CN116452467A - Container real-time positioning method based on laser data - Google Patents
Container real-time positioning method based on laser data Download PDFInfo
- Publication number
- CN116452467A CN116452467A CN202310714705.9A CN202310714705A CN116452467A CN 116452467 A CN116452467 A CN 116452467A CN 202310714705 A CN202310714705 A CN 202310714705A CN 116452467 A CN116452467 A CN 116452467A
- Authority
- CN
- China
- Prior art keywords
- container
- data
- matrix block
- window
- obtaining
- 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 37
- 239000011159 matrix material Substances 0.000 claims abstract description 155
- 238000012545 processing Methods 0.000 claims abstract description 26
- 238000001914 filtration Methods 0.000 claims abstract description 14
- 230000000877 morphologic effect Effects 0.000 claims description 29
- 239000013598 vector Substances 0.000 claims description 16
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000007405 data analysis Methods 0.000 claims description 3
- 230000003044 adaptive effect Effects 0.000 abstract description 5
- 238000012544 monitoring process Methods 0.000 abstract description 3
- 230000000694 effects Effects 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 4
- 238000010606 normalization Methods 0.000 description 4
- 238000012937 correction Methods 0.000 description 3
- 230000000737 periodic effect Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- 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
- G01S17/02—Systems using the reflection of electromagnetic waves other than radio waves
- G01S17/06—Systems determining position data of a target
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration using local operators
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/155—Segmentation; Edge detection involving morphological operators
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
-
- 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
- G06V10/255—Detecting or recognising potential candidate objects based on visual cues, e.g. shapes
-
- 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
- G06V10/30—Noise filtering
-
- 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
- G06V10/763—Non-hierarchical techniques, e.g. based on statistics of modelling distributions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10028—Range image; Depth image; 3D point clouds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20112—Image segmentation details
- G06T2207/20164—Salient point detection; Corner detection
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Electromagnetism (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Geometry (AREA)
- Probability & Statistics with Applications (AREA)
- Computer Networks & Wireless Communication (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The invention relates to the technical field of point cloud data processing based on laser radar, in particular to a container real-time positioning method based on laser data, which comprises the following steps: acquiring data to obtain a data matrix; clustering by utilizing the matrix to obtain a container matrix block; obtaining noise influence degree in a container matrix block; obtaining the real influence degree of noise according to the vertical inclination degree of the container; obtaining an adaptive Gaussian filter window according to the real influence degree of noise; denoising by using the adaptive window; and analyzing the denoised laser point cloud data. According to the invention, by analyzing the concave-convex structural characteristics of the upper surface of the container, noise points in the collected laser point cloud data are primarily identified, and then according to the influence degree of the vertical inclination of the container on noise identification, the noise points are accurately identified, the size of a self-adaptive filtering window is used, and then, gaussian filtering is used for accurately denoising, so that high-quality point cloud data are obtained and sent to a monitoring platform, and the position of the container is monitored in real time and accurately.
Description
Technical Field
The invention relates to the technical field of point cloud data processing based on laser radar, in particular to a container real-time positioning method based on laser data.
Background
The real-time positioning of the container plays an important role in modern logistics and freight transportation, the laser scanner is utilized to scan the container, point cloud data on the surface of the container is obtained, the real-time monitoring and recognition of the position and the posture of the container are realized, abnormal conditions can be found and processed in time, and the safety of the freight is ensured. Along with the continuous development of information technology, the container real-time positioning technology based on laser data is combined with other internet of things technologies, so that intelligent, visual and automatic logistics management can be formed.
However, in the process of collecting laser point cloud data, the obtained point cloud data is always prevented from having noise points due to the influence of the instrument or the gesture during scanning, and the accuracy of the subsequent real-time positioning of the container is affected. The common gaussian filtering can reduce noise and retain data edge information, but the size of a filtering window is selected according to manual experience, when the size of the filtering window is larger, the denoising effect is stronger, but some details can be smoothed, so that some important characteristics and detail information of the data are lost, and when the size of the window is selected inappropriately, the filtering effect is poor.
Disclosure of Invention
The invention provides a container real-time positioning method based on laser data, which aims to solve the existing problems.
The invention discloses a container real-time positioning method based on laser data, which adopts the following technical scheme:
one embodiment of the invention provides a container real-time positioning method based on laser data, which comprises the following steps:
collecting array type point cloud data and obtaining a laser point cloud data matrix;
clustering is carried out according to the laser point cloud data matrix to obtain a container matrix block;
obtaining the influence degree of noise of each data point on the basis of a window according to each data point in each container matrix block;
obtaining the long side direction of a container, counting according to the long side direction of the container and each data point in a container matrix block to obtain the distance between two adjacent convex structures on the upper surface of the container, obtaining the size of a morphological processing window according to the distance between two adjacent convex structures on the upper surface of the container, carrying out morphological processing on the container matrix block by utilizing the size of the morphological processing window to obtain a smooth matrix block, and obtaining the vertical inclination degree of the container according to the smooth matrix block;
obtaining the real influence degree of the noise according to the long side direction of the container and the up-down inclination degree of the container and the influence degree of the noise of each data point;
the method comprises the steps of obtaining the size of a Gaussian filter window required by each data point in each container matrix block according to the real influence degree of noise, and performing Gaussian filtering on each container matrix block by utilizing the size of the Gaussian filter window required by each data point in each container matrix block to obtain high-quality laser point cloud data;
and realizing the real-time positioning function of the container according to the high-quality laser point cloud data analysis.
Preferably, the acquiring array type point cloud data and acquiring the laser point cloud data matrix includes the following specific steps:
and (3) overlooking and collecting array type point cloud data of the container by using a laser scanner in a linear scanning mode to obtain a laser point cloud data matrix of the container, wherein each data point in the matrix represents the treatment height distance between the data point obtained by laser ranging and the ground.
Preferably, the clustering is performed according to the laser point cloud data matrix to obtain a container matrix block, which comprises the following specific steps:
clustering the laser point cloud data matrix of the container by using a DBSCAN density clustering algorithm to obtain a plurality of categories, representing each category as each divided matrix block, calculating the arithmetic mean value of data points in each divided matrix block, setting a first threshold value, classifying the matrix blocks with the mean value smaller than the first threshold value as ground areas, otherwise classifying the matrix blocks as container surface areas, acquiring each divided matrix block in the container area as each independent container surface area, and marking each independent container surface area as a container matrix block.
Preferably, the method for obtaining the influence degree of noise of each data point on a window basis according to each data point in each container matrix block comprises the following specific steps:
and acquiring each data point in each container matrix block, marking as a window center, acquiring each window in a fixed size at each window center, counting the minimum value in variances of data value sequences in four directions acquired by lines, columns and two diagonals of each window center, acquiring the minimum value in absolute values of differences between each data point and other data points in the window and the maximum value of the data points in each window, and acquiring the influence degree of noise of each data point according to the minimum value in the variances of the data value sequences in four directions, the minimum value in absolute values of differences between each data point and other data points in the window and the maximum value of the data points in each window.
Preferably, the specific calculation formula for obtaining the influence degree of the noise of each data point according to the minimum value in the variance of the data value sequence in four directions, the minimum value in the absolute value of the difference value between each data point and other data points in the window of each data point and the maximum value of the data points in each window is as follows:
where C is the degree of influence of noise of each data point, D is the minimum of the absolute values of the differences of each data point from other data points in its window, and F is the maximum of the data points in each windowThe value of the sum of the values,representing the minimum of the variances of the sequence of data points for each data point in its four directions within its window.
Preferably, the method for obtaining the long side direction of the container comprises the following specific steps:
and obtaining the gradient angle of each data point in each container matrix block by utilizing a derivative mode of a two-dimensional discrete function, obtaining an angle set, obtaining the gradient direction corresponding to the gradient angle of the mode in the angle set, and if a plurality of modes exist, re-obtaining the gradient direction corresponding to the arithmetic mean value of the gradient angles of the modes, wherein the gradient direction is the long side direction of the container.
Preferably, the counting is performed according to the long side direction of the container and each data point in the container matrix block to obtain the distance between two adjacent convex structures on the upper surface of the container, and the morphological processing window size is obtained according to the distance between two adjacent convex structures on the upper surface of the container, and the method comprises the following specific steps:
in each container matrix block, taking the long side direction of each container as the direction and each data point of the first row of each container matrix block as the starting point, obtaining a ray set of each container matrix block, sequentially carrying out least square fitting on the data points belonging to the container matrix block on each ray in the ray set of each container matrix block to obtain a curve, sequentially counting Euclidean distances between the data points corresponding to two adjacent peak points on each fitting curve, recording all Euclidean distances into one Euclidean distance set, taking the mode numbers in the Euclidean distance set, and if a plurality of mode numbers exist, re-taking the average value of the mode numbers, and recording the mode numbers as the distance between two adjacent raised grids on the upper surface of the container, wherein circular structural elements with the diameter larger than the distance between the two adjacent raised grids on the upper surface of the container are the morphological processing window size.
Preferably, the method for performing morphological processing on the container matrix block by using the size of the morphological processing window to obtain a smooth matrix block, and obtaining the vertical inclination degree of the container according to the smooth matrix block includes the following specific steps:
the method comprises the steps of obtaining a container matrix block, sequentially carrying out morphological closing operation and morphological opening operation on the container matrix block according to a morphological processing window with a window size being a morphological processing window size, obtaining a smooth matrix block, obtaining a direction vector with the largest gradient change in the smooth matrix block, marking the direction vector as the vertical inclination degree of the container, marking the gradient mean value in the smooth matrix block as a vector module of the vertical inclination degree of the container, and marking the direction with the largest gradient change as a vector direction of the vertical inclination degree of the container.
Preferably, the specific calculation formula for obtaining the real influence degree of the noise according to the long side direction of the container and the up-down inclination degree of the container and the influence degree of the noise of each data point is as follows:
where P is the true degree of influence of each data point, E represents the smaller value of the absolute value of the difference between each data point and the adjacent two data points in the straight line of each data point in the long-side direction of the upper surface of the container, K represents the maximum value of the data points in each container matrix block,an angle value representing the degree of vertical inclination of each container and the long side direction of each container,a vector module representing the degree of vertical tilting of the container, C being the degree of influence of noise for each data point within each container matrix block, U representing the influence of the degree of vertical tilting of the entire upper surface of each container on noise recognition,an exponential function based on a natural constant e is represented.
Preferably, the step of obtaining the size of the gaussian filter window required by each data point in each container matrix block according to the real influence degree of noise comprises the following specific steps:
obtaining the real influence degree of noise of each data point in each container matrix block, and calculating to obtain the self-adaptive window size specifically required by each data point in each container matrix block according to the real influence degree of noise of each data point in each container matrix block, wherein the window size is limited in a certain range, and the window is a square filter window.
The technical scheme of the invention has the beneficial effects that: according to the concave-convex structural characteristics of the upper surface of the container, noise points in collected laser point cloud data are identified, but the accuracy of noise point identification is affected due to the fact that the container is placed in a slightly vertical inclined mode, morphological processing is conducted on the upper surface data of the container, the influence degree of the vertical inclined mode of the container on noise identification is obtained, the noise points are accurately identified, larger filtering windows are given to data points with large noise influence degree, denoising effect is improved, smaller filtering windows are given to data points with small noise influence degree, and loss of detail information is prevented.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flow chart of steps of the method for positioning a container in real time based on laser data.
Detailed Description
In order to further describe the technical means and effects adopted by the invention to achieve the preset aim, the following detailed description is given below of the specific implementation, structure, characteristics and effects of the container real-time positioning method based on laser data according to the invention by combining the attached drawings and the preferred embodiment. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The following specifically describes a specific scheme of the container real-time positioning method based on laser data provided by the invention with reference to the accompanying drawings.
Referring to fig. 1, a flowchart of steps of a method for positioning a container based on laser data in real time according to an embodiment of the present invention is shown, the method includes the following steps:
step S001: and acquiring array type point cloud data and acquiring a laser point cloud data matrix.
In this embodiment, the laser scanner is used to collect the array type point cloud data of the container in a linear scanning manner, and the array type point cloud data are orderly arranged in a plane, so as to obtain a laser point cloud data matrix of the container as follows:
wherein,,is a laser point cloud data matrix of a container, and the size of the laser point cloud data matrix is as followsM and N are positive integers, M and N represent the rows and columns of the matrix respectively, and there are，M and n are positive integers, each data point in the matrixRepresenting the vertical height distance between the data point of the mth row and the nth column obtained by laser ranging and the ground.
Step S002: and performing DBSCAN density clustering according to the laser point cloud data matrix to obtain a container matrix block.
In this embodiment, the laser point cloud data matrix a of the container is clustered by using the DBSCAN density clustering algorithm to obtain several categories, each category represents each divided matrix block, and the value of a data point at the ground in the laser point cloud data matrix a of the container should be 0, but due to the complexity of actual data distribution, the embodiment gives the allowable range of the data point, namely, sets a first threshold valueWhere B is the height of one container in practice. Calculating an arithmetic mean of data points within each of the divided matrix blocks, the mean being less than a first thresholdIs classified as a ground area, otherwise is classified as a container surface area. According to the stacking and placing modes of the containers, the containers in different layers are different in height, a certain gap or a certain height fall exists between the containers, so that each matrix block divided in the container area is a surface area of each individual container, and each surface area of each individual container is marked as a container matrix block.
Step S003: the degree of influence of noise of each data point is obtained on a window basis according to each data point in each container matrix block.
The present embodiment obtains each data point in each container matrix block, noted as the window center, in a size of one dimensionIs to analyze each of the container matrix blocks on a window basisObtaining each window by data points, counting rows, columns and two diagonals at the center of each window to obtain four-direction data value sequences, respectively calculating variances of the four-direction data value sequences, and recording the minimum value in the variances of the four-direction data value sequences as。
The regular concave-convex structure of the upper surface of the container described in this embodiment means that the data values of the concave-convex structure along the wide side direction of the container are similar, and the data value variation along the long side direction of the container conforms to the periodic concave-convex variation. According to the shape of the upper surface of the container, the data points in each container matrix block have a certain similarity rule along the data point sequence in a certain direction and the data point sequence in a direction parallel to the data point sequence, and noise points can destroy the rule. Therefore, the influence degree C of noise of each data point obtained in this embodiment is:
wherein D represents the minimum of the absolute values of the differences of each data point from the other data points in its window, and F represents the maximum of the data points in each window, i.eFor the normalization process,representing the minimum of the variances of the sequence of data values for the data point in four directions within its window,the larger the probability that the data point is a noise point is, but since the concave-convex structure on the upper surface of the container only has the data value similar along the wide side direction of the container, when the adjacent data point of the normal data point in the direction is the noise point, the normal data point is mistakenly regarded as the noise point, namely the more likely the D value of the data point is larger is the noise point, the normalization is adoptedAs a means ofThe product of the two represents the degree of influence of the data point as noise.
Step S004: and deriving according to each data point in each container matrix block to obtain an angle set, and obtaining the long side direction of the container according to the angle set.
In this embodiment, a derivative manner of a two-dimensional discrete function (this is known technology) is used to obtain a gradient angle of each data point in each container matrix block, and an angle set is obtained. Any two data points within each container matrix block in the description of this embodiment will have a difference less than a first thresholdThe upper surface of the container is of a concave-convex structure meeting the periodic concave-convex change rule, so that the gradient directions of most data points in the container matrix block are equal and the gradient directions are the long-side directions of the container, and the gradient directions corresponding to the mode gradient angles in the angle set are takenIf a plurality of modes exist, the gradient direction corresponding to the arithmetic mean value of the gradient angles of the modes is re-takenRecord gradient directionIs the long side direction of the container,is 1.
Step S005: and counting according to the long side direction of the container and each data point in the container matrix block to obtain the distance between two adjacent convex structures on the upper surface of the container, obtaining the morphological processing window size according to the distance between two adjacent convex structures on the upper surface of the container, carrying out morphological processing on the container matrix block by utilizing the morphological processing window size to obtain a smooth matrix block, and obtaining the vertical tilting degree of the container according to the smooth matrix block.
In each container matrix block, taking the long side direction of each container as the direction and taking each data point of the first row of each container matrix block as the starting point to acquire a ray set of each container matrix block. Carrying out least square fitting on data points belonging to the container matrix block on each ray in a ray set of each container matrix block in sequence to obtain a curve, sequentially counting Euclidean distances between data points corresponding to two adjacent peak points on each fitting curve, recording all Euclidean distances into one Euclidean distance set, taking the number in the Euclidean distance set as G, if a plurality of modes exist, calculating the average value of the plurality of modes as G, and taking G as the distance between two adjacent raised gratings on the upper surface of the container. Thus, the distance between two adjacent raised grids on the upper surface of the container is obtained.
Since the data set structure will be smoothed when the size of structural element is larger than that of the data set structure in morphological processing, the distance between two adjacent raised grids on the upper surface of the containerThe use is known size is larger thanWhen each container matrix block is morphologically manipulated by the structural elements of (a) the smooth matrix block is obtained, so that the present embodiment uses a diameter ofThe circular structural elements of the container matrix block are subjected to morphological closing operation and then morphological opening operation to smooth the concave-convex structure of the upper surface of the container, so that a smooth matrix block is obtained, and the change of data points in the smooth matrix block can reflect the vertical inclination degree of the container. So that the calculation is performed in the smooth matrix blockThe direction vector with the largest gradient change is recorded as the vertical inclination degree of the containerThe vector mode is the average value of the gradient in the smooth matrix block, and the vector direction is the direction with the largest gradient change.
Step S006: the real influence degree of the noise is obtained according to the long side direction of the container and the up-down inclination degree of the container and the influence degree of the noise of each data point.
Because the container may be slightly inclined up and down, the actual upper surface height data of the container acquired by the laser may not conform to the sine change rule satisfied by the upper surface shape of the container, that is, in step S003, a certain error exists in the noise influence degree C obtained according to the sine change rule satisfied by the upper surface shape of the container, that is, a part of normal data points are processed into noise data points by mistake, so that the embodiment calculates the correction coefficient by analyzing the upper and lower inclination degree of the container, and obtains the accurate noise influence degree, that is, the real influence degree of noise. Therefore, the calculation formula of the real influence degree of the data point selected in each container matrix block as noise is as follows:
wherein P represents the real influence degree of noise of each data point in each container matrix block, E represents the long-side direction of each data point in each container matrix block on the upper surface of the containerThe smaller value of the absolute difference value between each data point and the adjacent front and back data points in the straight line, K represents the maximum value of the data points in each container matrix blockIs normalized.Representing the long side direction of each containerDegree of vertical inclination with containerThe included angle value of the two vectors is known to be within the range of(in this embodiment, the angle range is in radian), thenFor the normalization process,a vector module representing the degree of vertical tilting of the container, C being the degree of influence of noise for each data point within each container matrix block, U representing the influence of the degree of vertical tilting of the entire upper surface of each container on noise recognition,representing an exponential function based on a natural constant e, thereforeAnd correcting the correction coefficient of the influence of the overall vertical inclination degree of the upper surface of each container on noise recognition on the influence degree of noise of each data point. The value T of the vector model of the vertical inclination degree of the container can be known to represent the vertical inclination degree of the container, and when the inclination direction of the container is the long side direction of the container, namely H is 0, the data points at the concave-convex structure of the container do not influence the influence degree of noise of each data point, and when H is gradually increased to the degree thatWhen the container concave-convex structure is used, the data points have similar rules along the wide edge direction of the container (each containerThe data points in the matrix block have a certain rule similar to the sequence of the data points in the same direction along the sequence of the data points in a certain direction), so that the degree of destruction gradually increasesAnd carrying out normalization adjustment on T, wherein the product U of the T and the T represents the influence of the overall vertical inclination degree of the upper surface of the container on noise identification. It is known that when there is no up-down inclination of the container, the gradient direction of the normal data point at the concave-convex structure should be the long side direction of the container, and the smaller the value E in the absolute value of the difference between the data point and the adjacent data point in the long side direction of the container, the larger the influence of the up-down inclination of the container on the gradient direction of the data point, and since the gradient direction of the normal data value of the smooth surface at the non-concave structure of the upper surface of the container is influenced only by the up-down inclination of the container, it should be similar in the long side direction of the container, i.e., E is close to 0, so thatAdjusting U to makeAs E increases, the product of the two indicates that each data point is affected by the degree of tilt of the container up and down. The larger the value, the less trustworthy the C value. Therefore, inversely normalizedFor the correction factor of C, the product of the two represents the true degree to which the selected data point is noise.
Thus, the real influence degree that each data point in each container matrix block is noise is obtained.
Step S007: and obtaining the size of a Gaussian filter window required by each data point in each container matrix block according to the real influence degree of noise, and performing Gaussian filtering on each container matrix block by utilizing the size of the Gaussian filter window required by each data point in each container matrix block to obtain high-quality laser point cloud data.
Acquiring each containerEach data point in the bin matrix block is the true degree of influence of noiseRecorded in a collectionWherein，Is the number of data points within each container matrix block. The specific desired adaptive window size for each data point in this embodimentIs set to a window size of between 3 and 9 data points and is shaped as a square filter window, then the window size is adapted to the particular needs of each data point within each container matrix blockThe method comprises the following steps:
wherein,,indicating the actual degree of influence of noise at the x-th data point within each container matrix block,representing a rounding down calculation process. Since the filter window size should be odd, whenWhen even, pairAnd carrying out 1 adding operation. Thereby obtaining the adaptive window size required for each data point within each container matrix block. And carrying out Gaussian filtering by utilizing a filtering window with an adaptive filtering window size, so as to realize denoising treatment of each container matrix block.
Thus, the denoising processing of all container matrix blocks in the laser point cloud data matrix A of the container is completed, and the largest area, namely the area without important data, in the ground area in the laser point cloud data matrix A of the container is usedAnd the filter window of the filter is subjected to denoising treatment to realize the strongest denoising effect. Acquiring a laser point cloud data matrix after accurate denoising, and recording the laser point cloud data matrix as high-quality laser point cloud data。
Step S008: and realizing the real-time positioning function of the container according to the high-quality laser point cloud data analysis.
High quality laser point cloud data continues to be processed in this embodimentTransmitting the laser point cloud data to a monitoring platform, and obtaining high-quality laser point cloud dataAnd determining the position and shape of the container, further estimating the attitude information of the container, including rotation angle, translation distance, etc., and finally updating the position and the attitude of the container in real time by using the estimated attitude information. The estimation process is a known means, and the embodiment is not described in detail. So far, the embodiment realizes the real-time positioning function of the container.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.
Claims (10)
1. The container real-time positioning method based on the laser data is characterized by comprising the following steps of:
collecting array type point cloud data and obtaining a laser point cloud data matrix;
clustering is carried out according to the laser point cloud data matrix to obtain a container matrix block;
obtaining the influence degree of noise of each data point on the basis of a window according to each data point in each container matrix block;
obtaining the long side direction of a container, counting according to the long side direction of the container and each data point in a container matrix block to obtain the distance between two adjacent convex structures on the upper surface of the container, obtaining the size of a morphological processing window according to the distance between two adjacent convex structures on the upper surface of the container, carrying out morphological processing on the container matrix block by utilizing the size of the morphological processing window to obtain a smooth matrix block, and obtaining the vertical inclination degree of the container according to the smooth matrix block;
obtaining the real influence degree of the noise according to the long side direction of the container and the up-down inclination degree of the container and the influence degree of the noise of each data point;
the method comprises the steps of obtaining the size of a Gaussian filter window required by each data point in each container matrix block according to the real influence degree of noise, and performing Gaussian filtering on each container matrix block by utilizing the size of the Gaussian filter window required by each data point in each container matrix block to obtain high-quality laser point cloud data;
and realizing the real-time positioning function of the container according to the high-quality laser point cloud data analysis.
2. The method for positioning a container in real time based on laser data according to claim 1, wherein the collecting array type point cloud data and obtaining the laser point cloud data matrix comprises the following specific steps:
and (3) overlooking and collecting array type point cloud data of the container by using a laser scanner in a linear scanning mode to obtain a laser point cloud data matrix of the container, wherein each data point in the matrix represents the treatment height distance between the data point obtained by laser ranging and the ground.
3. The method for positioning the container in real time based on the laser data according to claim 1, wherein the clustering according to the laser point cloud data matrix to obtain the container matrix block comprises the following specific steps:
clustering the laser point cloud data matrix of the container by using a DBSCAN density clustering algorithm to obtain a plurality of categories, representing each category as each divided matrix block, calculating the arithmetic mean value of data points in each divided matrix block, setting a first threshold value, classifying the matrix blocks with the mean value smaller than the first threshold value as ground areas, otherwise classifying the matrix blocks as container surface areas, acquiring each divided matrix block in the container area as each independent container surface area, and marking each independent container surface area as a container matrix block.
4. The method for positioning containers in real time based on laser data according to claim 1, wherein the step of obtaining the influence degree of noise of each data point on a window basis according to each data point in each container matrix block comprises the following specific steps:
and acquiring each data point in each container matrix block, marking as a window center, acquiring each window in a fixed size at each window center, counting the minimum value in variances of data value sequences in four directions acquired by lines, columns and two diagonals of each window center, acquiring the minimum value in absolute values of differences between each data point and other data points in the window and the maximum value of the data points in each window, and acquiring the influence degree of noise of each data point according to the minimum value in the variances of the data value sequences in four directions, the minimum value in absolute values of differences between each data point and other data points in the window and the maximum value of the data points in each window.
5. The method for positioning a container in real time based on laser data according to claim 4, wherein the specific calculation formula for obtaining the influence degree of noise of each data point according to the minimum value in variance of the data value sequence in four directions, the minimum value in absolute value of difference value between each data point and other data points in the window, and the maximum value of data points in each window is as follows:
where C is the degree of influence of noise for each data point, D is the minimum of the absolute values of the differences of each data point from the other data points in its window, F is the maximum of the data points in each window,representing the minimum of the variances of the sequence of data points for each data point in its four directions within its window.
6. The method for positioning the container in real time based on the laser data according to claim 1, wherein the method for obtaining the long side direction of the container comprises the following specific steps:
and obtaining the gradient angle of each data point in each container matrix block by utilizing a derivative mode of a two-dimensional discrete function, obtaining an angle set, obtaining the gradient direction corresponding to the gradient angle of the mode in the angle set, and if a plurality of modes exist, re-obtaining the gradient direction corresponding to the arithmetic mean value of the gradient angles of the modes, wherein the gradient direction is the long side direction of the container.
7. The method for positioning a container in real time based on laser data according to claim 1, wherein the counting is performed according to the long side direction of the container and each data point in the matrix block to obtain the distance between two adjacent convex structures on the upper surface of the container, and the morphological processing window size is obtained according to the distance between two adjacent convex structures on the upper surface of the container, comprising the specific steps of:
in each container matrix block, taking the long side direction of each container as the direction and each data point of the first row of each container matrix block as the starting point, obtaining a ray set of each container matrix block, sequentially carrying out least square fitting on the data points belonging to the container matrix block on each ray in the ray set of each container matrix block to obtain a curve, sequentially counting Euclidean distances between the data points corresponding to two adjacent peak points on each fitting curve, recording all Euclidean distances into one Euclidean distance set, taking the mode numbers in the Euclidean distance set, and if a plurality of mode numbers exist, re-taking the average value of the mode numbers, and recording the mode numbers as the distance between two adjacent raised grids on the upper surface of the container, wherein circular structural elements with the diameter larger than the distance between the two adjacent raised grids on the upper surface of the container are the morphological processing window size.
8. The method for positioning a container in real time based on laser data according to claim 1, wherein the step of performing morphological processing on the container matrix block by using the size of the morphological processing window to obtain a smooth matrix block, and obtaining the vertical tilting degree of the container according to the smooth matrix block comprises the following specific steps:
the method comprises the steps of obtaining a container matrix block, sequentially carrying out morphological closing operation and morphological opening operation on the container matrix block according to a morphological processing window with a window size being a morphological processing window size, obtaining a smooth matrix block, obtaining a direction vector with the largest gradient change in the smooth matrix block, marking the direction vector as the vertical inclination degree of the container, marking the gradient mean value in the smooth matrix block as a vector module of the vertical inclination degree of the container, and marking the direction with the largest gradient change as a vector direction of the vertical inclination degree of the container.
9. The method for positioning a container in real time based on laser data according to claim 1, wherein the specific calculation formula for obtaining the real influence degree of noise according to the longitudinal direction of the container and the up-down inclination degree of the container combined with the influence degree of noise of each data point is as follows:
where P is the true degree of influence of each data point, E represents the smaller value of the absolute value of the difference between each data point and the adjacent two data points in the straight line of each data point in the long-side direction of the upper surface of the container, K represents the maximum value of the data points in each container matrix block,an angle value indicating the degree of vertical inclination of each container and the long side direction of each container +.>Vector module for representing the degree of vertical tilting of a container, C being the degree of influence of noise for each data point in each container matrix block, U representing the influence of the degree of vertical tilting of the upper surface of each container as a whole on noise recognition,'>An exponential function based on a natural constant e is represented.
10. The method for positioning containers in real time based on laser data according to claim 1, wherein the step of obtaining the size of the gaussian filter window required for each data point in each container matrix block according to the real influence degree of noise comprises the following specific steps:
obtaining the real influence degree of noise of each data point in each container matrix block, and calculating to obtain the self-adaptive window size specifically required by each data point in each container matrix block according to the real influence degree of noise of each data point in each container matrix block, wherein the window size is limited in a certain range, and the window is a square filter window.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310714705.9A CN116452467B (en) | 2023-06-16 | 2023-06-16 | Container real-time positioning method based on laser data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310714705.9A CN116452467B (en) | 2023-06-16 | 2023-06-16 | Container real-time positioning method based on laser data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116452467A true CN116452467A (en) | 2023-07-18 |
| CN116452467B CN116452467B (en) | 2023-09-22 |
Family
ID=87125958
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310714705.9A Active CN116452467B (en) | 2023-06-16 | 2023-06-16 | Container real-time positioning method based on laser data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116452467B (en) |
Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101251381A (en) * | 2007-12-29 | 2008-08-27 | 武汉理工大学 | Double container positioning system based on machine vision |
| CN101894255A (en) * | 2010-06-13 | 2010-11-24 | 电子科技大学 | A container number location method based on wavelet transform |
| US20110055172A1 (en) * | 2009-09-01 | 2011-03-03 | Containertrac, Inc. | Automatic error correction for inventory tracking and management systems used at a shipping container yard |
| US20110052001A1 (en) * | 2009-09-01 | 2011-03-03 | Container Trac, Inc. | Automatic error detection for inventory tracking and management systems used at a shipping container yard |
| CN102024144A (en) * | 2010-11-23 | 2011-04-20 | 上海海事大学 | Container number identification method |
| CN109272524A (en) * | 2018-08-27 | 2019-01-25 | 大连理工大学 | A Small Scale Point Cloud Noise Denoising Method Based on Threshold Segmentation |
| US20190188505A1 (en) * | 2017-12-14 | 2019-06-20 | COM-IoT Technologies | Distracted driver detection |
| CN111814936A (en) * | 2020-08-27 | 2020-10-23 | 上海西井信息科技有限公司 | Container identification method, system, device and storage medium based on space scanning |
| WO2021175403A1 (en) * | 2020-03-02 | 2021-09-10 | LASE Industrielle Lasertechnik GmbH | Container-loading system and method for monitoring operation therein |
| CN114219842A (en) * | 2021-12-14 | 2022-03-22 | 东南大学 | Visual recognition, ranging and positioning methods in port container automatic loading and unloading operations |
| WO2022057747A1 (en) * | 2020-09-17 | 2022-03-24 | 上海驭矩信息科技有限公司 | Container positioning method and apparatus based on multi-line laser data fusion |
| CN115222625A (en) * | 2022-07-15 | 2022-10-21 | 桂林理工大学 | A method for denoising LiDAR point cloud based on multi-scale noise |
| WO2023019709A1 (en) * | 2021-08-19 | 2023-02-23 | 中国铁路设计集团有限公司 | Automatic detection method of conductor height and pull-out value of overhead line system based on vehicle-mounted mobile laser point cloud |
| WO2023025030A1 (en) * | 2021-08-26 | 2023-03-02 | 上海交通大学 | Three-dimensional point cloud up-sampling method and system, device, and medium |
| CN116110053A (en) * | 2023-04-13 | 2023-05-12 | 济宁能源发展集团有限公司 | Container surface information detection method based on image recognition |
-
2023
- 2023-06-16 CN CN202310714705.9A patent/CN116452467B/en active Active
Patent Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101251381A (en) * | 2007-12-29 | 2008-08-27 | 武汉理工大学 | Double container positioning system based on machine vision |
| US20110055172A1 (en) * | 2009-09-01 | 2011-03-03 | Containertrac, Inc. | Automatic error correction for inventory tracking and management systems used at a shipping container yard |
| US20110052001A1 (en) * | 2009-09-01 | 2011-03-03 | Container Trac, Inc. | Automatic error detection for inventory tracking and management systems used at a shipping container yard |
| CN101894255A (en) * | 2010-06-13 | 2010-11-24 | 电子科技大学 | A container number location method based on wavelet transform |
| CN102024144A (en) * | 2010-11-23 | 2011-04-20 | 上海海事大学 | Container number identification method |
| US20190188505A1 (en) * | 2017-12-14 | 2019-06-20 | COM-IoT Technologies | Distracted driver detection |
| CN109272524A (en) * | 2018-08-27 | 2019-01-25 | 大连理工大学 | A Small Scale Point Cloud Noise Denoising Method Based on Threshold Segmentation |
| WO2021175403A1 (en) * | 2020-03-02 | 2021-09-10 | LASE Industrielle Lasertechnik GmbH | Container-loading system and method for monitoring operation therein |
| CN111814936A (en) * | 2020-08-27 | 2020-10-23 | 上海西井信息科技有限公司 | Container identification method, system, device and storage medium based on space scanning |
| WO2022057747A1 (en) * | 2020-09-17 | 2022-03-24 | 上海驭矩信息科技有限公司 | Container positioning method and apparatus based on multi-line laser data fusion |
| WO2023019709A1 (en) * | 2021-08-19 | 2023-02-23 | 中国铁路设计集团有限公司 | Automatic detection method of conductor height and pull-out value of overhead line system based on vehicle-mounted mobile laser point cloud |
| WO2023025030A1 (en) * | 2021-08-26 | 2023-03-02 | 上海交通大学 | Three-dimensional point cloud up-sampling method and system, device, and medium |
| CN114219842A (en) * | 2021-12-14 | 2022-03-22 | 东南大学 | Visual recognition, ranging and positioning methods in port container automatic loading and unloading operations |
| CN115222625A (en) * | 2022-07-15 | 2022-10-21 | 桂林理工大学 | A method for denoising LiDAR point cloud based on multi-scale noise |
| CN116110053A (en) * | 2023-04-13 | 2023-05-12 | 济宁能源发展集团有限公司 | Container surface information detection method based on image recognition |
Non-Patent Citations (1)
| Title |
|---|
| 薛萍;姚娟;邹学洲;王宏民;: "基于法矢修正的点云数据去噪平滑算法", 哈尔滨理工大学学报, no. 05 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116452467B (en) | 2023-09-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116664559B (en) | Machine vision-based memory bank damage rapid detection method | |
| CN116385472B (en) | Hardware stamping part deburring effect evaluation method | |
| CN117611563B (en) | Plastic mold production defect detection method based on image processing | |
| CN116110053B (en) | Container surface information detection method based on image recognition | |
| CN115187602A (en) | Injection molding part defect detection method and system based on image processing | |
| CN112927287B (en) | Phenotype data analysis method of target object, storage medium and terminal | |
| CN115994870B (en) | Image processing method for enhancing denoising | |
| CN118823481A (en) | A bridge detection image recognition method and system based on artificial intelligence | |
| CN116735612A (en) | Welding defect detection method for precise electronic components | |
| CN111047577A (en) | Abnormal urine red blood cell classification statistical method and system | |
| CN117874900B (en) | House construction engineering supervision method based on BIM technology | |
| CN117218029B (en) | Night dim light image intelligent processing method based on neural network | |
| CN117636075B (en) | Special equipment instrument identification system based on computer vision | |
| CN117115117B (en) | Pathological image recognition method based on small sample, electronic equipment and storage medium | |
| CN117635615B (en) | Defect detection method and system for realizing punching die based on deep learning | |
| CN109886320B (en) | Human femoral X-ray intelligent recognition method and system | |
| CN116452467B (en) | Container real-time positioning method based on laser data | |
| CN118691609B (en) | A method and device for detecting cracks in highway | |
| CN119600086A (en) | Intelligent detection method and system for cross-sectional ovality of medical wires | |
| CN118506030B (en) | Image defect target matching method and system based on geometric distance measurement | |
| CN117314899B (en) | Carbon fiber plate quality detection method based on image characteristics | |
| CN116900412B (en) | Steel plate equidistant automatic cutting regulation and control method based on visual processing | |
| CN113298838A (en) | Object contour line extraction method and system | |
| CN117095444B (en) | Image recognition method based on artificial intelligence | |
| CN118429300A (en) | Scoliosis identification method and system |
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 |