CN115205717B - Obstacle point cloud data processing method and flight equipment - Google Patents
Obstacle point cloud data processing method and flight equipment Download PDFInfo
- Publication number
- CN115205717B CN115205717B CN202211118209.9A CN202211118209A CN115205717B CN 115205717 B CN115205717 B CN 115205717B CN 202211118209 A CN202211118209 A CN 202211118209A CN 115205717 B CN115205717 B CN 115205717B
- Authority
- CN
- China
- Prior art keywords
- point cloud
- cloud data
- obstacle
- processed
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/17—Terrestrial scenes taken from planes or by drones
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
- G06V10/7635—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks based on graphs, e.g. graph cuts or spectral clustering
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Health & Medical Sciences (AREA)
- Remote Sensing (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Optical Radar Systems And Details Thereof (AREA)
- Image Analysis (AREA)
Abstract
The application discloses a method for processing obstacle point cloud data and flight equipment, wherein the method for processing the obstacle point cloud data comprises the following steps: calculating the occupation proportion of the space occupied by the obstacle point cloud data to be processed in the total space of the obstacle; when the occupation ratio is determined to be smaller than a preset segmentation threshold value, segmenting obstacle point cloud data to be processed into a plurality of sub-point cloud data; and the sub-point cloud data is used as new obstacle point cloud data to be processed, so that the obstacle point cloud data corresponding to the irregular obstacle is segmented, the proportion of space-time white areas of the subsequently constructed obstacle frame is greatly reduced, and the accuracy of the subsequent obstacle identification is improved.
Description
Technical Field
The application relates to the technical field of point cloud data processing, in particular to a method for processing obstacle point cloud data and flight equipment.
Background
Generally, the flight device can carry on detection device, and detection device such as sensor etc. detection device detects the surrounding environment, obtains some cloud data, carries out the detection and the discernment of barrier through some cloud data, and the discernment result can supply the low reaches module to use, if carry out the orbit planning to the automation that realizes flight device keeps away barrier etc..
In the related technology, the point cloud data are aggregated through a density clustering algorithm to obtain obstacle point cloud data, and the aggregated obstacle point cloud data are directly used for constructing a frame of an obstacle; and realizing the detection and identification of the obstacles according to the construction result.
Generally, the frame of the obstacle can be a rectangular frame, but the rectangular frame is suitable for a regular object, but has a large defect for an irregular object, and affects the use of a subsequent downstream module.
Disclosure of Invention
In view of the above problems, the present invention provides an obstacle point cloud data processing method and a flight device to improve the above problems.
In a first aspect, an embodiment of the present application provides an obstacle point cloud data processing method, where the method includes: calculating the occupation proportion of the space occupied by the obstacle point cloud data to be processed in the total space of the obstacle; when the occupation ratio is smaller than a preset segmentation threshold value, segmenting obstacle point cloud data to be processed into a plurality of sub-point cloud data; and taking the sub-point cloud data as new obstacle point cloud data to be processed.
In a second aspect, embodiments of the present application further provide a flight device that includes one or more processors, memory, and one or more applications. Wherein one or more applications are stored in the memory and configured to be executed by the one or more processors, the one or more applications being configured to perform the method of obstacle point cloud data processing as described above in the first aspect.
The obstacle point cloud data processing method provided by the invention comprises the following steps: calculating the occupation proportion of the space occupied by the to-be-processed obstacle point cloud data in the total space of the obstacles; when the occupation ratio is determined to be smaller than a preset segmentation threshold value, segmenting obstacle point cloud data to be processed into a plurality of sub-point cloud data; and the sub-point cloud data is used as new obstacle point cloud data to be processed, so that the obstacle point cloud data corresponding to the irregular obstacle is segmented, the proportion of space-time white areas of the subsequently constructed obstacle frame is greatly reduced, and the accuracy of the subsequent obstacle identification is improved.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required in the description of the embodiments are briefly introduced below, and it is apparent that the drawings in the description below are only some embodiments of the present application, and not all embodiments. All other embodiments and drawings obtained by a person skilled in the art based on the embodiments of the present application without inventive step belong to the protection scope of the present invention.
Fig. 1 shows a schematic flow chart of a method for processing obstacle point cloud data according to an embodiment of the present disclosure.
Fig. 2 shows a schematic flowchart of determining an occupation ratio of obstacle point cloud data to be processed in step 110 according to an embodiment of the present application.
Fig. 3 shows a schematic flow chart of another obstacle point cloud data processing method proposed in the embodiment of the present application.
Fig. 4 shows a schematic structural diagram of an obstacle point cloud data processing apparatus according to an embodiment of the present application.
Fig. 5 shows a schematic structural diagram of a flight device according to an embodiment of the present application.
Fig. 6 shows a schematic structural diagram of a computer-readable storage medium according to an embodiment of the present application.
Detailed Description
In order to make the technical solutions better understood by those skilled in the art, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application.
Generally, the flight device can carry on detection equipment, such as a sensor, and the detection equipment detects the surrounding environment, obtains point cloud data, carries out the detection and the discernment of barrier through point cloud data, and the recognition result can supply the low reaches module to use, if carry out the orbit planning to the automation that realizes flight device keeps away barrier etc..
In the related technology, the point cloud data are aggregated through a density clustering algorithm to obtain obstacle point cloud data, and the aggregated obstacle point cloud data are directly used for constructing a frame of an obstacle; and realizing the detection and identification of the obstacles according to the construction result.
Generally, the frame of the obstacle may be a rectangular frame, but the rectangular frame is suitable for regular objects, such as vehicles, pedestrians, and the like. However, there are relatively large defects for irregular objects, for example, a large empty area, such as an L-shaped fence, a tree with continuous positions and undulation, may appear in the total space of the obstacle defined by the rectangular frame. The blank areas can influence the use of subsequent downstream modules, for example, the blank areas can mislead the planning modules of the downstream automatic obstacle avoidance, namely, the blank safe areas are regarded as the impassable areas, so that the maneuverability and flexibility of the flight equipment are reduced, and the normal flight of the aircraft is influenced. And may affect the extraction of information on the obstacle, for example, may affect the accuracy of the extracted height information of the obstacle.
In some cases, in order to reduce the influence of the blank area, if the point cloud data exceeding a certain number is directly filtered, namely the number of the point cloud data of the obstacles after clustering is set, if the number exceeds a preset threshold value, the obstacles which are unreasonable are considered to be oversized, and then the obstacles are removed. However, many objects with irregular shapes but a large amount of obstacle point cloud data, such as L-shaped walls with an excessively long length, are missed, and thus the flight safety of the flight equipment is affected.
In other cases, if the density clustering radius of the density clustering algorithm is reduced to reduce the influence of the blank area, the generation of too large obstacles can be avoided to some extent, but cannot be completely eliminated. And will increase the number of finely divided small obstacles. And is not friendly to distant sparse point cloud data. The point cloud data can present the characteristics of close density and distant sparsity, and if the density clustering radius is set to be too small, the sparse point cloud data of the same obstacle at a distance can be divided into a plurality of obstacles, so that the integrity of the obstacles is influenced.
In still other cases, in order to reduce the influence of the blank area, the point cloud data of the non-interest area may be filtered in advance by using map information, for example, in the case of an L-shaped wall, and the wall, trees, etc. may be removed in advance by using the map information, so as to only pay attention to the obstacle on the driving path. However, the perceived task of the flying device is focused on the full range of detection, rather than on only a portion of the region of interest. By adopting the map information mode, an additional data source is needed, and the cost is increased. Although the number of obstacles is reduced to a certain extent and the speed is increased, the situation that the detection is not timely occurs aiming at the obstacles which suddenly appear from the non-interested region to the interested region, and the safety is affected.
In order to improve the above problem, the inventor proposes an obstacle point cloud data processing method and a flight device provided by the present application, wherein the obstacle point cloud data processing method includes: calculating the occupation proportion of the space occupied by the obstacle point cloud data to be processed in the total space of the obstacle; when the occupation ratio is smaller than a preset segmentation threshold value, segmenting obstacle point cloud data to be processed into a plurality of sub-point cloud data; and the sub-point cloud data is used as new obstacle point cloud data to be processed, so that the obstacle point cloud data corresponding to the irregular obstacle is segmented, the proportion of space-time white areas of the subsequently constructed obstacle frame is greatly reduced, and the accuracy of the subsequent obstacle identification is improved.
Referring to fig. 1, an obstacle point cloud data processing method is provided in the embodiment of the present application, and the method may be applied to flight equipment, and it may be understood that the obstacle point cloud data processing method in the embodiment of the present application may also be applied to other equipment that needs to perform obstacle point cloud data processing, which is not limited herein. Specifically, the method comprises the following steps: step 110 to step 130.
And 110, calculating the occupation proportion of the space occupied by the cloud data of the obstacle points to be processed in the total space of the obstacle.
In an embodiment of the present application, before step 110, the method for processing obstacle point cloud data provided in the embodiment of the present application may further include: and filtering the point cloud data acquired by the sensor to obtain initial obstacle point cloud data to be processed.
In the embodiment of the application, the flight device may be provided with a sensor, and the sensor may detect the detection area to acquire point cloud data of the detection area. The point cloud data is a data set of points of the detection area under preset coordinates. The point cloud data may include information such as three-dimensional coordinates, colors, classification values, intensity values, time, and the like.
In some embodiments, the sensor may be a lidar sensor, a three-dimensional imaging sensor, such as a binocular camera, a three-dimensional scanner, an RGB-D (three primary colors and depth) camera, and the like.
In some embodiments, the point cloud data collected by the sensor may not be used directly, and the initial obstacle point cloud data to be processed may be obtained by filtering.
In some embodiments, the filtering process includes removing ground point cloud data from the point cloud data, resulting in initial point cloud data.
In an embodiment of the application, the point cloud data comprises information of all points of the detection area, such as ground, pedestrians, houses, etc. In actual use, information in which influence is exerted on the flight, such as obstacles on the ground, is more focused. The ground information may affect the sensing of the obstacle, and therefore, the ground point cloud data in the point cloud data needs to be removed to obtain initial point cloud data related to the obstacle.
In some embodiments, the ground point cloud data may be removed by using a normal vector method, a RANSAC algorithm (random sample consensus algorithm), or the like, which is not limited herein.
In some embodiments, the filtering further includes clustering the initial point cloud data to obtain initial obstacle point cloud data to be processed.
In the embodiment of the application, the initial point cloud data belongs to point cloud data of an obstacle, but a detection area may include a plurality of obstacles, so that the initial point cloud data needs to be subjected to target classification, that is, the point cloud data belonging to the same obstacle are gathered together, that is, clustering processing is performed, so as to form a plurality of obstacle point cloud data, and each obstacle point cloud data is used for representing the point cloud data belonging to the same obstacle, so that subsequent analysis can be performed.
In the embodiment of the present application, the clustering process may adopt a density clustering algorithm, such as K-means clustering, DBSCAN clustering, region growing clustering, and the like, and may specifically be selected according to actual use needs, which is not limited in the present application.
Blank areas may exist in the plurality of obstacle point cloud data obtained through clustering, and in order not to affect the use of the subsequent downstream module, the obstacle point cloud data can be further processed, so that the obstacle point cloud data can meet the use requirements of the downstream module. In the embodiment of the application, a tag can be set for the obstacle point cloud data, the tag of the obstacle point cloud data which is not processed is marked as a to-be-processed state, and the tag of the obstacle point cloud data which is processed is marked as a processed state, so that the processing states of the obstacle point cloud data are distinguished.
In the embodiment of the application, the plurality of obstacle point cloud data obtained through filtering are initial obstacle point cloud data to be processed. In an embodiment of the present application, the number of the initial obstacle point cloud data to be processed may be one or more. The initial obstacle point cloud data to be processed is not subjected to subsequent further processing, and when it is determined through subsequent processing that the obstacle point cloud data meets the use requirements of the downstream module, the obstacle point cloud data can be marked to be in a processed state to serve as processed obstacle point cloud data, and the subsequent processing is specifically described.
When the number of the obstacle point cloud data to be processed is multiple, each obstacle point cloud data to be processed can be traversed in sequence, so that each obstacle point cloud data can meet the use requirements of the downstream modules.
In the embodiment of the application, the occupation proportion is the occupation proportion of the space occupied by the obstacle point cloud data to be processed in the total space of the obstacle, so that the occupation proportion is used for measuring the occupation of a white area in the total space of the obstacle.
In some embodiments, step 110 comprises: step 111 to step 114.
And 111, performing principal component analysis on the obstacle point cloud data to be processed, and constructing a reduction matrix according to an analysis result.
In the embodiment of the application, the principal component analysis is to project the obstacle point cloud data to be processed to a characteristic direction (for example, a first direction and a second direction of a preset two-dimensional space), and a projection of each data point in the obstacle point cloud data to be processed on a vector is a principal component.
The obstacle point cloud data to be processed includes a plurality of high-dimensional data points, which may be three-dimensional data points, for example, and the plurality of data points form a matrix as an input of the principal component analysis. Through principal component analysis, the eigenvalue and the eigenvector of the matrix can be obtained, and the matrix can be described through combination of the eigenvalue and the eigenvector.
In some embodiments, step 111 comprises the following steps.
(1) And performing principal component analysis on the obstacle point cloud data to be processed to obtain the characteristic value and the characteristic vector of the obstacle point cloud data to be processed.
(2) And constructing a reduction matrix according to the eigenvalue and the eigenvector.
And 112, performing projection transformation processing on the obstacle point cloud data to be processed based on the reduction matrix to obtain point cloud projection data.
In the embodiment of the application, the obstacle point cloud data to be processed can be subjected to projection transformation processing to a preset two-dimensional space based on the reduction matrix, and the data projected on the two-dimensional space is point cloud projection data, so that the point cloud data information is compressed.
And 113, determining the two-dimensional attribute of the obstacle point cloud data to be processed according to the point cloud projection data.
In some embodiments, the two-dimensional attributes include a first length of the point cloud projection data in a first direction in the two-dimensional space and a second length in a second direction in the two-dimensional space.
In some embodiments, step 113 may include the following steps.
(1) And determining a boundary value of the point cloud projection data.
In an embodiment of the application, the boundary values of the point cloud projection data include a maximum value and a minimum value of the point cloud projection data in a first direction of the two-dimensional space, and a maximum value and a minimum value of the point cloud projection data in a second direction of the two-dimensional space.
(2) And determining a first length of the point cloud projection data in a first direction of the two-dimensional space and a second length of the point cloud projection data in a second direction of the two-dimensional space according to the boundary value and the preset segmentation size.
In the embodiment of the application, the preset segmentation size can be set according to the area size of the point cloud projection data, the calculation speed can be influenced if the preset segmentation size is too large, the preset segmentation size is too small, and the number of data points of each unit is too small. Alternatively, the preset cut size may be set to 0.1 to 0.8m, such as 0.2m, 0.3m, and 0.5m. Preferably, the preset dicing size may be set to 0.4m. It can be understood that the preset slicing size may be set according to an actual application scenario and a use requirement, which is not limited in this application.
In some embodiments, a first length of the point cloud projection data in a first direction of the two-dimensional space may be determined by the following equation one, and a second length of the point cloud projection data in a second direction of the two-dimensional space may be determined by the following equation two.
Wherein the content of the first and second substances,
characterizing the first length;
representing the maximum value of the point cloud projection data in the first direction;
representing the minimum value of the point cloud projection data in the first direction;
characterizing the second length;
representing the maximum value of the point cloud projection data in the second direction;
representing the minimum value of the point cloud projection data in the second direction;
and representing the preset cutting size.
And step 114, determining the occupation proportion of the total space of the obstacles in the space occupied by the point cloud data of the obstacles to be processed according to the point cloud data of the obstacles to be processed and the two-dimensional attributes.
In the embodiment of the application, the total space of the obstacles corresponding to the obstacle point cloud data to be processed can be determined according to the two-dimensional attributes.
Further, each data point in the obstacle point cloud data to be processed may be projected to the two-dimensional space, so as to determine an area actually occupied by the obstacle point cloud data to be processed in the two-dimensional space (i.e., a space actually occupied by the obstacle point cloud data to be processed).
In some embodiments, the step of determining the occupation ratio of the obstacle point cloud data to be processed according to the obstacle point cloud data to be processed and the two-dimensional attribute comprises the following steps.
(1) The total number of occupancies is determined from the first length and the second length.
(2) And determining the actual occupied number of the obstacle point cloud data to be processed in the two-dimensional space.
(3) And determining the occupation proportion of the space occupied by the cloud data of the obstacle points to be processed in the total space of the obstacle according to the actual occupation number and the total occupation number.
In some embodiments, for convenience of calculation, the point cloud projection data may be divided into a plurality of units according to a preset segmentation size, and the length of each unit in the first direction and the length of each unit in the second direction are both the preset segmentation sizes.
The total occupied number, i.e. the number of divided units, may then be determined from the first length and the second length.
Specifically, the total occupied number may be determined according to the following equation three.
Wherein the content of the first and second substances,
characterizing a total occupancy number;
characterizing the first length;
the second length is characterized.
In the embodiment of the application, data points in the obstacle point cloud data to be processed can be sequentially projected to the two-dimensional space, and the number of units occupied by the projection of the data points on the two-dimensional space is counted, so that the actual occupied number is determined.
Exemplarily, the obstacle point cloud data to be processed includes a first data point and a second data point, and the number of cells occupied by the projection of the first data point on the two-dimensional space is 2; the number of cells occupied by the projection of the second data point on the two-dimensional space is 5, and the actual occupied number of the obstacle point cloud data to be processed on the two-dimensional space is 7.
Further, the occupation ratio of the preprocessed point cloud data in the two-dimensional space can be determined according to the following formula four.
Wherein the content of the first and second substances,
characterizing the occupation ratio;
characterizing the actual occupancy amount;
the total occupancy number is characterized.
In some embodiments, the size of some fine-grained obstacles is small, the area of the blank area is relatively small, and the subsequent influence is small, so that whether the point cloud data of the obstacle to be processed belongs to the fine-grained obstacles can be judged firstly, and the occupation ratio is determined only when the point cloud data does not belong to the fine-grained obstacles, so that the calculation amount can be saved.
In some embodiments, step 114 comprises the following steps.
(1) And determining whether the point cloud data of the obstacle to be processed meets the boundary condition according to the two-dimensional attribute.
In some embodiments, the two-dimensional attributes include a first length of the point cloud projection data in a first direction of the two-dimensional space and a second length in a second direction of the two-dimensional space.
In some embodiments, the boundary condition may be one of the first length being greater than a first preset threshold and the second length being greater than a second preset threshold.
For example, the boundary condition may be that the first length is greater than a first preset threshold.
For example, the boundary condition may be that the second length is greater than a second preset threshold.
For example, the boundary condition may be that the first length is greater than a first preset threshold and the second length is greater than a second preset threshold.
(2) And if the to-be-processed obstacle point cloud data meets the boundary condition, determining the occupation proportion according to the to-be-processed obstacle point cloud data and the two-dimensional attributes.
If the to-be-processed obstacle point cloud data meets the boundary conditions, it can be determined that the obstacle corresponding to the to-be-processed obstacle point cloud data does not belong to a fine-crushing obstacle, and subsequent processing can be performed.
In some embodiments, the obstacle point cloud data processing method provided in the embodiments of the present application further includes: and if the obstacle point cloud data to be processed do not meet the boundary conditions, taking the obstacle point cloud data to be processed as processed obstacle point cloud data.
And if at least one of the first length is smaller than a first preset threshold value and the second length is smaller than a second preset threshold value is met, determining that the point cloud data of the obstacle to be processed does not meet the boundary condition. That is to say, the obstacle corresponding to the obstacle point cloud data to be processed belongs to a finely-divided obstacle, and does not need to be processed, that is, the obstacle point cloud data to be processed may be used as processed obstacle point cloud data (for example, the tag state of the obstacle point cloud data may be changed, and the state to be processed may be changed to the processed state), so as to be further processed by a subsequent downstream module.
In some embodiments, the first preset threshold may be 0.2 to 0.5m. Optionally, the first preset threshold may take a value of 0.4m.
In some embodiments, the second preset threshold may be 0.2 to 0.5m. Optionally, the second preset threshold may take a value of 0.4m.
It is understood that the first preset threshold and the second preset threshold may be set according to actual detection requirements, and the present application is not limited thereto.
(3) And determining the occupation proportion of the obstacle point cloud data to be processed according to the actual occupation number and the total occupation number.
And 120, if the occupation ratio is smaller than a preset segmentation threshold, segmenting the obstacle point cloud data to be processed into a plurality of sub-point cloud data.
In the embodiment of the application, the preset segmentation threshold value can be 0.2 to 0.4. Optionally, the preset segmentation threshold may take a value of 0.4.
It is understood that the preset segmentation threshold may be set according to actual detection requirements, and the present application is not limited thereto.
If the occupation ratio is smaller than a preset segmentation threshold, determining that a large-area blank area exists in the obstacle point cloud data to be processed, and performing segmentation into sub-point cloud data with a smaller area so as to reduce the proportion of the blank area in the total space of the obstacle and avoid influencing subsequent processing.
In some embodiments, the partitioning of the obstacle point cloud data to be processed into a plurality of sub-point cloud data in step 120 includes the following steps.
(1) And carrying out segmentation processing on the point cloud projection data of the obstacle point cloud data to be processed to obtain a plurality of sub-segmentation data.
In some embodiments, the step of segmenting the obstacle point cloud data to be processed to obtain a plurality of sub-segmented data; comprises the following steps.
And (1.1) determining a segmentation direction according to the two-dimensional attribute of the point cloud data of the obstacle to be processed.
In some embodiments, the two-dimensional attributes include a first length of the point cloud projection data in a first direction in the two-dimensional space and a second length in a second direction in the two-dimensional space.
In some embodiments, the first length may be compared to the second length; and determining the direction corresponding to the larger one of the first length and the second length as the dividing direction.
Illustratively, if the first length is greater than the second length, the first direction is taken as the dividing direction.
Illustratively, if the first length is less than the second length, the second direction is taken as the dividing direction.
In some embodiments, if the first length is equal to the second length, any one of the first direction and the second direction may be arbitrarily selected as the dividing direction.
And (1.2) carrying out segmentation processing on the point cloud projection data based on the segmentation direction to obtain a plurality of sub-segmentation data.
In some embodiments, the point cloud projection data may be equally divided, i.e. the plurality of sub-divided data are equal in length in the dividing direction.
Illustratively, the division direction isAnd in the first direction, the point cloud projection data are segmented based on the first direction and are two sub-segmented data. For example, the point cloud projection data has a maximum value in a first direction
(ii) a The minimum value of the point cloud projection data in the first direction is
Then the length of each of the divided sub-data after the division in the first direction is
。
(2) And reducing the plurality of sub-segmentation data according to the reduction matrix to obtain a plurality of sub-point cloud data.
In the embodiment of the application, the reduction matrix obtained in the above step is used to perform transposition processing on the plurality of sub-divided data to obtain a plurality of sub-point cloud data.
And step 130, taking the sub-point cloud data as new obstacle point cloud data to be processed.
In the embodiment of the application, original obstacle point cloud data to be processed belonging to the same obstacle are divided into a plurality of obstacles, and each piece of sub-point cloud data corresponds to one obstacle. And the segmented sub-point cloud data may need to be segmented again.
Therefore, the sub-point cloud data can be used as obstacle point cloud data to be processed, the obstacle point cloud data to be processed needs to be processed and determined through the steps, and processed obstacle point cloud data with the blank area ratio meeting the requirements are finally obtained.
In some embodiments, the obstacle point cloud data processing method provided in the embodiments of the present application further includes: and if the occupation ratio is greater than or equal to the preset segmentation threshold, taking the obstacle point cloud data to be processed as processed obstacle point cloud data.
In the embodiment of the application, if the occupation ratio is greater than or equal to the preset segmentation threshold, it is determined that the occupation ratio of the blank area in the obstacle point cloud data to be processed is smaller, and the obstacle point cloud data to be processed can meet the subsequent processing requirement, so that segmentation is not needed, and the obstacle point cloud data to be processed can be used as processed obstacle point cloud data to be processed by a subsequent downstream module for continuous processing.
In some embodiments, the obstacle point cloud data processing method provided in the embodiments of the present application further includes: and if the obstacle point cloud data to be processed do not exist, performing obstacle identification processing according to the processed obstacle point cloud data.
In the embodiment of the application, after traversing all obstacle point cloud data to be processed, the irregular obstacle can be divided once or for multiple times to form more accurate obstacle point cloud data with a plurality of blank areas smaller.
When the obstacle point cloud data to be processed does not exist in the plurality of obstacle point cloud data, it is shown that blank areas in all the obstacle point cloud data at the moment meet requirements, and then the plurality of processed obstacle point cloud data can be sent to a downstream module for subsequent processing, so that accurate obstacle point cloud data is sent to the downstream module, and interference to subsequent operation is reduced. For example, the obstacle identification process may be performed according to the plurality of processed obstacle point cloud data, such as assignment of attributes of length, width, speed, angle, and the like of the obstacle.
According to the obstacle point cloud data processing method provided by the embodiment of the application, a large-range blank area in the obstacle point cloud data can be avoided, the processed obstacle point cloud data is more accurate, and various attributes of the obstacle can be more accurately extracted (for example, the extracted height information is closer to the real height of the obstacle, and the occurrence of abnormal height is reduced); meanwhile, the required parameters are few, the operation efficiency is high, the consumed time is short (can be less than 1 ms), and the real-time requirement of automatic driving is met. Moreover, by dividing the obstacle data which do not meet the requirements, ultra-large clustering obstacles can not be filtered, and the condition of missing detection can not be caused. Moreover, the clustering radius does not need to be reduced, and excessive fine and broken small obstacles at a distance cannot be generated. The algorithm can realize the full-range sensing task which can be achieved by the laser without depending on other data sources; further, the sensing task with the full range of the detection range can be realized without depending on other data sources.
In some embodiments, the present application further provides an obstacle point cloud data processing method, including steps 201 to 220.
And step 201, obstacle point cloud data to be processed.
In the embodiment of the application, the obstacle point cloud data to be processed can be obtained by filtering the initial point cloud data to obtain a plurality of initial obstacle point cloud data to be processed.
In some embodiments, the obstacle point cloud data to be processed may also be derived from new obstacle point cloud data to be processed obtained by re-dividing the obstacle point cloud data with a larger blank area.
In some embodiments, each obstacle point cloud data may set a cluster index. For example, clustering the initial point cloud data to obtain a plurality of obstacle point cloud data may randomly set a cluster index for each obstacle point cloud data.
Exemplarily, the initial point cloud data is clustered to obtain three obstacle point cloud data, which are respectively the first obstacle point cloud data, the second obstacle point cloud data and the third obstacle point cloud data, so that the cluster index of the first obstacle point cloud data can be set to 1, the cluster index of the second obstacle point cloud data can be set to 2, and the cluster index of the third obstacle point cloud data can be set to 3. The clustering number is 3, namely the number of the obstacle point cloud data.
In some embodiments, if the obstacle point cloud data with a large blank area is re-segmented to obtain new obstacle point cloud data to be processed, a cluster index may be set for the new obstacle point cloud data to be processed, and the value of the cluster number may be adjusted.
Exemplarily, if the first obstacle point cloud data is re-segmented to obtain new fourth obstacle point cloud data and fifth obstacle point cloud data to be processed, the cluster index of the fourth obstacle point cloud data may be set to 4, and the index of the fifth obstacle point cloud data may be set to 5. And if the newly added obstacle point cloud data to be processed exist, updating the clustering number according to the number of the newly added obstacle point cloud data to be processed. For example, if the first obstacle point cloud data is re-segmented to obtain new fourth obstacle point cloud data and fifth obstacle point cloud data to be processed, the clustering number is updated to 5.
Step 202, judging whether the cluster index is less than the cluster number.
In the embodiment of the application, the obstacle point cloud data to be processed can be determined according to the cluster index and processed, so that each obstacle point cloud data to be processed is ensured to be traversed.
In some embodiments, the obstacle point cloud data to be processed may be processed sequentially according to the order of the cluster indexes from small to large.
For example, the first obstacle point cloud data with the cluster index of 1 may be processed first, and then the obstacle point cloud data with the cluster index of 2, 3 \8230 \ 8230may be processed in sequence.
In some embodiments, if the cluster index is less than the number of clusters, it indicates that there is obstacle point cloud data to be processed, and step 203 is executed.
If the cluster index is equal to the cluster number, it indicates that there is no obstacle point cloud data to be processed, and the processing of the obstacle point cloud data to be processed can be finished.
And step 203, principal component analysis.
In the embodiment of the application, the obstacle point cloud data to be processed can be subjected to principal component analysis, and a reduction matrix is constructed.
In some embodiments, the reduction matrix may also be used for the reduction process of step 209.
And step 204, calculating the two-dimensional attribute.
In the embodiment of the application, the point cloud data of the obstacle to be processed can be subjected to projection transformation processing to obtain point cloud projection data, and the two-dimensional attributes are determined according to the point cloud projection data.
In the embodiment of the application, whether the to-be-processed obstacle point cloud data belongs to a fine-crushing obstacle is determined based on the two-dimensional attribute. The size of the fine obstacles is small and the fine obstacles do not need to be divided again.
If the obstacle belongs to a fine-crushing obstacle, executing step 206, namely taking the current obstacle point cloud data to be processed as processed obstacle point cloud data; otherwise, step 207 is performed.
And step 206, determining the processed obstacle point cloud data.
And step 207, determining the actual occupied number.
In the embodiment of the application, the total occupied number can be determined according to the two-dimensional attribute of the obstacle point cloud data to be processed, each data point in the obstacle point cloud data to be processed can be projected to a two-dimensional space, and the actual occupied number occupied by all the data points can be counted.
And step 208, determining whether the occupation ratio is smaller than a preset segmentation threshold value.
In an embodiment of the present application, the occupation ratio may be determined according to the actual occupation number and the total occupation number, and the occupation ratio may be compared with a preset division threshold.
If the occupancy ratio is smaller than the preset segmentation threshold, it indicates that a large blank area exists in the obstacle point cloud data, and step 209 is executed; otherwise, if the point cloud data of the obstacle to be processed meets the requirement of the subsequent processing, step 206 is executed.
And step 209, segmentation processing.
In the embodiment of the application, the unsatisfactory obstacle point cloud data to be processed is continuously segmented into a plurality of sub-segmented data.
And 210, reduction treatment.
In an embodiment of the present application, the plurality of sub-divided data may be reduced through the reduction matrix obtained in step 201 to obtain a plurality of sub-point cloud data.
And step 211, performing elevation reassignment.
In an embodiment of the application, a height array is provided, wherein the height array comprises a height maximum and a height minimum of each obstacle point cloud data.
The original obstacle point cloud data to be processed is divided into a plurality of sub-point cloud data, namely, one original obstacle is divided into a plurality of obstacles, and the numerical value in the height data is updated according to the plurality of sub-point cloud data after division. For example, the height value of the original obstacle point cloud may be removed. And re-determining the height value of each sub-point cloud data, and updating the height value of each sub-point cloud data into a height array.
And returning the plurality of sub-point cloud data serving as new obstacle point cloud data to be processed to step 201, labeling the clustering index, updating the clustering number, and repeating the previous steps serving as the obstacle point cloud data to be processed.
Referring to fig. 4, an embodiment of the present application further provides an obstacle point cloud data processing apparatus 300, where the apparatus 300 includes: a scale determination module 310, a segmentation module 320, and a processing module 330.
The proportion determining module 310 is configured to calculate an occupation proportion of a space occupied by the to-be-processed obstacle point cloud data in a total space of the obstacle.
The segmentation module 320 is configured to segment the obstacle point cloud data to be processed into a plurality of sub-point cloud data when it is determined that the occupancy ratio is smaller than a preset segmentation threshold.
The processing module 330 is configured to use the sub-point cloud data as new obstacle point cloud data to be processed.
It should be noted that, for the device-type embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and reference may be made to part of the description of the method embodiment for relevant points. For any processing manner described in the method embodiment, all the processing manners may be implemented by corresponding processing modules in the apparatus embodiment, and details in the apparatus embodiment are not described again.
In addition, functional modules in the embodiments of the present application may be integrated into one processing module, or each of the modules may exist alone physically, or two or more modules are integrated into one module. The integrated module can be realized in a hardware mode, and can also be realized in a software functional module mode.
Referring to fig. 5, based on the above obstacle point cloud data processing method, an embodiment of the present application further provides a flight device 400 capable of executing the above obstacle point cloud data processing method.
In an embodiment of the present application, the flight device 400 includes one or more processors 410, memory 420, and one or more applications. One or more application programs are stored in the memory 420, wherein the memory 420 stores programs that can execute the contents of the foregoing embodiments, and the processor 410 can execute the programs stored in the memory.
The Memory 420 may include a Random Access Memory (RAM) or a Read-Only Memory (Read-Only Memory). The memory 420 may be used to store instructions, programs, code, sets of codes, or sets of instructions. The memory may include a stored program area and a stored data area, where the stored program area may store instructions for implementing an operating system, instructions for implementing at least one function (such as split processing, occupancy determination, etc.), instructions for implementing the various method embodiments described below, and the like. The storage data area can also store data (such as obstacle point cloud data to be processed, occupation ratio and the like) created by the terminal in use.
Referring to fig. 6, a block diagram of a computer-readable storage medium 500 according to an embodiment of the present disclosure is shown. The computer-readable storage medium 500 has stored therein a program code 610, the program code 510 being capable of being invoked by a processor to perform the obstacle point cloud data processing method described in the above-described method embodiments.
The computer-readable storage medium 500 may be an electronic memory such as a flash memory, an EEPROM (electrically erasable and programmable read only memory), an EPROM, a hard disk, or a ROM. Alternatively, the computer-readable storage medium includes a non-volatile computer-readable medium. The computer-readable storage medium 500 has a storage space of program codes for executing any of the method steps in the above-described obstacle point cloud data processing method. The program code 510 can be read from or written to one or more computer program products. The program code may be compressed, for example, in a suitable form.
In summary, an obstacle point cloud data processing method and flight equipment provided by the embodiments of the present application include: calculating the occupation proportion of the space occupied by the to-be-processed obstacle point cloud data in the total space of the obstacles; when the occupation ratio is determined to be smaller than a preset segmentation threshold value, segmenting obstacle point cloud data to be processed into a plurality of sub-point cloud data; and the sub-point cloud data is used as new obstacle point cloud data to be processed, so that the obstacle point cloud data corresponding to the irregular obstacle is segmented, the proportion of space-time white areas of the subsequently constructed obstacle frame is greatly reduced, and the accuracy of the subsequent obstacle identification is improved.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit the same; although the present application has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not depart from the spirit and scope of the corresponding technical solutions in the embodiments of the present application.
Claims (13)
1. An obstacle point cloud data processing method, characterized by comprising:
performing principal component analysis on obstacle point cloud data to be processed to obtain a characteristic value and a characteristic vector of the obstacle point cloud data to be processed;
constructing a reduction matrix according to the eigenvalue and the eigenvector;
performing projection transformation processing on the obstacle point cloud data to be processed based on the reduction matrix to obtain point cloud projection data;
determining two-dimensional attributes of the to-be-processed obstacle point cloud data according to the point cloud projection data;
determining the occupation proportion of the space occupied by the obstacle point cloud data to be processed in the total space of the obstacle according to the obstacle point cloud data to be processed and the two-dimensional attributes; wherein the total space of the obstacle is a space defined by a frame of the obstacle;
when the occupation ratio is determined to be smaller than a preset segmentation threshold value, segmenting the obstacle point cloud data to be processed into a plurality of sub-point cloud data;
and taking the sub-point cloud data as new obstacle point cloud data to be processed.
2. The method of claim 1, further comprising:
and filtering the point cloud data acquired by the sensor to obtain initial obstacle point cloud data to be processed.
3. The method according to claim 1, wherein the two-dimensional attributes comprise a first length of the obstacle point cloud projection data to be processed in a first direction of a two-dimensional space and a second length in a second direction of the two-dimensional space;
the determining the two-dimensional attribute of the obstacle point cloud data to be processed according to the point cloud projection data comprises the following steps:
determining a boundary value of the point cloud projection data;
and determining a first length of the to-be-processed obstacle point cloud projection data in a first direction of a two-dimensional space and a second length of the point cloud projection data in a second direction of the two-dimensional space according to the boundary value and a preset segmentation size.
4. The method according to claim 1, wherein the determining the occupancy ratio of the obstacle point cloud data to be processed according to the obstacle point cloud data to be processed and the two-dimensional attributes comprises:
determining whether the to-be-processed obstacle point cloud data meets boundary conditions or not according to the two-dimensional attributes;
and if the to-be-processed obstacle point cloud data meet the boundary condition, determining the occupation proportion according to the to-be-processed obstacle point cloud data and the two-dimensional attribute.
5. The method of claim 4, further comprising:
and if the to-be-processed obstacle point cloud data does not meet the boundary condition, taking the to-be-processed obstacle point cloud data as processed obstacle point cloud data.
6. The method of claim 5, wherein the two-dimensional attributes comprise a first length of the point cloud projection data in a first direction in two-dimensional space and a second length in a second direction in two-dimensional space;
determining whether the cloud data of the obstacle points to be processed meet boundary conditions according to the two-dimensional attributes comprises the following steps: and if at least one of the first length is smaller than a first preset threshold value and the second length is smaller than a second preset threshold value is met, determining that the point cloud data of the obstacle to be processed does not meet the boundary condition.
7. The method according to claim 3, wherein the determining the occupation ratio of the space occupied by the obstacle point cloud data to be processed in the total space of the obstacle according to the obstacle point cloud data to be processed and the two-dimensional attribute comprises:
determining a total occupancy number from the first length and the second length;
determining the actual occupied number of the obstacle point cloud data to be processed in a two-dimensional space;
and determining the occupation proportion of the space occupied by the to-be-processed obstacle point cloud data in the total space of the obstacle according to the actual occupation number and the total occupation number.
8. The method of claim 1, wherein the segmenting the obstacle point cloud data to be processed into a plurality of sub-point cloud data comprises:
carrying out segmentation processing on the point cloud projection data of the obstacle point cloud data to be processed to obtain a plurality of sub-segmentation data;
and restoring the plurality of sub-segmentation data according to the restoration matrix to obtain a plurality of sub-point cloud data.
9. The method according to claim 8, wherein the point cloud projection of the obstacle point cloud data to be processed is subjected to segmentation processing to obtain a plurality of sub-segmentation data; the method comprises the following steps:
determining a segmentation direction according to the two-dimensional attribute of the point cloud data of the obstacle to be processed;
and carrying out segmentation processing on the point cloud projection data based on the segmentation direction to obtain a plurality of sub-segmentation data.
10. The method of claim 9, wherein the two-dimensional attributes comprise a first length of the point cloud projection data in a first direction in two-dimensional space and a second length in a second direction in two-dimensional space;
determining a segmentation direction according to the two-dimensional attribute of the point cloud data of the obstacle to be processed; the method comprises the following steps:
comparing the first length to the second length; and determining the direction corresponding to the larger numerical value as the dividing direction.
11. The method according to any one of claims 1 to 10, further comprising:
and if the occupation ratio is greater than or equal to a preset segmentation threshold, taking the obstacle point cloud data to be processed as processed obstacle point cloud data.
12. The method of claim 11, further comprising:
and if the obstacle point cloud data to be processed does not exist, performing obstacle identification processing according to the processed obstacle point cloud data.
13. A flying apparatus, comprising:
one or more processors;
a memory;
one or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, the one or more applications configured to perform the obstacle point cloud data processing method of any of claims 1-12.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211118209.9A CN115205717B (en) | 2022-09-14 | 2022-09-14 | Obstacle point cloud data processing method and flight equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211118209.9A CN115205717B (en) | 2022-09-14 | 2022-09-14 | Obstacle point cloud data processing method and flight equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115205717A CN115205717A (en) | 2022-10-18 |
| CN115205717B true CN115205717B (en) | 2022-12-20 |
Family
ID=83572608
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211118209.9A Active CN115205717B (en) | 2022-09-14 | 2022-09-14 | Obstacle point cloud data processing method and flight equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115205717B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115797814B (en) * | 2023-02-13 | 2023-05-23 | 广东电网有限责任公司肇庆供电局 | Real-time evaluation and calculation method for hidden danger of power transmission line tree obstacle based on unmanned aerial vehicle |
| CN116954265B (en) * | 2023-09-20 | 2023-12-05 | 天津云圣智能科技有限责任公司 | Method and device for rescheduling local motion trail and electronic equipment |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020258314A1 (en) * | 2019-06-28 | 2020-12-30 | 西门子（中国）有限公司 | Cutting method, apparatus and system for point cloud model |
| CN112166457A (en) * | 2019-08-29 | 2021-01-01 | 深圳市大疆创新科技有限公司 | Point cloud segmentation method and system and movable platform |
| CN112287939A (en) * | 2020-10-29 | 2021-01-29 | 平安科技（深圳）有限公司 | Three-dimensional point cloud semantic segmentation method, device, equipment and medium |
| CN112712089A (en) * | 2020-12-31 | 2021-04-27 | 的卢技术有限公司 | Obstacle detection method, obstacle detection device, computer device, and storage medium |
| CN113658195A (en) * | 2021-08-16 | 2021-11-16 | 金地（集团）股份有限公司 | Image segmentation method and device and electronic equipment |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109828592B (en) * | 2019-04-22 | 2019-07-26 | 深兰人工智能芯片研究院（江苏）有限公司 | A kind of method and apparatus of detection of obstacles |
| CN113808077A (en) * | 2021-08-05 | 2021-12-17 | 西人马帝言(北京)科技有限公司 | Target detection method, device, equipment and storage medium |
-
2022
- 2022-09-14 CN CN202211118209.9A patent/CN115205717B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020258314A1 (en) * | 2019-06-28 | 2020-12-30 | 西门子（中国）有限公司 | Cutting method, apparatus and system for point cloud model |
| CN112166457A (en) * | 2019-08-29 | 2021-01-01 | 深圳市大疆创新科技有限公司 | Point cloud segmentation method and system and movable platform |
| CN112287939A (en) * | 2020-10-29 | 2021-01-29 | 平安科技（深圳）有限公司 | Three-dimensional point cloud semantic segmentation method, device, equipment and medium |
| CN112712089A (en) * | 2020-12-31 | 2021-04-27 | 的卢技术有限公司 | Obstacle detection method, obstacle detection device, computer device, and storage medium |
| CN113658195A (en) * | 2021-08-16 | 2021-11-16 | 金地（集团）股份有限公司 | Image segmentation method and device and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115205717A (en) | 2022-10-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115205717B (en) | Obstacle point cloud data processing method and flight equipment | |
| CN110348294B (en) | Method and device for positioning chart in PDF document and computer equipment | |
| CN109948497B (en) | Object detection method and device and electronic equipment | |
| CN111144242B (en) | Three-dimensional target detection method, device and terminal | |
| CN108470174B (en) | Obstacle segmentation method and device, computer equipment and readable medium | |
| JP7253573B2 (en) | Matching method, device, electronic device and computer readable storage medium | |
| CN113761999A (en) | Target detection method and device, electronic equipment and storage medium | |
| CN112711034A (en) | Object detection method, device and equipment | |
| US20230419659A1 (en) | Method and system for processing point-cloud data | |
| CN113467875A (en) | Training method, prediction method, device, electronic equipment and automatic driving vehicle | |
| CN111932545A (en) | Image processing method, target counting method and related device thereof | |
| CN114419599A (en) | Obstacle identification method and device and electronic equipment | |
| CN114241448A (en) | Method and device for obtaining heading angle of obstacle, electronic equipment and vehicle | |
| CN113870196A (en) | Image processing method, device, equipment and medium based on anchor point cutting graph | |
| CN113806464A (en) | Road tooth determining method, device, equipment and storage medium | |
| CN113449702A (en) | Target detection method and device for remote sensing image, storage medium and electronic equipment | |
| CN111783648A (en) | Method and device for extracting guardrail in road point cloud | |
| CN111340139A (en) | Method and device for judging complexity of image content | |
| CN116595104A (en) | Visual processing method of mine safety monitoring system | |
| CN115116080A (en) | Table analysis method and device, electronic equipment and storage medium | |
| CN114549884A (en) | Abnormal image detection method, device, equipment and medium | |
| CN112132215B (en) | Method, device and computer readable storage medium for identifying object type | |
| CN114511862A (en) | Form identification method and device and electronic equipment | |
| CN116863325A (en) | Method for multiple target detection and related product | |
| CN113989300A (en) | Lane line segmentation method and device, electronic equipment and storage medium |
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 |