CN117408913A - Method, system and device for denoising point cloud of object to be measured - Google Patents
Method, system and device for denoising point cloud of object to be measured Download PDFInfo
- Publication number
- CN117408913A CN117408913A CN202311684329.XA CN202311684329A CN117408913A CN 117408913 A CN117408913 A CN 117408913A CN 202311684329 A CN202311684329 A CN 202311684329A CN 117408913 A CN117408913 A CN 117408913A
- Authority
- CN
- China
- Prior art keywords
- point cloud
- point
- blocks
- regional
- denoising
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 50
- 238000013507 mapping Methods 0.000 claims abstract description 44
- 230000012010 growth Effects 0.000 claims abstract description 28
- 238000001914 filtration Methods 0.000 claims abstract description 25
- 238000004458 analytical method Methods 0.000 claims abstract description 21
- 238000009826 distribution Methods 0.000 claims abstract description 13
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 4
- 238000004590 computer program Methods 0.000 claims description 13
- 238000007781 pre-processing Methods 0.000 claims description 9
- 238000012163 sequencing technique Methods 0.000 claims description 6
- 238000003860 storage Methods 0.000 claims description 5
- 238000007619 statistical method Methods 0.000 claims description 4
- 238000010586 diagram Methods 0.000 description 9
- 238000004141 dimensional analysis Methods 0.000 description 8
- 241000196324 Embryophyta Species 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 241000209094 Oryza Species 0.000 description 3
- 235000007164 Oryza sativa Nutrition 0.000 description 3
- 235000009566 rice Nutrition 0.000 description 3
- 238000000605 extraction Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 241000209140 Triticum Species 0.000 description 1
- 235000021307 Triticum Nutrition 0.000 description 1
- 240000008042 Zea mays Species 0.000 description 1
- 235000005824 Zea mays ssp. parviglumis Nutrition 0.000 description 1
- 235000002017 Zea mays subsp mays Nutrition 0.000 description 1
- 238000007792 addition Methods 0.000 description 1
- 230000001174 ascending effect Effects 0.000 description 1
- 235000005822 corn Nutrition 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001066 destructive effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000035040 seed growth Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- 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
-
- 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
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/7715—Feature extraction, e.g. by transforming the feature space, e.g. multi-dimensional scaling [MDS]; Mappings, e.g. subspace methods
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a denoising method, a denoising system and a denoising device for point clouds of an object to be measured, wherein the denoising method comprises the following steps: the original object point cloud to be detected is obtained through three-dimensional overall reconstruction, and the original object point cloud to be detected is obtained after pretreatment; performing region growth analysis according to the initial position information of the point cloud of the object to be detected, clustering by adopting a clustering algorithm based on the color distribution characteristics to obtain the point cloud of the region of the object to be detected and the clustering point cloud of the object to be detected, and calculating the density distribution of the clustering point cloud blocks; and mapping the regional point cloud blocks with the clustering point cloud blocks, filtering the regional point cloud blocks from the initial object point cloud to be detected, and finally denoising to obtain the denoised object point cloud to be detected if the number of points in the regional point cloud blocks is smaller than a preset first threshold value and the density of the corresponding clustering point cloud blocks is smaller than a preset second threshold value. The method combines the color distribution characteristics of the object to be detected, can be suitable for rapid denoising of the point cloud of the object to be detected, and has an important role in the three-dimensional phenotype analysis of the subsequent object to be detected.
Description
Technical Field
The invention relates to the technical field of three-dimensional analysis, in particular to a method, a system and a device for denoising point clouds of an object to be detected.
Background
At present, three-dimensional application is mainly focused on three-dimensional modeling, in agricultural application, the modeling cost is high due to complex and changeable application scenes and complex characteristics such as plant color, smoothness and the like, modeling data are more noisy and more holes are formed, and then point cloud analysis is difficult and the accuracy is low. The three-dimensional analysis task is mainly used in the field of vision SLAM application, three-dimensional reconstruction and obstacle detection are carried out by adopting a plurality of sensors, and corn crops are used for three-dimensional analysis application at present. In agricultural application, the current technology carries out curved surface reconstruction and skeleton extraction on crops by manually removing point cloud noise, and further calculates parameters such as crop area, included angle and the like; or the crop is subjected to destructive decomposition, the stems and the leaves are manually segmented, and the crop phenotype information is further analyzed; or based on basic characteristics of crops, analyzing the plant types of the crops by utilizing a skeleton extraction technology to obtain stem leaf phenotype parameters.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a method, a system and a device for denoising point cloud of an object to be measured.
In order to solve the technical problems, the invention is solved by the following technical scheme:
a denoising method for an object point cloud to be measured comprises the following steps:
performing three-dimensional reconstruction based on the full-view image information of the object to be detected to obtain an initial point cloud, and preprocessing the initial point cloud to obtain an initial point cloud;
performing region growth analysis based on the position information of the initial point cloud to obtain a region point cloud, and clustering based on the color information of the initial point cloud to obtain a clustered point cloud;
sorting according to the density of the clustering point cloud blocks in the clustering point cloud, and mapping the regional point cloud blocks with the clustering point cloud blocks by combining regional point cloud block information in the regional point cloud to obtain a point cloud block mapping relation;
and counting the density of the clustered point cloud blocks and the number of the regional point cloud blocks according to the point cloud block mapping relation, filtering the regional point cloud blocks meeting the conditions, and filtering the filtered regional point cloud to obtain the denoising point cloud.
As an implementation manner, the three-dimensional reconstruction is performed based on the full view image information of the object to be detected to obtain an initial point cloud, and the initial point cloud is obtained by preprocessing the initial point cloud, which includes the following steps:
acquiring full-view image information of an object to be detected, and reconstructing to obtain an original point cloud containing color information and position information according to a multi-view three-dimensional modeling algorithm;
and deleting the point cloud which does not accord with the growth characteristics of the object to be detected based on the position information of the original point cloud to obtain the initial point cloud.
As an implementation manner, the performing area growth analysis based on the position information of the initial point cloud to obtain an area point cloud, and clustering based on the color information of the initial point cloud to obtain a clustered point cloud includes the following steps:
dividing the point clouds with different distributions based on the point cloud position information of the object to be detected and combining the point cloud discrete degree and the point cloud uniformity characteristics to obtain regional point clouds;
and taking the color distribution characteristics of the object to be detected as priori information, setting the number of color clusters, and clustering the initial point cloud based on the color information to obtain a clustered point cloud.
As an implementation manner, the area growth analysis is performed based on the position information of the initial point cloud to obtain an area point cloud, and the method includes the following steps:
calculating the normal line and curvature of each point in the initial point cloud to obtain a normal line set and a curvature set, and selecting the corresponding point with the lowest curvature as a seed point after sequencing the curvature set;
setting a normal angle threshold, searching a neighborhood point of the seed point, calculating an angle between the normal line of the neighborhood point and the normal line of the seed point, and attributing the neighborhood point smaller than the normal angle threshold to the current area;
setting a curvature threshold value, analyzing the curvature of a neighborhood point, establishing a seed point sequence, adding the neighborhood point which is smaller than a normal angle threshold value and smaller than the curvature threshold value into the seed point sequence, deleting the current seed point, and calculating new seed points to continue growing;
repeating the above processes until the seed point sequence is empty, and completing the current region growth;
repeating the steps for the rest initial point cloud until all points are traversed, namely finishing the initial point cloud region growth analysis, and obtaining the region point cloud.
As an implementation manner, the sorting is performed according to the density of the cluster point cloud blocks in the cluster point cloud, the regional point cloud block and the cluster point cloud block are mapped by combining regional point cloud block information in the regional point cloud, and a point cloud block mapping relation is obtained, and the method comprises the following steps:
calculating the density of each cluster point cloud block in the cluster point cloud, and sequencing the cluster point cloud blocks according to the density;
mapping each regional point cloud block in the regional point cloud with the ordered cluster point cloud blocks, and determining the mapping relation between the regional point cloud blocks and the cluster point cloud blocks.
As an implementation manner, the counting of the density of the clustered point cloud blocks and the number of the regional point cloud blocks according to the point cloud block mapping relationship, filtering the regional point cloud blocks meeting the condition, and filtering the filtered regional point cloud to obtain the denoising point cloud, and the method comprises the following steps:
based on the mapping relation between the regional point cloud blocks and the clustering point cloud blocks, carrying out corresponding marking on the regional point cloud blocks and the clustering point cloud blocks;
analyzing the number of regional point cloud blocks and the density of the corresponding mark clustering point cloud blocks, and filtering the regional point cloud blocks which accord with the noise distribution characteristics;
and filtering the filtered regional point cloud, and removing the discrete distribution point cloud to obtain a denoising point cloud.
As an implementation manner, the filtering out the regional point cloud blocks meeting the condition includes the following steps:
setting a first threshold value, carrying out statistical analysis on the number of points contained in the regional point cloud blocks, and if the number of points is smaller than the first threshold value, further obtaining the density of the corresponding cluster point cloud blocks;
and setting a second threshold, and if the density of the corresponding cluster point cloud blocks is smaller than the second threshold, filtering the current area point cloud blocks, wherein the current area point cloud blocks do not accord with the growth characteristics of the object to be detected, and are noise point cloud.
A point cloud denoising system for an object to be measured comprises a point cloud acquisition module, a point cloud processing module, a point cloud mapping module and a point cloud denoising module.
The point cloud acquisition module is used for carrying out three-dimensional reconstruction based on the full-view image information of the object to be detected to obtain an original point cloud, and preprocessing the original point cloud to obtain the original point cloud;
the point cloud processing module performs area growth analysis based on the position information of the initial point cloud to obtain area point cloud, and performs clustering based on the color information of the initial point cloud to obtain clustered point cloud;
the point cloud mapping module is used for sorting according to the density of the clustered point cloud blocks in the clustered point cloud, and mapping the regional point cloud blocks with the clustered point cloud blocks by combining regional point cloud block information in the regional point cloud to obtain a point cloud block mapping relation;
and the point cloud denoising module counts the density of clustered point clouds and the number of regional point clouds through the point cloud block mapping relation, filters regional point clouds meeting the condition, and filters the filtered regional point clouds to obtain denoised point clouds.
As an implementation manner, the point cloud processing module is configured to:
calculating the normal line and curvature of each point in the initial point cloud to obtain a normal line set and a curvature set, and selecting the corresponding point with the lowest curvature as a seed point after sequencing the curvature set;
setting a normal angle threshold, searching a neighborhood point of the seed point, calculating an angle between the normal line of the neighborhood point and the normal line of the seed point, and attributing the neighborhood point smaller than the normal angle threshold to the current area;
setting a curvature threshold value, analyzing the curvature of a neighborhood point, establishing a seed point sequence, adding the neighborhood point which is smaller than a normal angle threshold value and smaller than the curvature threshold value into the seed point sequence, deleting the current seed point, and calculating new seed points to continue growing;
repeating the above processes until the seed point sequence is empty, and completing the current region growth;
repeating the steps for the rest initial point cloud until all points are traversed, namely finishing the initial point cloud region growth analysis, and obtaining the region point cloud.
As an implementation manner, the point cloud denoising module is configured to:
setting a first threshold value, carrying out statistical analysis on the number of points contained in the regional point cloud blocks, and if the number of points is smaller than the first threshold value, further obtaining the density of the corresponding cluster point cloud blocks;
and setting a second threshold, and if the density of the corresponding cluster point cloud blocks is smaller than the second threshold, filtering the current area point cloud blocks, wherein the current area point cloud blocks do not accord with the growth characteristics of the object to be detected, and are noise point cloud.
A computer readable storage medium storing a computer program which when executed by a processor performs the method of:
performing three-dimensional reconstruction based on the full-view image information of the object to be detected to obtain an initial point cloud, and preprocessing the initial point cloud to obtain an initial point cloud;
performing region growth analysis based on the position information of the initial point cloud to obtain a region point cloud, and clustering based on the color information of the initial point cloud to obtain a clustered point cloud;
sorting according to the density of the clustering point cloud blocks in the clustering point cloud, and mapping the regional point cloud blocks with the clustering point cloud blocks by combining regional point cloud block information in the regional point cloud to obtain a point cloud block mapping relation;
and counting the density of the clustered point cloud blocks and the number of the regional point cloud blocks according to the point cloud block mapping relation, filtering the regional point cloud blocks meeting the conditions, and filtering the filtered regional point cloud to obtain the denoising point cloud.
An object point cloud denoising device to be measured, comprising a memory, a processor and a computer program stored in the memory and running on the processor, wherein the processor realizes the following method when executing the computer program:
performing three-dimensional reconstruction based on the full-view image information of the object to be detected to obtain an initial point cloud, and preprocessing the initial point cloud to obtain an initial point cloud;
the point cloud processing module performs area growth analysis based on the position information of the initial point cloud to obtain area point cloud, and performs clustering based on the color information of the initial point cloud to obtain clustered point cloud;
the point cloud mapping module is used for sorting according to the density of the clustered point cloud blocks in the clustered point cloud, and mapping the regional point cloud blocks with the clustered point cloud blocks by combining regional point cloud block information in the regional point cloud to obtain a point cloud block mapping relation;
and the point cloud denoising module counts the density of clustered point clouds and the number of regional point clouds through the point cloud block mapping relation, filters regional point clouds meeting the condition, and filters the filtered regional point clouds to obtain denoised point clouds.
The invention has the remarkable technical effects due to the adoption of the technical scheme:
the method solves the noise problem in the three-dimensional analysis task, is suitable for automatic removal of point cloud noise of each object to be detected, has high efficiency, and has great significance for subsequent analysis of phenotype parameters of the object to be detected.
The method and the device realize rapid, accurate and efficient filtering of the point cloud noise of the object to be detected in the scene, and meet the requirements of researchers on three-dimensional analysis of the object to be detected.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only some embodiments of the invention, and that other drawings can be obtained according to these drawings without inventive faculty for a person skilled in the art.
FIG. 1 is a schematic flow chart of the method of the present invention;
FIG. 2 is an overall schematic of the system of the present invention;
FIG. 3 is a schematic diagram of the architecture of a three-dimensional phenotyping system of the invention;
FIG. 4 is a schematic diagram of an initial point cloud;
fig. 5 is a denoising point cloud schematic diagram.
Detailed Description
The present invention will be described in further detail with reference to the following examples, which are illustrative of the present invention and are not intended to limit the present invention thereto.
Example 1:
a denoising method for an object point cloud to be measured, as shown in fig. 1, comprises the following steps:
s100: performing three-dimensional reconstruction based on the full-view image information of the object to be detected to obtain an initial point cloud, and preprocessing the initial point cloud to obtain an initial point cloud, as shown in fig. 4;
s200: performing region growth analysis based on the position information of the initial point cloud to obtain a region point cloud, and clustering based on the color information of the initial point cloud to obtain a clustered point cloud;
s300: sorting according to the density of the clustering point cloud blocks in the clustering point cloud, and mapping the regional point cloud blocks with the clustering point cloud blocks by combining regional point cloud block information in the regional point cloud to obtain a point cloud block mapping relation;
s400: and counting the density of the clustered point cloud blocks and the number of the regional point cloud blocks according to the point cloud block mapping relation, filtering the regional point cloud blocks meeting the conditions, and filtering the filtered regional point cloud to obtain denoising point cloud, as shown in fig. 5.
The method solves the noise problem in the three-dimensional analysis task, realizes the rapid, accurate and efficient filtering of the point cloud noise of the object to be detected in the scene, meets the requirement of researchers on the three-dimensional analysis of the object to be detected, and has great significance for the subsequent analysis of the phenotype parameters.
In step S100, three-dimensional reconstruction is performed based on the full view image information of the object to be detected to obtain an initial point cloud, and the initial point cloud is preprocessed to obtain the initial point cloud, including the following steps:
s110: the three-dimensional reconstruction system is shown in fig. 3, wherein 301 is an object to be detected containing a calibration object, in this embodiment, the object to be detected is a rice pot plant, 303 is a calibration object, 302 is a turntable for driving the object to rotate at a constant speed of 360 degrees, 304 is a monocular RGB camera for image acquisition, and 305 is a server for generating a rice point cloud; of course, in other embodiments, other objects may be used, such as other crops like wheat, or still or other objects;
s120: driving 301 the rice pot plant containing the calibration object through a 302 turntable, fixing a camera, and acquiring an all-visual-angle image of the object to be detected;
s130: carrying out three-dimensional reconstruction and three-dimensional analysis on the acquired full-view image to obtain an original point cloud;
s140: filtering the original point cloud from infinity to obtain an original point cloudAs shown in fig. 4, the specific expression is:
wherein,is->Any one of (4)>For the point abscissa, +.>For the ordinate of the point, +.>For the point vertical coordinate, +.>、/>、/>The values are separated for each channel of the point in RGB image space.
In step S200, performing region growth analysis based on the position information of the initial point cloud to obtain a region point cloud, and performing clustering based on the color information of the initial point cloud to obtain a clustered point cloud, including the steps of:
s210: based on initial point cloudPerforming region growth according to the position distribution to obtain region point cloud and region point cloud block set ∈ ->Comprising the following steps:
step1: according toCalculating the normal corresponding to each point>Calculating the corresponding curvature of each point；
Step2: defining seed point cloud proximity point cloud search rulesIn the method, a k neighbor search method is adopted, wherein the k value is set to be k=30;
step3: setting a normal angleThreshold valueAnd curvature threshold->In the present method set to +.>、As the judging basis of seed growth stop;
step4: sorting the curvatures according to ascending order, selecting the lowest curvature point as a seed point, and comparing the neighborhood point with the seed point;
step5: according to the threshold value of the normal angleSearching a neighborhood point of the current seed point, calculating an included angle between the normal line of the neighborhood point and the normal line of the current seed point, and adding the neighborhood point smaller than a threshold value into the current area;
step6: according to the threshold value of curvatureChecking the curvature of each neighborhood point to be less than +.>Adding the neighborhood points of the seed points into the seed point sequence, deleting the current seed points, and continuing to grow with new seed points;
step7: taking the point meeting Step5 and Step6 simultaneously as a seed point;
step8: repeating the above process until the seed point sequence is emptied, then one region grows to completion, repeating the above steps for the remaining points until all points are traversed.
The method comprises the following specific steps:
input:
initial point cloudNormal set { N }, curvature set{ C }, k neighbor search rule +.>Normal threshold->Threshold of curvature->；
Initializing: the region list R is set to be empty and the available point cloud list A is set to be empty, wherein；
The algorithm flow is as follows:
while A non-empty do
Order the
Order the
Taking the point of least curvature from A
Updating
Updating
Will beReject from A
for i=0 to size
SearchingK-nearest neighbor->
for j=0 to size
Taking outOne at the middle->
if&&/>Then
Will beReject from A
ifThen
endif
endif
end for
end for
Adding the current region to the global list block:
end while
output R
S220: according to the color distribution characteristics of the object to be detected, the clustering number is set to be clusters=5 based onKmeans clustering is carried out by containing color information to obtain cluster point cloud +.>The method comprises the following specific steps:
step1: inputting an initial point cloudClustering numbers clusters;
step2: randomly selecting clusterings of center points;
step3: assigning each data point to its nearest center point according to the point cloud color;
step4: recalculating the average of the distances from the points in each class to the center points of the class
Step5: assigning each data to its nearest central point;
step6: repeating steps 4 and 5 until all observations are no longer assigned or the maximum number of iterations is reached (the maximum number of iterations is set to 120);
step7: output K, where，/>。
In step S300, sorting is performed according to the density of the cluster point cloud blocks in the cluster point cloud, and the regional point cloud blocks and the cluster point cloud blocks are mapped by combining the regional point cloud block information in the regional point cloud to obtain a point cloud block mapping relation, which comprises the following steps:
according to the cluster point cloud obtained in the step S200, the cluster point cloud block density is calculated and ordered, and the method comprises the following steps:
inputting cluster point cloud { K }, neighborhood searching radius r, initial density list；
for index=0 tosize{K}
Taking the index-th point set K from K
Initializing pointsum=0, distancesum=0
for i=0 to size k
for j=0 to size
pointSum++
distanceSum+=distance()
Wherein,for the set of points within the k radius r, according to +.>After descending order of K, obtain ordered Density +.>Cluster cloud block set>；
Region-based cloud block setAnd cluster point cloud block set->Mapping is performed when->Mapping to->The medium density is less than the secondThreshold->And->The number of points contained is smaller than a first threshold +.>When in use, will->Filtering from the initial point cloud to obtain filtered point cloud +.>The method comprises the following specific steps of:
input area block point cloud collectionCluster cloud block set->A threshold number of points in a point cloud block num=1000;
for i=0 to size{R}
point cloud block from { R }, point cloud block is obtained
Initializing an empty list { r = {0}, list size n
for j=0 to size
From the slaveGet one little->
for k to size
if
break
end if
end for
end for
if
if size>1000
end if
end if
end for
And (3) outputting:
after the denoising process, the point cloudThere are more discrete points, the pair +.>Removing isolated points to obtain result point cloud->The resulting point cloud is the denoising point cloud, as shown in fig. 5.
Example 2:
the system for denoising the point cloud of the object to be measured comprises a point cloud acquisition module 100, a point cloud processing module 200, a point cloud mapping module 300 and a point cloud denoising module 400 as shown in fig. 2;
the point cloud acquisition module 100 performs three-dimensional reconstruction based on the full view image information of the object to be detected to obtain an initial point cloud, and performs preprocessing on the initial point cloud to obtain the initial point cloud;
the point cloud processing module 200 performs area growth analysis based on the position information of the initial point cloud to obtain an area point cloud, and performs clustering based on the color information of the initial point cloud to obtain a clustered point cloud;
the point cloud mapping module 300 performs sorting according to the density of the clustered point cloud blocks in the clustered point cloud, and maps the regional point cloud blocks with the clustered point cloud blocks by combining regional point cloud block information in the regional point cloud to obtain a point cloud block mapping relation;
the point cloud denoising module 400 counts the density of clustered point clouds and the number of regional point clouds according to the point cloud mapping relationship, filters regional point clouds meeting the condition, and filters the filtered regional point clouds to obtain denoised point clouds.
All changes and modifications that come within the spirit and scope of the invention are desired to be protected and all equivalent thereto are deemed to be within the scope of the invention.
In this specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different manner from other embodiments, so that identical and similar parts of each embodiment are mutually referred to.
It will be apparent to those skilled in the art that embodiments of the present invention may be provided as a method, apparatus, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, terminal devices (systems), and computer program products according to the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
It should be noted that:
reference in the specification to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the invention. Thus, the appearances of the phrase "one embodiment" or "an embodiment" in various places throughout this specification are not necessarily all referring to the same embodiment.
In addition, the specific embodiments described in the present specification may differ in terms of parts, shapes of components, names, and the like. All equivalent or simple changes of the structure, characteristics and principle according to the inventive concept are included in the protection scope of the present invention. Those skilled in the art may make various modifications or additions to the described embodiments or substitutions in a similar manner without departing from the scope of the invention as defined in the accompanying claims.
Claims (12)
1. The object point cloud denoising method to be measured is characterized by comprising the following steps:
performing three-dimensional reconstruction based on the full-view image information of the object to be detected to obtain an initial point cloud, and preprocessing the initial point cloud to obtain an initial point cloud;
performing region growth analysis based on the position information of the initial point cloud to obtain a region point cloud, and clustering based on the color information of the initial point cloud to obtain a clustered point cloud;
sorting according to the density of the clustering point cloud blocks in the clustering point cloud, and mapping the regional point cloud blocks with the clustering point cloud blocks by combining regional point cloud block information in the regional point cloud to obtain a point cloud block mapping relation;
and counting the density of the clustered point cloud blocks and the number of the regional point cloud blocks according to the point cloud block mapping relation, filtering the regional point cloud blocks meeting the conditions, and filtering the filtered regional point cloud to obtain the denoising point cloud.
2. The method for denoising the point cloud of the object to be measured according to claim 1, wherein the method for denoising the point cloud of the object to be measured based on the full view image information of the object to be measured comprises the steps of:
acquiring full-view image information of an object to be detected, and reconstructing to obtain an original point cloud containing color information and position information according to a multi-view three-dimensional modeling algorithm;
and deleting the point cloud which does not accord with the growth characteristics of the object to be detected based on the position information of the original point cloud to obtain the initial point cloud.
3. The method for denoising point cloud of object to be measured according to claim 1, wherein the method for denoising point cloud of object to be measured comprises the steps of:
dividing the point clouds with different distributions based on the point cloud position information of the object to be detected and combining the point cloud discrete degree and the point cloud uniformity characteristics to obtain regional point clouds;
and taking the color distribution characteristics of the object to be detected as priori information, setting the number of color clusters, and clustering the initial point cloud based on the color information to obtain a clustered point cloud.
4. The method for denoising the point cloud of the object to be measured according to claim 3, wherein the area growth analysis is performed based on the position information of the initial point cloud to obtain the area point cloud, and the method comprises the following steps:
calculating the normal line and curvature of each point in the initial point cloud to obtain a normal line set and a curvature set, and selecting the corresponding point with the lowest curvature as a seed point after sequencing the curvature set;
setting a normal angle threshold, searching a neighborhood point of the seed point, calculating an angle between the normal line of the neighborhood point and the normal line of the seed point, and attributing the neighborhood point smaller than the normal angle threshold to the current area;
setting a curvature threshold value, analyzing the curvature of a neighborhood point, establishing a seed point sequence, adding the neighborhood point which is smaller than a normal angle threshold value and smaller than the curvature threshold value into the seed point sequence, deleting the current seed point, and calculating new seed points to continue growing;
repeating the above processes until the seed point sequence is empty, and completing the current region growth;
repeating the steps for the rest initial point cloud until all points are traversed, namely finishing the initial point cloud region growth analysis, and obtaining the region point cloud.
5. The method for denoising the point cloud of the object to be measured according to claim 1, wherein the method is characterized in that the method comprises the following steps of sorting according to the density of the cluster point clouds in the cluster point cloud, mapping the regional point clouds with the cluster point clouds by combining regional point cloud information in the regional point cloud to obtain a point cloud mapping relation, and the method comprises the following steps:
calculating the density of each cluster point cloud block in the cluster point cloud, and sequencing the cluster point cloud blocks according to the density;
mapping each regional point cloud block in the regional point cloud with the ordered cluster point cloud blocks, and determining the mapping relation between the regional point cloud blocks and the cluster point cloud blocks.
6. The method for denoising the point cloud of the object to be measured according to claim 1, wherein the method for denoising the point cloud of the object to be measured is characterized by comprising the steps of:
based on the mapping relation between the regional point cloud blocks and the clustering point cloud blocks, carrying out corresponding marking on the regional point cloud blocks and the clustering point cloud blocks;
analyzing the number of regional point cloud blocks and the density of the corresponding mark clustering point cloud blocks, and filtering the regional point cloud blocks which accord with the noise distribution characteristics;
and filtering the filtered regional point cloud, and removing the discrete distribution point cloud to obtain a denoising point cloud.
7. The method for denoising the point cloud of the object to be measured according to claim 6, wherein the filtering the point cloud blocks of the area meeting the condition comprises the following steps:
setting a first threshold value, carrying out statistical analysis on the number of points contained in the regional point cloud blocks, and if the number of points is smaller than the first threshold value, further obtaining the density of the corresponding cluster point cloud blocks;
and setting a second threshold, and if the density of the corresponding cluster point cloud blocks is smaller than the second threshold, filtering the current area point cloud blocks, wherein the current area point cloud blocks do not accord with the growth characteristics of the object to be detected, and are noise point cloud.
8. The point cloud denoising system for the object to be measured is characterized by comprising a point cloud acquisition module, a point cloud processing module, a point cloud mapping module and a point cloud denoising module;
the point cloud acquisition module is used for carrying out three-dimensional reconstruction based on the full-view image information of the object to be detected to obtain an original point cloud, and preprocessing the original point cloud to obtain the original point cloud;
the point cloud processing module performs area growth analysis based on the position information of the initial point cloud to obtain area point cloud, and performs clustering based on the color information of the initial point cloud to obtain clustered point cloud;
the point cloud mapping module is used for sorting according to the density of the clustered point cloud blocks in the clustered point cloud, and mapping the regional point cloud blocks with the clustered point cloud blocks by combining regional point cloud block information in the regional point cloud to obtain a point cloud block mapping relation;
and the point cloud denoising module counts the density of clustered point clouds and the number of regional point clouds through the point cloud block mapping relation, filters regional point clouds meeting the condition, and filters the filtered regional point clouds to obtain denoised point clouds.
9. The system for denoising a point cloud of an object to be measured according to claim 8, wherein the point cloud processing module is configured to:
calculating the normal line and curvature of each point in the initial point cloud to obtain a normal line set and a curvature set, and selecting the corresponding point with the lowest curvature as a seed point after sequencing the curvature set;
setting a normal angle threshold, searching a neighborhood point of the seed point, calculating an angle between the normal line of the neighborhood point and the normal line of the seed point, and attributing the neighborhood point smaller than the normal angle threshold to the current area;
setting a curvature threshold value, analyzing the curvature of a neighborhood point, establishing a seed point sequence, adding the neighborhood point which is smaller than a normal angle threshold value and smaller than the curvature threshold value into the seed point sequence, deleting the current seed point, and calculating new seed points to continue growing;
repeating the above processes until the seed point sequence is empty, and completing the current region growth;
repeating the steps for the rest initial point cloud until all points are traversed, namely finishing the initial point cloud region growth analysis, and obtaining the region point cloud.
10. The object to be measured point cloud denoising system according to claim 8, wherein the point cloud denoising module is configured to:
setting a first threshold value, carrying out statistical analysis on the number of points contained in the regional point cloud blocks, and if the number of points is smaller than the first threshold value, further obtaining the density of the corresponding cluster point cloud blocks;
and setting a second threshold, and if the density of the corresponding cluster point cloud blocks is smaller than the second threshold, filtering the current area point cloud blocks, wherein the current area point cloud blocks do not accord with the growth characteristics of the object to be detected, and are noise point cloud.
11. A computer readable storage medium storing a computer program, which when executed by a processor implements the method of any one of claims 1 to 7.
12. A device for denoising an object point cloud to be measured, comprising a memory, a processor and a computer program stored in the memory and running on the processor, wherein the processor implements the method according to any one of claims 1 to 7 when executing the computer program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311684329.XA CN117408913B (en) | 2023-12-11 | 2023-12-11 | Method, system and device for denoising point cloud of object to be measured |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311684329.XA CN117408913B (en) | 2023-12-11 | 2023-12-11 | Method, system and device for denoising point cloud of object to be measured |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117408913A true CN117408913A (en) | 2024-01-16 |
| CN117408913B CN117408913B (en) | 2024-02-23 |
Family
ID=89487365
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311684329.XA Active CN117408913B (en) | 2023-12-11 | 2023-12-11 | Method, system and device for denoising point cloud of object to be measured |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117408913B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117808703A (en) * | 2024-02-29 | 2024-04-02 | 南京航空航天大学 | Multi-scale large-scale component assembly gap point cloud filtering method |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105354883A (en) * | 2015-11-25 | 2016-02-24 | 武汉大学 | 3ds Max fast and precise three-dimensional modeling method and system based on point cloud |
| CN111145129A (en) * | 2020-04-03 | 2020-05-12 | 速度时空信息科技股份有限公司 | Point cloud denoising method based on hyper-voxels |
| WO2021012254A1 (en) * | 2019-07-25 | 2021-01-28 | 深圳市大疆创新科技有限公司 | Target detection method, system, and mobile platform |
| WO2021017471A1 (en) * | 2019-07-30 | 2021-02-04 | 五邑大学 | Point cloud filtering method based on image processing, apparatus, and storage medium |
| WO2021097618A1 (en) * | 2019-11-18 | 2021-05-27 | 深圳市大疆创新科技有限公司 | Point cloud segmentation method and system, and computer storage medium |
| WO2021142996A1 (en) * | 2020-01-13 | 2021-07-22 | 五邑大学 | Point cloud denoising method, system, and device employing image segmentation, and storage medium |
| CN113192179A (en) * | 2021-04-28 | 2021-07-30 | 沈阳工业大学 | Three-dimensional reconstruction method based on binocular stereo vision |
| CN113902937A (en) * | 2021-09-23 | 2022-01-07 | 中国科学院国家空间科学中心 | Self-adaptive target detection method and system for TOF point cloud |
| WO2022016310A1 (en) * | 2020-07-20 | 2022-01-27 | 深圳元戎启行科技有限公司 | Point cloud data-based three-dimensional reconstruction method and apparatus, and computer device |
| CN116486268A (en) * | 2023-05-05 | 2023-07-25 | 中交信息技术国家工程实验室有限公司 | Laser point cloud ground filtering method and device |
| CN116486395A (en) * | 2022-09-21 | 2023-07-25 | 北京理工大学 | Spatial Obstacle Discrimination Method Based on Gaussian Mixture Model and Depth-Color Image |
| CN116612035A (en) * | 2023-05-30 | 2023-08-18 | 同济大学 | A Noise Reduction Method for 3D Point Cloud Data of Tunnel Surrounding Rock |
| WO2023179405A1 (en) * | 2022-03-21 | 2023-09-28 | 深圳市正浩创新科技股份有限公司 | Obstacle identification method, device and storage medium |
-
2023
- 2023-12-11 CN CN202311684329.XA patent/CN117408913B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105354883A (en) * | 2015-11-25 | 2016-02-24 | 武汉大学 | 3ds Max fast and precise three-dimensional modeling method and system based on point cloud |
| WO2021012254A1 (en) * | 2019-07-25 | 2021-01-28 | 深圳市大疆创新科技有限公司 | Target detection method, system, and mobile platform |
| WO2021017471A1 (en) * | 2019-07-30 | 2021-02-04 | 五邑大学 | Point cloud filtering method based on image processing, apparatus, and storage medium |
| WO2021097618A1 (en) * | 2019-11-18 | 2021-05-27 | 深圳市大疆创新科技有限公司 | Point cloud segmentation method and system, and computer storage medium |
| WO2021142996A1 (en) * | 2020-01-13 | 2021-07-22 | 五邑大学 | Point cloud denoising method, system, and device employing image segmentation, and storage medium |
| CN111145129A (en) * | 2020-04-03 | 2020-05-12 | 速度时空信息科技股份有限公司 | Point cloud denoising method based on hyper-voxels |
| WO2022016310A1 (en) * | 2020-07-20 | 2022-01-27 | 深圳元戎启行科技有限公司 | Point cloud data-based three-dimensional reconstruction method and apparatus, and computer device |
| CN113192179A (en) * | 2021-04-28 | 2021-07-30 | 沈阳工业大学 | Three-dimensional reconstruction method based on binocular stereo vision |
| CN113902937A (en) * | 2021-09-23 | 2022-01-07 | 中国科学院国家空间科学中心 | Self-adaptive target detection method and system for TOF point cloud |
| WO2023179405A1 (en) * | 2022-03-21 | 2023-09-28 | 深圳市正浩创新科技股份有限公司 | Obstacle identification method, device and storage medium |
| CN116486395A (en) * | 2022-09-21 | 2023-07-25 | 北京理工大学 | Spatial Obstacle Discrimination Method Based on Gaussian Mixture Model and Depth-Color Image |
| CN116486268A (en) * | 2023-05-05 | 2023-07-25 | 中交信息技术国家工程实验室有限公司 | Laser point cloud ground filtering method and device |
| CN116612035A (en) * | 2023-05-30 | 2023-08-18 | 同济大学 | A Noise Reduction Method for 3D Point Cloud Data of Tunnel Surrounding Rock |
Non-Patent Citations (4)
| Title |
|---|
| 李幸刚;张亚萍;杨雨薇;: "一种基于邻域扩展聚类的去噪算法", 系统仿真学报, no. 11, 8 November 2017 (2017-11-08) * |
| 牛晓静;王美丽;何东健;: "一种聚类与滤波融合的点云去噪平滑方法", 计算机应用与软件, no. 10, 15 October 2016 (2016-10-15) * |
| 王?;李松;田昕;张智宇;张文豪;: "方向自适应的星载光子计数激光测高植被冠层高度估算", 红外与毫米波学报, no. 03, 15 June 2020 (2020-06-15) * |
| 王雯;任小玲;陈逍遥;: "一种改进的区域增长彩色3D点云分割算法", 国外电子测量技术, no. 11, 15 November 2018 (2018-11-15) * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117808703A (en) * | 2024-02-29 | 2024-04-02 | 南京航空航天大学 | Multi-scale large-scale component assembly gap point cloud filtering method |
| CN117808703B (en) * | 2024-02-29 | 2024-05-10 | 南京航空航天大学 | A multi-scale large-scale component assembly gap point cloud filtering method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117408913B (en) | 2024-02-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109146948B (en) | Vision-based analysis method of crop growth phenotype parameter quantification and yield correlation | |
| CN110135341B (en) | Weed identification method, device and terminal equipment | |
| Bao et al. | Field-based architectural traits characterisation of maize plant using time-of-flight 3D imaging | |
| CN111724433B (en) | Crop phenotype parameter extraction method and system based on multi-view vision | |
| CN105701478B (en) | The method and apparatus of rod-shaped Objects extraction | |
| CN111598915A (en) | Point cloud single wood segmentation method, device, equipment and computer readable medium | |
| CN111598780B (en) | Terrain adaptive interpolation filtering method suitable for airborne LiDAR point cloud | |
| CN117408913B (en) | Method, system and device for denoising point cloud of object to be measured | |
| CN115496796B (en) | Method and system for measuring trunk volume through laser point cloud | |
| JP5745290B2 (en) | Image processing apparatus, cell classification apparatus, incubator, image processing method, cell classification method, image processing program, and cell classification program | |
| CN113269791A (en) | Point cloud segmentation method based on edge judgment and region growth | |
| CN117437271B (en) | Three-dimensional target measuring method based on artificial intelligence | |
| Wang et al. | DFSP: A fast and automatic distance field-based stem-leaf segmentation pipeline for point cloud of maize shoot | |
| CN114519712A (en) | Point cloud data processing method and device, terminal equipment and storage medium | |
| CN116523898A (en) | Tobacco phenotype character extraction method based on three-dimensional point cloud | |
| CN111950589B (en) | Optimal segmentation method of point cloud region growth combined with K-means clustering | |
| CN117710243A (en) | Point cloud denoising method and device, electronic equipment and readable storage medium | |
| CN117115212A (en) | Automatic detection model pose abnormality and registration method | |
| CN117315049B (en) | Three-dimensional overall measurement scale calibration method, system and device | |
| CN116310849A (en) | Tree point cloud monomerization extraction method based on three-dimensional morphological characteristics | |
| CN117495891A (en) | Point cloud edge detection method and device and electronic equipment | |
| CN118135207A (en) | Point cloud leaf segmentation method and system based on geometric interaction and adaptive graph convolution | |
| CN117095145A (en) | Training method and terminal of tooth grid segmentation model | |
| CN117349273A (en) | A power data fusion method and equipment | |
| CN117315002B (en) | Method, system and device for measuring three-dimensional phenotype of object to be measured |
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 |