CN114002708A - Tail wave filtering method for unmanned ship application - Google Patents
Tail wave filtering method for unmanned ship application Download PDFInfo
- Publication number
- CN114002708A CN114002708A CN202111220246.6A CN202111220246A CN114002708A CN 114002708 A CN114002708 A CN 114002708A CN 202111220246 A CN202111220246 A CN 202111220246A CN 114002708 A CN114002708 A CN 114002708A
- Authority
- CN
- China
- Prior art keywords
- point cloud
- point
- tail wave
- height difference
- cloud data
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 50
- 238000001914 filtration Methods 0.000 title claims abstract description 47
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 38
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000012216 screening Methods 0.000 description 14
- 230000008447 perception Effects 0.000 description 9
- 230000011218 segmentation Effects 0.000 description 7
- 230000008859 change Effects 0.000 description 6
- 230000000694 effects Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 4
- 238000001514 detection method Methods 0.000 description 3
- 238000003384 imaging method Methods 0.000 description 3
- 230000004888 barrier function Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
- G01S17/88—Lidar systems specially adapted for specific applications
- G01S17/93—Lidar systems specially adapted for specific applications for anti-collision purposes
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/48—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00
- G01S7/4802—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Networks & Wireless Communication (AREA)
- General Physics & Mathematics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Electromagnetism (AREA)
- Radar Systems Or Details Thereof (AREA)
- Optical Radar Systems And Details Thereof (AREA)
Abstract
The invention discloses a tail wave filtering method for unmanned ship application, and belongs to the field of unmanned ship water surface environment sensing and understanding. The method comprises the following steps: s1, point cloud data of a water surface environment around the unmanned ship are obtained, and the point cloud data are obtained through scanning of a multi-line rotary laser radar; s2, removing clutter noise from the point cloud data; s3, identifying tail wave points of the filtered point cloud data, including: s31, calculating the actual height difference of two adjacent laser reflection points in the vertical direction in the point cloud data; s32, calculating the ratio of the actual height difference to the theoretical height difference of each point cloud data point to serve as a normalized height difference; s33, identifying the point with the normalized height difference lower than a first threshold value as a tail wave point; and S4, removing the tail wave points from the point cloud data. The invention provides a method for identifying tail wave point clouds by taking the ratio of the actual height difference and the theoretical height difference of each point cloud data point as a normalized height difference characteristic, which can accurately identify the tail wave point clouds, ensure that all tail wave points are found, and further simply and efficiently remove the tail wave point clouds.
Description
Technical Field
The invention belongs to the field of unmanned ship water surface environment sensing and understanding, and particularly relates to a tail wave filtering method for unmanned ship application.
Background
The laser radar is one of the most important environment perception sensors for autonomous obstacle avoidance navigation of the unmanned ship. The laser radar periodically scans the surrounding water surface environment to acquire the azimuth and geometric shape information of the obstacle target, and the unmanned ship identifies and positions the obstacles in the surrounding environment through an environment perception algorithm, so that autonomous obstacle avoidance navigation is completed. However, when a ship sails on the water surface, the generated tail wave can reflect laser to generate echo point cloud in the field of view of the laser radar, and the echo point cloud is often misjudged as a water surface obstacle by an environment perception algorithm to cause a target false alarm; moreover, when an obstacle exists in the tail wave region, the tail wave point cloud interferes with the environment perception algorithm, and the target misjudgment is caused. Therefore, a certain algorithm is required to filter the tail wave point cloud.
Tail wave filtering is always a technical difficulty in the field of unmanned boat water surface environment sensing and understanding, and no better solution is provided at present. In actual unmanned boat applications, often only a few simple and inefficient processes can be performed on the wake cloud. For example, the point cloud within the fixed sector of the stern is directly subtracted based on the prior generation of stern waves from the stern. Obviously, as the size of the tail wave area changes in real time along with the change of the ship speed and the sea condition, the sector range is difficult to determine, so that the method can be only used in the scenes of low ship speed and low sea condition; and the method can reduce the perception capability of the unmanned boat to the close-range target, and has larger application limitation. Meanwhile, because the method cannot completely filter the tail wave point cloud, certain problems of target false alarm and target misjudgment still exist, and time-consuming target judgment and false alarm removal processing still needs to be carried out in subsequent steps. Therefore, a simple and efficient method for filtering the tail waves is needed.
The water surface scene point cloud can be regarded as comprising three parts of a barrier target point cloud, a tail wave point cloud and a clutter noise point. Clutter noise points are usually expressed as 'outliers', can be well removed by using a filter method, and at the moment, the tail wave filtering problem can be modeled as a segmentation problem of target point clouds and tail wave point clouds, namely a two-classification problem of the target point and the tail wave point. Therefore, the key to solve the problem of tail wave filtering is to find out the characteristics of strong classification capability on target points and tail wave points, high scene adaptability, namely strong accuracy and good robustness. Currently, there are few related studies specifically aiming at tail wave filtering. One method is based on the assumption that the tail wave point cloud is in a position with relatively low height in the point cloud of the whole water surface scene, and the point cloud with relatively low height in the whole scene is directly removed by utilizing a height threshold screening method; this may cause that part of the point cloud in the target close to the water surface is also removed, that is, "over-segmentation", resulting in "damage" of the target, thereby affecting subsequent detection and identification of the target, and in a high sea state, the height range covered by the tail wave point cloud is usually overlapped with the water surface target to some extent, resulting in that the above assumption is no longer satisfied, and the algorithm is invalid. One method is based on the characteristic that the height difference of the local area of the tail wave point cloud is small, and the height difference of the local area of the target point cloud is usually large, voxel filtering or grid filtering is carried out on the whole scene point cloud, the height difference of a voxel grid or a point cloud cluster in a grid is calculated, and the voxel grid or the point cloud cluster in the grid with the height difference lower than a certain threshold value is removed; similarly, a simple threshold strategy is difficult to adapt to various sea conditions, and the method usually leaves partial tail wave point clouds, namely 'under-segmentation', and also 'over-segmentation', removes partial point clouds with small local height difference in the target, and affects subsequent target detection and identification, so the method has the same poor effect. The existing methods can not find out the characteristics of strong classification capability on target points and tail wave points and high scene adaptability, so that the filtering effect of partial tail waves can be realized only in a specific scene, the problems of under-segmentation or over-segmentation occur, and the problem of tail wave filtering can not be solved well.
Disclosure of Invention
Aiming at the defects and improvement requirements of the prior art, the invention provides a tail wave filtering method for unmanned ship application, aiming at simply and efficiently filtering tail wave point cloud in a water surface scene by constructing the characteristics of strong classification capability on target points and tail wave points and high scene adaptability; the problems of 'under segmentation' and 'over segmentation' in the conventional tail wave filtering method are solved; and time-consuming target discrimination and false alarm removal processing in a subsequent environment perception algorithm is avoided.
To achieve the above object, according to a first aspect of the present invention, there is provided a method for filtering tail waves for an unmanned surface vehicle, the method including:
s1, point cloud data of a water surface environment around an unmanned ship are obtained, and the point cloud data are obtained through scanning of a multi-line rotary laser radar;
s2, removing clutter noise from the point cloud data;
s3, tail wave point identification is carried out on the point cloud data after clutter noise points are filtered, and the method comprises the following steps:
s31, calculating the actual height difference of two adjacent laser reflection points in the vertical direction in the point cloud data according to plane geometry;
s32, calculating the ratio of the actual height difference to the theoretical height difference of each point cloud data point to serve as a normalized height difference;
s33, identifying the point with the normalized height difference lower than the first threshold as a tail wave point, and adding a tail wave point set;
and S4, removing the identified tail wave point set from the point cloud data of the water surface environment.
Preferably, after step S33, the method further includes:
s34, calculating the absolute height of each point in the Z-axis direction according to plane geometry;
s35, removing points with absolute height larger than or equal to a second threshold value in the tail wave point set.
Preferably, absolute height
The calculation formula is as follows:
wherein d isn,
Respectively representing the target distance and the vertical emission angle obtained by the nth laser beam.
Has the advantages that: aiming at the problem of 'over-segmentation' which is easy to occur in tail wave filtering (namely, a part of a point cloud of a water surface obstacle target is classified into a tail wave point cloud), the method is based on the prior knowledge that the obstacle target is usually far higher than a tail wave plane, and based on the absolute height of the point cloud in the Z-axis direction, the part with the absolute height characteristic value larger than or equal to a second threshold value is removed from the tail wave point cloud set determined in the step S33, and the rest is a real tail wave point set. Because the target point cloud (namely the water surface obstacle point cloud) which is wrongly classified as the tail wave point is removed, the accuracy of tail wave point cloud identification is ensured on the premise of not reducing the recall ratio, and the problem of over-segmentation in tail wave filtering is avoided.
Preferably, the second threshold is determined by: and maximizing the variance of the absolute height between the target point and the tail wave point, thereby determining a second threshold value.
Has the advantages that: the absolute height characteristic values of the tail wave point cloud and the target point cloud are distributed in respective certain value ranges, the value ranges are continuously changed due to the change of conditions such as target types and sea conditions, and a simple threshold strategy is usually difficult to deal with. According to the invention, a corresponding screening threshold value is calculated for each frame of water surface scene point cloud in a self-adaptive manner by maximizing the variance of the absolute heights between a target point and a tail wave point, so that robust adaptation can be realized for various condition changes, and accurate distinguishing of the tail wave point cloud and the target point cloud is ensured.
Preferably, the difference in height is normalized
The calculation formula is as follows:
wherein d isn,
Respectively representing the target distance and the vertical emission angle obtained by the nth laser beam, dn+k,
Respectively representing the targets obtained by the n + k laser beamsDistance and vertical launch angle; the beam numbers n, n + k indicate two vertically adjacent laser beams striking the obstacle.
Has the advantages that: aiming at the problem of 'under segmentation' which is easy to occur in tail wave filtering (namely tail wave point clouds in a scene are not all identified), the problem is analyzed because the characteristics of classifying a target point and a tail wave point in the prior art are not accurate enough, and further the normalized height difference characteristic of the reflection points of two adjacent beams of laser in the vertical direction is constructed. Because the normalized height difference characteristic value of the target point is usually very large and close to 1, and the normalized height difference characteristic value of the tail wave point is very small and close to 0, the screening mode can effectively classify the target point and the tail wave point, ensure to find out all the tail wave points, and ensure the 'recall ratio' of the tail wave points, thereby avoiding the 'under-segmentation' problem in tail wave filtering.
Preferably, step S2 employs a statistical filter.
Has the advantages that: aiming at the clutter noise problem in the point cloud data, the invention is based on the prior knowledge of clutter noise in the point cloud data of the water surface environment, generally, scattered points caused by electromagnetic noise, water surface ripple reflection and the like, and step S2 preferably adopts a statistical filter, and removes 'outliers' with too far distance by counting the average distance from each data point to the nearest k points in Euclidean space, thereby accurately filtering the scattered clutter noise in the point cloud of the water surface scene. Compared with the problem that the original point cloud is subjected to down-sampling by using general point cloud filtering methods such as voxel filtering, grid filtering and the like, the method cannot influence the original target point cloud and the tail wave point cloud, and ensures accurate subsequent filtering of the tail wave point cloud.
To achieve the above object, according to a second aspect of the present invention, there is provided a computer readable storage medium storing one or more first programs, the one or more first programs being executed by one or more processors to implement the steps of a method for tailwave filtering for an unmanned ship application as described in the first aspect.
Generally, by the above technical solution conceived by the present invention, the following beneficial effects can be obtained:
(1) aiming at the problem that the classification characteristics of a target point and a tail wave point in the prior art are not accurate enough, the invention constructs the normalized height difference characteristic of the reflection points of two adjacent beams of laser in the vertical direction. Two adjacent beams of laser in the vertical direction show the characteristic of great difference in vertical height when striking obstacles and tail waves, and target points and tail wave points in the water surface scene point cloud can be accurately distinguished through a threshold screening strategy.
(2) Aiming at the problem that classification characteristics of a target point and a tail wave point are not robust enough in the prior art, the height difference characteristic value is normalized to a value range of 0-1 by calculating the ratio of the actual height difference to the theoretical height difference, namely the normalized height difference, the problem that the height difference characteristic value changes along with the target distance due to the inherent characteristic that the imaging distance of the laser radar is more far and more sparse is solved, and the effect of the normalized height difference characteristic value on the robustness of the imaging distance is achieved. Meanwhile, when the unmanned ship jolts due to sea condition changes, the normalized height difference feature provided by the invention can still accurately distinguish the target point cloud and the tail wave point cloud, so that the tail wave filtering effect is ensured, and the effect of the normalized height difference feature on sea condition robustness is achieved.
The normalized height difference characteristic provided by the invention can accurately classify the target point cloud and the tail wave point cloud, ensures the recall ratio of tail wave point cloud identification, and avoids the problem of under-segmentation caused by inaccurate classification in tail wave filtering. Meanwhile, the method has good characteristic robustness, can overcome the influence of condition changes such as target distance, sea conditions and the like, and ensures the effect of tail wave filtering.
(3) Aiming at the problem of easy over-segmentation in tail wave filtering, the invention constructs the absolute height characteristic of point cloud in the Z-axis direction based on the priori knowledge that the obstacle target is usually far higher than the tail wave plane, removes the target point cloud wrongly classified as the tail wave point through a threshold screening strategy, ensures the accuracy of tail wave point cloud identification on the premise of not reducing the recall ratio, and avoids the problem of over-segmentation in the tail wave filtering.
In conclusion, the tail wave filtering method for unmanned ship application, provided by the invention, has the advantages that the constructed normalized height difference and absolute height characteristics can accurately identify the tail wave point cloud in the scene, the robustness is good, the problems of under-segmentation and over-segmentation are effectively solved, the tail wave filtering task is simply and efficiently completed, and the time-consuming target discrimination and false alarm removal processing in the subsequent environment perception algorithm is avoided. The method is simple and easy to transplant, can be directly used as a sub-link, and is applied to the environment perception requirement of the unmanned ship.
Drawings
Fig. 1 is a flowchart of a tail wave filtering method applied to an unmanned surface vehicle according to the present invention.
Fig. 2 is a schematic diagram of the height difference of the target point calculated when the laser beam is irradiated on the obstacle.
FIG. 3 is a schematic diagram of the calculated target point height difference when a laser beam strikes an obstacle when the unmanned surface vehicle jolts.
Fig. 4 is a schematic diagram of the calculated height difference of the reflection point when the laser beam strikes the tail waves.
FIG. 5 is a schematic diagram of the laser beam impinging on the flat surface of an obstacle to calculate the height difference of the target point.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
As shown in fig. 1, the invention provides a tail wave filtering method for unmanned ship application, which comprises the following steps:
step 1: and acquiring point cloud data of the water surface environment around the unmanned ship, wherein the point cloud data is obtained by scanning through a multi-line rotary laser radar.
The lidar used in this example was a Pandar40 model from shanghai standing grain company. The radar adopts a multi-line rotary scanning mode, and has N groups of laser transmitters, wherein the value of N is 40. The laser transmitters are vertically arranged at certain angle intervals, periodically and simultaneously emit laser beams, the time difference from the emission of the laser beams to the return is measured, and the distance of the target is calculated by multiplying the time difference by the speed of light. The radar rotates at a constant speed at a certain speed and can scan the surrounding environment to obtain point cloud data. In the embodiment, the radar is horizontally arranged at the position without shielding at the top of the unmanned ship so as to obtain the point cloud data of the water surface environment within the 360-degree view field around the unmanned ship.
The laser beam strikes a target object and is reflected, and a point cloud data point can be obtained, wherein the data point comprises 4 attributes (index, theta)horizontal,θverticalAnd d) respectively representing the laser beam number, the horizontal emission angle, the vertical emission angle and the target distance of the point. All data points acquired by rotating the radar by 360 degrees form a frame of point cloud data.
Step 2: and (3) removing clutter noise from the point cloud data acquired in the step (1).
Any method can be adopted to carry out filtering and denoising processing on the point cloud data. Clutter noise points in the point cloud data of the water surface environment are scattered points generally caused by electromagnetic noise, water surface ripple reflection and the like, and the clutter noise points in the point cloud are removed by preferably adopting a statistical filter method in the embodiment. Specifically, by counting the average distance from each data point in the euclidean space to its nearest k points, the "outliers" that are too far away are eliminated. Compared with the problem that the original point cloud is subjected to down-sampling by using general point cloud filtering methods such as voxel filtering, grid filtering and the like, the method cannot influence the point cloud of the original water surface scene, and ensures accurate subsequent filtering of the tail wave point cloud.
After clutter noise points are removed, the tail wave filtering problem is simplified into a two-classification problem of target points and tail wave points.
And step 3: and (3) calculating the characteristic of normalized height difference of the point cloud data subjected to denoising processing in the step (2), and segmenting a suspected tail wave point set from the scene point cloud.
Calculate the "normalized height difference" feature:
as shown in figure 2, the laser radar rotates to scan the surrounding water surface environment, when two beams of laser light adjacent in the vertical direction strike the obstacle, if the obstacle is in the event of collisionWhen the reflecting surface of the obstacle is perpendicular to the water surface, the difference in height between the two reflecting points by radar is the line segment D'nI.e. "theoretical height difference". In general, the surface of the obstacle has a certain inclination, so the actual height difference of the reflection points of the two laser beams is the line segment D in the figuren。
As shown in fig. 2, the water surface obstacle is usually inclined to a large degree, and when the laser is applied to the obstacle, the actual height difference D is obtainednValue of (A) is generally different from the theoretical height by D'nIs relatively close and much greater than 0; in contrast to fig. 4, the wake appears as a local waviness of the water surface, generally "flat", and when the laser hits the wake, the actual height difference D is obtainednIs generally much less than the theoretical height difference D'nAnd is close to 0. Therefore, the target point cloud and the tail wave point cloud in the scene point cloud can be segmented by utilizing the characteristic.
The water surface environment is complex and changeable, the unmanned ship is often in a bumpy state when sailing, the laser radar is not always vertical to the horizontal plane, as shown in figure 3, when the unmanned ship is in a bumpy state, two adjacent laser beams in the vertical direction strike an obstacle, and the obtained actual height difference D is obtainednStill satisfies the theoretical height difference D'nThe quantity relationship is not large and is far larger than 0; similarly, when the laser is applied to the tail wave, the actual height difference D is obtainednStill satisfies the condition that the height is far less than the theoretical height difference D'nAnd is close to a numerical relationship of 0. Therefore, when the unmanned ship jolts due to the change of sea conditions, the segmentation of the target point cloud and the tail wave point cloud can be carried out by utilizing the characteristic.
The imaging characteristic of the laser radar is that the included angle between the laser beams in the vertical direction is fixed, and the farther the same barrier is away from the radar, the obtained height difference D of the reflection points of the two adjacent beams of laser isnThe larger, and therefore only by a single height difference feature DnTo perform threshold screening, it is difficult to select an appropriate threshold to cope with a change in the obstacle distance. Therefore, the invention proposes a "normalized height difference" feature, denoted as DnIs defined as finding the actual height difference DnAnd theoretical height difference D'nThe ratio of (a) to (b). The characteristic can eliminate the influence of target distance change, and the specific calculation formula is as follows:
wherein d isn,
Respectively representing the target distance and the vertical emission angle obtained by the nth laser beam, dn+k,
Respectively representing the target distance and the vertical emission angle obtained by the n + k beams of laser; the beam numbers n, n + k indicate two vertically adjacent laser beams striking the obstacle. Since not every laser beam will be reflected, the value of k is not always 1, and it can be seen that,
the value range of (1) is 0-1.
Segmenting a suspected tail wave point:
specifically, traversing the point cloud data set subjected to denoising processing in the step 2, calculating normalized height difference characteristics of each point cloud data point, performing characteristic screening on the point cloud data set, and segmenting a 'suspected tail wave point' set. The normalized height difference eigenvalues for the "target point" are typically large, close to 1; the normalized height difference characteristic value of the suspected tail wave point is usually very small and is close to 0; therefore, the present embodiment adopts a threshold value screening method to screen and classify the "target point" and the "suspected tail wave point". Any threshold screening method may be used herein. The present embodiment simply employs a fixed threshold for screening, preferably 0.5. From the above analysis, data points with normalized height difference eigenvalues less than 0.5 are considered as "suspected wake points", and data points with normalized height difference eigenvalues greater than or equal to 0.5 are considered as "target points".
The normalized height difference characteristics can effectively classify target points and tail wave points, and the step can ensure that all tail wave points are found out, namely the 'recall ratio' of the tail wave points is ensured, and the 'under-segmentation' problem of tail wave filtering is solved.
And 4, step 4: and (3) calculating the absolute height characteristic of the 'suspected tail wave point' set obtained by segmentation in the step (3), and segmenting the 'tail wave point' set from the 'suspected tail wave point' set.
Calculate "absolute height" feature:
as shown in fig. 5, when the surface of the obstacle is relatively flat, the calculated actual height difference D is obtainednAnd when the judgment is carried out by utilizing the normalized height difference characteristic, the part of the obstacle point cloud is wrongly judged as the tail wave point cloud to be filtered, so that the point cloud of the obstacle is damaged, the subsequent detection and identification steps are influenced, and the problem of over-segmentation is solved. It can be seen that the water surface obstacle is usually far higher than the horizontal plane, and is represented as that the height of the obstacle point cloud in the Z-axis direction of the laser radar three-dimensional rectangular coordinate system is far higher than that of the tail wave point cloud, so that the ' flat ' part in the obstacle point cloud can be distinguished from the tail wave point cloud through the priori knowledge, and the invention defines the height of the point cloud in the Z-axis direction as the ' absolute height ', and records the absolute height as the ' absolute height
The calculation formula is as follows:
wherein d isn,
Respectively representing the target distance and the vertical emission angle obtained by the nth laser beam.
The "suspected wake wave points" set obtained by the segmentation in step 3 contains the point cloud reflected by the flat surface in the obstacle, and the second threshold screening is performed on the "suspected wake wave points" set according to the absolute height feature, where any one threshold screening method may be adopted. The absolute height characteristic values of the tail wave point cloud and the target point cloud are distributed in respective certain value ranges, the value ranges are continuously changed due to the change of conditions such as target types and sea conditions, and a simple threshold strategy is usually difficult to deal with. The invention preferably adopts a maximum inter-class variance method, calculates a corresponding screening threshold value for each frame of water surface scene point cloud in a self-adaptive manner by maximizing the variance difference of absolute height characteristics between a target point and a tail wave point, can adapt to various condition changes in a robust manner, and ensures accurate distinguishing of the tail wave point cloud and the target point cloud. From the analysis, the absolute height characteristic value of the point cloud of the water surface obstacle is far larger than that of the tail wave point cloud, the part of the suspected tail wave point set, of which the absolute height characteristic value is larger than or equal to the screening threshold value, is removed, and the rest is the tail wave point set.
The step eliminates the target point cloud which is wrongly classified as the tail wave point through the absolute height characteristic, ensures the accuracy of tail wave filtering on the premise of not reducing the recall ratio, and solves the problem of over-segmentation.
And 5: and removing the tail wave points in the point cloud of the water surface scene according to the tail wave point set obtained in the step 3-4.
The method for filtering the tail waves applied to the unmanned ship can simply and efficiently remove the tail wave part in the point cloud of the water surface scene, can be used as a sub-link, and can be directly applied to the environment perception requirement of the unmanned ship.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (7)
1. A tail wave filtering method for unmanned ship application is characterized by comprising the following steps:
s1, point cloud data of a water surface environment around an unmanned ship are obtained, and the point cloud data are obtained through scanning of a multi-line rotary laser radar;
s2, removing clutter noise from the point cloud data;
s3, tail wave point identification is carried out on the point cloud data after clutter noise points are filtered, and the method comprises the following steps:
s31, calculating the actual height difference of two adjacent laser reflection points in the vertical direction in the point cloud data according to plane geometry;
s32, calculating the ratio of the actual height difference to the theoretical height difference of each point cloud data point to serve as a normalized height difference;
s33, identifying the point with the normalized height difference lower than the first threshold as a tail wave point, and adding a tail wave point set;
and S4, removing the identified tail wave point set from the point cloud data of the water surface environment.
2. The method of claim 1, after step S33, further comprising:
s34, calculating the absolute height of each point in the Z-axis direction according to plane geometry;
s35, removing points with absolute height larger than or equal to a second threshold value in the tail wave point set.
3. The method of claim 2, wherein the absolute height
The calculation formula is as follows:
wherein d isn,
Respectively representing the target distance and the vertical emission angle obtained by the nth laser beam.
4. The method of claim 2, wherein the second threshold is determined by: and maximizing the variance of the absolute height between the target point and the tail wave point, thereby determining a second threshold value.
5. The method of any of claims 1 to 4, wherein the height difference is normalized
The calculation formula is as follows:
wherein d isn,
Respectively representing the target distance and the vertical emission angle obtained by the nth laser beam, dn+k,
Respectively representing the target distance and the vertical emission angle obtained by the n + k beams of laser; the beam numbers n, n + k indicate two vertically adjacent laser beams striking the obstacle.
6. The method of any one of claims 1 to 4, wherein step S2 employs a statistical filter.
7. A computer readable storage medium, characterized in that the computer readable storage medium stores one or more first programs, which are executed by one or more processors, to implement the steps of the method for tail wave filtering for unmanned-boat-oriented applications of any of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111220246.6A CN114002708B (en) | 2021-10-20 | 2021-10-20 | Tail wave filtering method for unmanned ship application |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111220246.6A CN114002708B (en) | 2021-10-20 | 2021-10-20 | Tail wave filtering method for unmanned ship application |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114002708A true CN114002708A (en) | 2022-02-01 |
| CN114002708B CN114002708B (en) | 2024-06-14 |
Family
ID=79923280
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111220246.6A Active CN114002708B (en) | 2021-10-20 | 2021-10-20 | Tail wave filtering method for unmanned ship application |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114002708B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110275153A (en) * | 2019-07-05 | 2019-09-24 | 上海大学 | A kind of waterborne target detection and tracking based on laser radar |
| CN112464994A (en) * | 2020-11-05 | 2021-03-09 | 北京航天控制仪器研究所 | Boat stern wave identification and removal method based on PointNet network |
| WO2021072709A1 (en) * | 2019-10-17 | 2021-04-22 | 深圳市大疆创新科技有限公司 | Method for detecting and tracking target, system, device, and storage medium |
-
2021
- 2021-10-20 CN CN202111220246.6A patent/CN114002708B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110275153A (en) * | 2019-07-05 | 2019-09-24 | 上海大学 | A kind of waterborne target detection and tracking based on laser radar |
| WO2021072709A1 (en) * | 2019-10-17 | 2021-04-22 | 深圳市大疆创新科技有限公司 | Method for detecting and tracking target, system, device, and storage medium |
| CN112464994A (en) * | 2020-11-05 | 2021-03-09 | 北京航天控制仪器研究所 | Boat stern wave identification and removal method based on PointNet network |
Non-Patent Citations (2)
| Title |
|---|
| BING LIANG 等: "Metric Reliability Analysis of Autonomous Marine LiDAR Systems under Extreme Wind Loads", J. MAR. SCI. ENG, 25 December 2023 (2023-12-25) * |
| 周利兰;高高;: "高速排水型船舶尾浪影响因素的数值研究", 水动力学研究与进展A辑, no. 01, 30 January 2013 (2013-01-30) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114002708B (en) | 2024-06-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111239766B (en) | Water surface multi-target rapid identification and tracking method based on laser radar | |
| CN108509918B (en) | Target detection and tracking method fusing laser point cloud and image | |
| CN112882059B (en) | Unmanned ship inland river obstacle sensing method based on laser radar | |
| CN108562913B (en) | Unmanned ship false target detection method based on three-dimensional laser radar | |
| CN112464994B (en) | PointNet network-based boat tail wave recognition and removal method | |
| CN112513679B (en) | Target identification method and device | |
| CN111712731A (en) | Target detection method and system and movable platform | |
| CN115761550A (en) | Water surface target detection method based on laser radar point cloud and camera image fusion | |
| CN111524084A (en) | Complex scene photon counting laser radar point cloud denoising algorithm based on multimodal Gaussian fitting | |
| CN112881993B (en) | Method for automatically identifying false flight path caused by radar distribution clutter | |
| CN113671481B (en) | 3D multi-target tracking processing method based on millimeter wave radar | |
| CN113031004A (en) | Unmanned ship water surface target detection and path planning method based on three-dimensional laser radar | |
| CN111913177A (en) | Method and device for detecting target object and storage medium | |
| CN108074232A (en) | A kind of airborne LIDAR based on volume elements segmentation builds object detecting method | |
| CN111929653A (en) | Target detection and tracking method and system based on unmanned ship marine radar | |
| CN115248436A (en) | Imaging sonar-based fish resource assessment method | |
| CN110596728A (en) | Water surface small target detection method based on laser radar | |
| CN113570005A (en) | Long-distance ship type identification method based on airborne photon radar | |
| CN113888589A (en) | Water surface obstacle detection and multi-target tracking method based on laser radar | |
| CN114002708B (en) | Tail wave filtering method for unmanned ship application | |
| CN115267827B (en) | Laser radar harbor area obstacle sensing method based on high density screening | |
| WO2024060209A1 (en) | Method for processing point cloud, and radar | |
| CN116206094B (en) | Fan blade angle measuring method, device and system and electronic equipment | |
| CN114879180B (en) | Seamless situation awareness method for real-time fusion of unmanned ship-borne multi-element multi-scale radar | |
| CN111723797B (en) | Method and system for determining bounding box of three-dimensional target |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |