provided a preferred embodiment, the method further comprising:
and 5: and repeating the steps 1 to 4 until the preset repetition times are reached.
Further, a preferred embodiment is provided, and in the step 1, the manner of obtaining the point cloud specifically is: and completing the neighbor search of the point by using a KD-Tree data index structure to obtain the point cloud.
Further, a preferred embodiment is provided, and the manner of obtaining the set in step 2 specifically is: and obtaining the weighted distance according to the point cloud.
Based on the same inventive concept, the invention also provides a point cloud data preprocessing device for removing the noise point of the point cloud data on the water surface, and the device comprises:
module 1: the system is used for selecting a point in a target water surface space and obtaining a point cloud according to the point;
and (3) module 2: the laser point set is used for obtaining a laser point set with a distance meeting a preset threshold value according to the point cloud;
and a module 3: the system is used for judging whether the number of the laser points in the laser point set meets a preset requirement or not, if so, the set is reserved as a target point cloud cluster, and the function of a module 4 is performed; if not, returning to execute the functions of the modules 1 to 3;
and (4) module: and the system is used for acquiring the target water surface information according to the target point cloud cluster.
Further, there is provided in a preferred embodiment that the apparatus further comprises:
and a module 5: and the function of the modules 1 to 4 is repeated until the preset repetition times are reached.
Further, a preferred embodiment is provided, and in the module 1, the manner of obtaining the point cloud specifically is: and finishing the neighbor search of the point by using a KD-Tree data index structure to obtain the point cloud.
Based on the same inventive concept, the invention also provides a water surface target laser radar, and the point cloud data preprocessing method is loaded in a central processing unit of the radar.
Based on the same inventive concept, the invention also provides a computer storage medium for storing a computer program, wherein when the computer program stored in the storage medium is read by a processor of a computer, the computer executes the point cloud data preprocessing method.
Based on the same inventive concept, the invention also provides a computer, which comprises a processor and a storage medium, wherein the storage medium stores a computer program, and when the processor reads the computer program stored in the storage medium and is read by the processor of the computer, the computer executes the point cloud data preprocessing method.
Compared with the prior art, the technical scheme provided by the invention has the advantages that:
according to the point cloud data preprocessing method provided by the invention, point cloud data are classified through the weighted Euclidean distance, and noise point removal and point cloud clustering are completed based on a KD-Tree structure, so that the point cloud data preprocessing method has the advantages of being more suitable for a specific scene of a water surface target identification task and enabling a processing result to be faster and more accurate.
According to the point cloud data preprocessing method, the characteristics of scattered and sparse ship tail waves and low reflectivity are utilized, the weighted Euclidean distance is calculated based on the point cloud intensity, and the KD-Tree data index structure is used for accelerating the noise point removal and point cloud clustering of the laser radar point cloud data, so that the rapidness and the accuracy of the detection of the water surface target are improved.
The point cloud data preprocessing method provided by the invention adopts an improved weighted Euclidean clustering method, and the improved clustering method can further screen the point cloud clusters obtained after intensity weighted clustering and also can effectively inhibit scattered and stray noise points in the original data of the laser radar.
According to the point cloud data preprocessing method provided by the invention, scattered and outlier noise points in the laser radar original data can be effectively inhibited through further screening the point cloud clusters obtained after intensity weighted clustering, the result obtained by the method is accurate, the real-time performance is strong, and the requirement of point cloud data preprocessing in the multi-sensor data fusion process can be met.
According to the water surface target laser radar, point cloud data are classified through the weighted Euclidean distance, noise point removal and point cloud clustering are completed based on the KD-Tree structure, the processing result is quicker and more accurate, and the rapidity and the accuracy of water surface target detection are improved.
The method is suitable for being applied to a point cloud processing technology and applied to a shipborne water surface laser radar sensor.
Drawings
FIG. 1 is a typical surface target lidar cloud point diagram according to an eleventh embodiment;
wherein (a) represents an actual shot picture and (b) represents a lidar cloud point map;
FIG. 2 is a comparison of the improvement effect of weighted clustering according to the eleventh embodiment;
wherein (a) represents a target bounding box obtained by a classical Euclidean clustering method, and (b) represents a target bounding box obtained by a weighted clustering method;
FIG. 3 is a schematic diagram of a KD-Tree search flow mentioned in the eleventh embodiment;
fig. 4 is a schematic diagram of a weighted euclidean clustering process according to an eleventh embodiment.
Detailed Description
In order to make the advantages and benefits of the technical solutions provided by the present invention more clear, the technical solutions provided by the present invention will be further described in detail with reference to the accompanying drawings, specifically:
the first embodiment provides a point cloud data preprocessing method for removing noise of water surface point cloud data, and the method comprises the following steps:
step 1: selecting a point in a target water surface space, and obtaining a point cloud according to the point;
step 2: obtaining a laser point set with a distance meeting a preset threshold according to the point cloud;
and step 3: judging whether the number of laser points in the laser point set meets a preset requirement, if so, keeping the set as a target point cloud cluster, and performing step 4; if not, returning to repeat the step 1 to the step 3;
and 4, step 4: and acquiring the target water surface information according to the target point cloud cluster.
In a second embodiment, the method for preprocessing point cloud data provided in the first embodiment is further limited, and the method further includes:
and 5: and repeating the steps 1 to 4 until the preset repetition times are reached.
In a third embodiment, the method for preprocessing point cloud data provided in the first embodiment is further limited, and in the step 1, the manner of obtaining the point cloud is specifically as follows: and finishing the neighbor search of the point by using a KD-Tree data index structure to obtain the point cloud.
In a fourth embodiment, the point cloud data preprocessing method provided in the first embodiment is further limited, and in the step 2, the manner of obtaining the set specifically includes: and obtaining the weighted distance according to the point cloud.
Fifth, the present embodiment provides a point cloud data preprocessing apparatus for removing noise in point cloud data on a water surface, the apparatus including:
module 1: the system is used for selecting a point in a target water surface space and obtaining a point cloud according to the point;
and (3) module 2: the laser point set is used for obtaining a laser point set with a distance meeting a preset threshold value according to the point cloud;
and a module 3: the system is used for judging whether the number of the laser points in the laser point set meets a preset requirement or not, if so, the set is reserved as a target point cloud cluster, and the function of a module 4 is performed; if not, returning to execute the functions of the modules 1 to 3;
and (4) module: and the system is used for acquiring the target water surface information according to the target point cloud cluster.
Sixth, the present embodiment is further limited to the point cloud data preprocessing device according to fifth, further including:
and a module 5: and the function of the modules 1 to 4 is repeated until the preset repetition times are reached.
Seventh, the present embodiment is further limited to the point cloud data preprocessing apparatus provided in the fifth embodiment, and the manner of obtaining the point cloud in the module 1 is specifically as follows: and completing the neighbor search of the point by using a KD-Tree data index structure to obtain the point cloud.
The eighth embodiment provides a water surface target laser radar, and a point cloud data preprocessing method provided in any one of the first to fourth embodiments is loaded in a central processing unit of the radar.
Ninth embodiment provides a computer storage medium storing a computer program, wherein when the computer program stored in the storage medium is read by a processor of a computer, the computer executes the point cloud data preprocessing method provided in any one of the first to fourth embodiments.
Tenth embodiment the present embodiment provides a computer, comprising
The point cloud data preprocessing method comprises a processor and a storage medium, wherein a computer program is stored in the storage medium, and when the processor reads the computer program stored in the storage medium and is read by a processor of a computer, the computer executes the point cloud data preprocessing method provided by any one of the first to fourth embodiments.
In an eleventh embodiment, the present embodiment is described with reference to fig. 1 and 2, and the present embodiment provides a specific example for the point cloud data preprocessing method provided in any one of the first to fourth embodiments, so as to prove the advantages of the point cloud data preprocessing method provided in any one of the first to fourth embodiments compared with the prior art, and at the same time, is used to explain the point cloud data preprocessing method provided in any one of the first to fourth embodiments, specifically:
the embodiment provides an improved laser radar point cloud data preprocessing method based on a KD-Tree data index structure. In the working environment of the unmanned surface vehicle, the interference of the tail wave of the vehicle often exists, the tail wave of the vehicle can be identified as sparse scattered noise points with lower reflectivity by the laser radar, as shown in fig. 1 (b), the continuous point cloud at the tail of the unmanned surface vehicle in the figure is the tail wave point cloud, and the tail wave point cloud is low in intensity and is displayed as dark red during visualization; according to the method, the characteristics of scattered and sparse ship tail waves and low reflectivity are utilized, the weighted Euclidean distance is calculated based on the point cloud intensity, the noise point removal and the point cloud clustering of the laser radar point cloud data are accelerated through the KD-Tree data index structure, and the rapidity and the accuracy of the detection of the water surface target are improved. The specific implementation method is as follows.
Firstly, selecting a certain point P in the space, and accelerating to complete neighbor search of the point by using a KD-Tree data index structure;
calculating the weighted distance of the point cloud based on the point cloud intensity to obtain a laser point set with the distance meeting the threshold setting; euclidean clustering, i.e., a method for clustering point clouds based on euclidean distances between the point clouds, which is the most common distance metric method, is the euclidean distance metric between the point clouds, and the coordinate of the euclidean distance metric is (x) under the laser radar point cloud three-dimensional coordinate system p ,y p ,z p ) Point P of (a) and coordinates of (x) q ,y q ,z q ) Distance d between Q points PQ Is as follows
On the basis, a classical Euclidean clustering method is optimized, two factors of point cloud intensity and Euclidean distance are comprehensively considered, and a weighted distance calculation formula of the point cloud is as follows:
in the formula: d intensity Representing the weighted Euclidean distance (m), d represents the Euclidean distance (m), and intensity represents the point cloud intensity;
FIG. 2 is a comparison graph of the improvement effect of weighted clustering;
as shown in fig. 2, by using the improved weighted euclidean clustering method, the improved clustering method can effectively suppress scattered outlier noise in the raw data of the laser radar by further screening the point cloud clusters obtained after the intensity weighted clustering.
And judging the number of the laser point sets obtained by calculation, and if the number requirement is met, keeping the set as a target point cloud cluster. If not, removing the set and reselecting the initial point;
and analyzing the obtained detection result, and processing the point cloud data of the next frame. The Euclidean clustering method based on the point cloud reflectivity can well inhibit the interference of ship tail waves on clustering results, and can also effectively inhibit scattered and stray noise points in laser radar original data by further screening point cloud clusters obtained after intensity weighted clustering.
Wherein, FIG. 3 is a schematic diagram of a KD-Tree search flow; for a given point P to be queried, comparison needs to be performed from a root node of the KD-Tree, and a value P (k) corresponding to the query point P on a current node division dimension k is compared with a threshold value m. If P (k) < m, accessing the left sub-tree, otherwise accessing the right sub-tree until reaching a leaf node Q, wherein Q is the nearest neighbor point of P, and the minimum distance D is the distance between Q and P. Then, returning to the root node along the original search path, and in the process, if a child node meeting the requirement that the distance between the child node and the P is smaller than D is included in the search range, and updating the nearest neighbor point; and if no other remaining points exist, namely all the search paths are empty, ending the KT-Tree search process.
FIG. 4 is a flow chart of weighted Euclidean clustering; aiming at a certain point P in the space, a KD-Tree neighbor search algorithm is used for finding a neighbor point and calculating the weighted distance of the point cloud as a distance judgment standard, and the neighbor point smaller than a set threshold is divided into an output set Q. Judging the number of elements in Q, if the number of elements in Q continuously increases, selecting other laser points except P in Q, and repeating the process; if the elements in Q are not increased any more and the number of elements meets the threshold requirement, the set is retained as the output result, otherwise the set is removed. The above process is repeated until there are no more remaining laser spots in space.
The technical solutions provided by the present invention are further described in detail through several specific embodiments in order to highlight the advantages and benefits of the technical solutions provided by the present invention, but the several specific embodiments described above are only used for describing the technical solutions provided by the present invention, and are not used as limitations of the present invention; any reasonable modification and improvement, reasonable combination of embodiments, and equivalent substitution, etc. of the present invention, which are based on the spirit and principle of the present invention, should be included in the protection scope of the present invention.
Claims (10)
1. The point cloud data preprocessing method is used for removing noise points of water surface point cloud data, and is characterized by comprising the following steps of:
step 1: selecting a point in a target water surface space, and obtaining a point cloud according to the point;
step 2: obtaining a laser point set with a distance meeting a preset threshold according to the point cloud;
and 3, step 3: judging whether the number of laser points in the laser point set meets a preset requirement, if so, keeping the set as a target point cloud cluster, and performing step 4; if not, returning to repeat the step 1 to the step 3;
and 4, step 4: and acquiring the target water surface information according to the target point cloud cluster.
2. The method of point cloud data preprocessing of claim 1, the method further comprising:
and 5: and repeating the steps 1 to 4 until the preset repetition times are reached.
3. The point cloud data preprocessing method according to claim 1, wherein in the step 1, the manner of obtaining the point cloud is specifically: and finishing the neighbor search of the point by using a KD-Tree data index structure to obtain the point cloud.
4. The point cloud data preprocessing method of claim 1, wherein in the step 2, the manner of obtaining the set is specifically: and obtaining the weighted distance according to the point cloud.
5. Point cloud data preprocessing device for surface of water point cloud data noise removes, its characterized in that, the device includes:
module 1: the method comprises the steps of selecting a point in a target water surface space, and obtaining a point cloud according to the point;
and (3) module 2: the laser point set is used for obtaining a laser point set with a distance meeting a preset threshold value according to the point cloud;
and a module 3: the system is used for judging whether the number of the laser points in the laser point set meets a preset requirement, if so, the set is reserved as a target point cloud cluster, and the function of a module 4 is performed; if not, returning to execute the functions of the modules 1 to 3;
and (4) module: and the system is used for acquiring the target water surface information according to the target point cloud cluster.
6. The point cloud data preprocessing apparatus of claim 1, wherein the apparatus further comprises:
and a module 5: for repeating the functions of the modules 1 to 4 until a preset number of repetitions is reached.
7. The point cloud data preprocessing device according to claim 1, wherein in the module 1, the manner of obtaining the point cloud is specifically: and finishing the neighbor search of the point by using a KD-Tree data index structure to obtain the point cloud.
8. The surface target laser radar is characterized in that the point cloud data preprocessing method of any one of claims 1-4 is loaded in a central processing unit of the radar.
9. Computer storage medium for storing a computer program, wherein the computer executes the point cloud data preprocessing method according to any one of claims 1 to 4 when the computer program stored in the storage medium is read by a processor of the computer.
10. A computer comprising a processor and a storage medium, wherein the storage medium stores a computer program, wherein when the processor reads the computer program stored in the storage medium and the computer program is read by the processor of the computer, the computer performs the point cloud data preprocessing method according to any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211298880.6A CN115877404A (en) | 2022-10-24 | 2022-10-24 | Point cloud data preprocessing method and device and water surface target laser radar |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211298880.6A CN115877404A (en) | 2022-10-24 | 2022-10-24 | Point cloud data preprocessing method and device and water surface target laser radar |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115877404A true CN115877404A (en) | 2023-03-31 |
Family
ID=85758844
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211298880.6A Pending CN115877404A (en) | 2022-10-24 | 2022-10-24 | Point cloud data preprocessing method and device and water surface target laser radar |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115877404A (en) |
-
2022
- 2022-10-24 CN CN202211298880.6A patent/CN115877404A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111239766B (en) | Water surface multi-target rapid identification and tracking method based on laser radar | |
| US20220130156A1 (en) | Three-dimensional object detection and intelligent driving | |
| KR20220155559A (en) | Autonomous navigation method using image segmentation | |
| CN112464994B (en) | PointNet network-based boat tail wave recognition and removal method | |
| CN108549084B (en) | Target detection and attitude estimation method based on sparse two-dimensional laser radar | |
| CN109283538A (en) | A kind of naval target size detection method of view-based access control model and laser sensor data fusion | |
| CN108765458A (en) | High sea situation unmanned boat sea-surface target dimension self-adaption tracking based on correlation filtering | |
| CN108830879A (en) | A kind of unmanned boat sea correlation filtering method for tracking target suitable for blocking scene | |
| CN112394726B (en) | Unmanned ship obstacle fusion detection method based on evidence theory | |
| CN113031004B (en) | Unmanned ship water surface target detection and path planning method based on three-dimensional laser radar | |
| KR102466804B1 (en) | Autonomous navigation method using image segmentation | |
| CN111913177A (en) | Method and device for detecting target object and storage medium | |
| Clunie et al. | Development of a perception system for an autonomous surface vehicle using monocular camera, lidar, and marine radar | |
| CN111736167B (en) | Method and device for obtaining laser point cloud density | |
| CN111929653A (en) | Target detection and tracking method and system based on unmanned ship marine radar | |
| CN116467848B (en) | Millimeter wave radar point cloud simulation method and device | |
| Mu et al. | Surface navigation target detection and recognition based on SSD | |
| CN115877404A (en) | Point cloud data preprocessing method and device and water surface target laser radar | |
| CN115267827A (en) | Laser radar harbor area obstacle sensing method based on height density screening | |
| CN112651405B (en) | Target detection method and device | |
| US11874407B2 (en) | Technologies for dynamic, real-time, four-dimensional volumetric multi-object underwater scene segmentation | |
| CN113888589A (en) | Water surface obstacle detection and multi-target tracking method based on laser radar | |
| CN112766100A (en) | 3D target detection method based on key points | |
| Perera et al. | A gm-phd filter with estimation of probability of detection and survival for individual targets | |
| EP4036595A1 (en) | Object detection with multiple ranges and resolutions |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |