CN116468761A - Registration method, equipment and storage medium based on probability distribution distance feature description - Google Patents
Registration method, equipment and storage medium based on probability distribution distance feature description Download PDFInfo
- Publication number
- CN116468761A CN116468761A CN202310383322.8A CN202310383322A CN116468761A CN 116468761 A CN116468761 A CN 116468761A CN 202310383322 A CN202310383322 A CN 202310383322A CN 116468761 A CN116468761 A CN 116468761A
- Authority
- CN
- China
- Prior art keywords
- point cloud
- probability distribution
- target
- source
- source point
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000009826 distribution Methods 0.000 title claims abstract description 270
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000003860 storage Methods 0.000 title claims abstract description 15
- 239000011159 matrix material Substances 0.000 claims abstract description 158
- 238000002156 mixing Methods 0.000 claims abstract description 82
- 238000004364 calculation method Methods 0.000 claims abstract description 40
- 230000011218 segmentation Effects 0.000 claims description 76
- 239000013598 vector Substances 0.000 claims description 34
- 238000010586 diagram Methods 0.000 claims description 26
- 239000000203 mixture Substances 0.000 claims description 10
- 238000013519 translation Methods 0.000 claims description 9
- 230000009466 transformation Effects 0.000 claims description 8
- 238000007781 pre-processing Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 abstract description 5
- 238000004891 communication Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 230000007613 environmental effect Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
-
- 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
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- 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
- G06T2207/20081—Training; Learning
-
- 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
- G06T2207/20084—Artificial neural networks [ANN]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of data processing, and discloses a registration method, equipment and a storage medium based on probability distribution distance feature description, wherein the method comprises the following steps: generating a source point cloud mixing probability distribution map according to the source point cloud, and generating a target point cloud mixing probability distribution map according to the target point cloud; respectively carrying out nonlinear calculation on the source point cloud mixed probability distribution map and the target point cloud mixed probability distribution map, and taking calculation results as source point cloud characteristics and target point cloud characteristics; generating a point cloud corresponding relation matrix according to the source point cloud characteristics and the target point cloud characteristics; determining a registration result from the source point cloud to the target point cloud according to the point cloud corresponding relation matrix; by the method, the point cloud corresponding relation matrix is generated according to the source point cloud characteristics and the target point cloud characteristics, and then the registration result from the source point cloud to the target point cloud is determined according to the point cloud corresponding relation matrix, so that the accuracy, the stability and the generalization capability of the registration point cloud can be effectively improved.
Description
Technical Field
The present invention relates to the field of data processing technologies, and in particular, to a registration method, apparatus, and storage medium based on probability distribution distance feature description.
Background
Under the double development and promotion of depth sensing technology and metauniverse industry, a three-dimensional model is widely applied to the fields of daily life such as entertainment and consumption, is hopeful to become a center of future world digital information circulation, and three-dimensional point cloud data based on depth images are usually obtained by sampling at a plurality of view angles, and often accompanies the problems of noise, environmental shielding, target overlapping and the like. Therefore, the point cloud registration is the most important data processing link before three-dimensional imaging and application thereof, the related technology commonly used for registration is an algorithm based on iterative closest points (Iterative Closest Point, ICP), the requirements of high overlapping rate, no noise, complete data and the like are required to be met during point cloud registration, so that popularization of ICP is restricted, in addition, the registration algorithm based on a probability model converts the registration problem into the maximum estimation of solving probability density, the noise robustness is good, but registration failure easily occurs under the condition of data missing, coarse registration is performed based on local feature point extraction and feature description thereof, then iterative calculation is used for fine registration, the coarse-to-fine registration mode can effectively reduce calculation cost, the data overlapping rate and the integrity robustness are good, but the interference of noise is easy, the point feature lack of global characteristics easily causes a mismatching relationship, and finally a series of problems of low accuracy, poor stability, weak generalization capability and the like of the point cloud are caused.
The foregoing is provided merely for the purpose of facilitating understanding of the technical solutions of the present invention and is not intended to represent an admission that the foregoing is prior art.
Disclosure of Invention
The invention mainly aims to provide a registration method, equipment and storage medium based on probability distribution distance feature description, and aims to solve the technical problems of low accuracy, poor stability and weak generalization capability of the point cloud registration in the prior art.
In order to achieve the above object, the present invention provides a registration method based on probability distribution distance feature description, the registration method based on probability distribution distance feature description comprising the steps of:
generating a source point cloud mixing probability distribution map according to the source point cloud, and generating a target point cloud mixing probability distribution map according to the target point cloud;
respectively carrying out nonlinear calculation on the source point cloud mixing probability distribution map and the target point cloud mixing probability distribution map, and respectively taking calculation results as a source point cloud characteristic and a target point cloud characteristic;
generating a point cloud corresponding relation matrix according to the source point cloud characteristics and the target point cloud characteristics;
and determining a registration result from the source point cloud to the target point cloud according to the point cloud corresponding relation matrix.
Optionally, the generating a source point cloud mixing probability distribution map according to the source point cloud and generating a target point cloud mixing probability distribution map according to the target point cloud includes:
acquiring an initial source point cloud and an initial point cloud, and calibrating the initial source point cloud and the initial point cloud;
respectively carrying out quality enhancement on the calibrated initial source point cloud and initial point cloud to obtain a source point cloud and a target point cloud;
generating a source point cloud model according to the source point cloud, and generating a target point cloud model according to the target point cloud;
semantic segmentation is respectively carried out on the source point cloud model and the target point cloud model through a target proximity algorithm, so that a source point cloud segmentation result and a target point cloud segmentation result are obtained;
and generating a source point cloud mixing probability distribution map according to the source point cloud segmentation result, and generating a target point cloud mixing probability distribution map according to the target point cloud segmentation result.
Optionally, the generating a source point cloud mixing probability distribution map according to the source point cloud segmentation result, and generating a target point cloud mixing probability distribution map according to the target point cloud segmentation result include:
calculating the source point cloud segmentation result to obtain Gaussian distribution of a source point cloud segmentation part and neighborhood Gaussian distribution of each point of the source point cloud;
Splicing the neighborhood Gaussian distribution of each point of the source point cloud and the Gaussian distribution of the source point cloud segmentation part to which each point of the source point cloud belongs to obtain the source point cloud mixed Gaussian distribution;
generating a source point cloud mixing probability distribution diagram according to the source point cloud mixing Gaussian distribution and a preset diagram structure;
calculating the target point cloud segmentation result to obtain Gaussian distribution of a target point cloud segmentation part and neighborhood Gaussian distribution of each point of the target point cloud;
splicing the neighborhood Gaussian distribution of each point of the target point cloud and the Gaussian distribution of the target point cloud segmentation part to which each point of the target point cloud belongs to obtain the target point cloud Gaussian mixture distribution;
and generating a target point cloud mixing probability distribution map according to the target point cloud mixing Gaussian distribution and a preset map structure.
Optionally, the performing nonlinear calculation on the source point cloud mixing probability distribution map and the target point cloud mixing probability distribution map respectively, and taking the calculation result as a source point cloud feature and a target point cloud feature, including:
calculating a first probability distribution distance between a target vertex in the source point cloud mixed probability distribution diagram and each vertex in the target point cloud mixed probability distribution diagram;
generating a source vertex vector according to the first probability distribution distance and the number of source vertices;
Calculating a second probability distribution distance between a target vertex and a source point cloud segmentation part in the source point cloud mixed probability distribution map;
generating a first vertex matrix according to the second probability distribution distance and the number of source vertices;
calculating a third probability distribution distance between a target vertex and a target point cloud segmentation part in the source point cloud mixed probability distribution map;
generating a second vertex matrix according to the third probability distribution distance and the number of source vertices;
expanding the target vertex vector to the first vertex matrix and the second vertex matrix respectively;
generating a source point cloud feature matrix according to the expanded first vertex matrix, the expanded second vertex matrix and the target vertex vector;
calculating a fourth probability distribution distance between a target vertex in the target point cloud mixing probability distribution diagram and each vertex in the source point cloud mixing probability distribution diagram;
generating a target vertex vector according to the fourth probability distribution distance and the number of the target vertices;
calculating a fifth probability distribution distance between a target vertex and a target point cloud segmentation part in the target point cloud mixed probability distribution map;
generating a third vertex matrix according to the fifth probability distribution distance and the number of target vertices;
calculating a sixth probability distribution distance between a target vertex and a source point cloud segmentation part in the target point cloud mixed probability distribution map;
Generating a fourth vertex matrix according to the sixth probability distribution distance and the number of target vertices;
expanding the target vertex vector to the third vertex matrix and the fourth vertex matrix respectively;
generating a target point cloud feature matrix according to the expanded third vertex matrix and fourth vertex matrix and the target vertex vector;
and obtaining source point cloud characteristics and target point cloud characteristics according to the source point cloud characteristic matrix and the target point cloud characteristic matrix.
Optionally, the obtaining the source point cloud feature and the target point cloud feature according to the source point cloud feature matrix and the target point cloud feature matrix includes:
performing multi-layer perceptron calculation on the source point cloud characteristic matrix and the target point cloud characteristic matrix respectively to obtain a source point cloud characteristic description matrix and a target point cloud characteristic description matrix;
preprocessing the source point cloud characteristic description matrix and the target point cloud characteristic description matrix respectively to obtain source point cloud characteristic description and target point cloud characteristic description;
and obtaining the source point cloud characteristics and the target point cloud characteristics according to the source point cloud characteristics description and the target point cloud characteristics description.
Optionally, the generating a point cloud correspondence matrix according to the source point cloud feature and the target point cloud feature includes:
Performing dot product calculation on the source point cloud features and the target point cloud features to obtain a plurality of feature similarities;
constructing a feature similarity matrix according to the feature similarities;
and performing row-column calculation on the feature similarity matrix through a sinkhorn algorithm to obtain a point cloud corresponding relation matrix.
Optionally, the determining, according to the point cloud correspondence matrix, a registration result of the source point cloud to the target point cloud includes:
obtaining a matched point cloud pair according to the point cloud corresponding relation matrix;
calculating the matched point cloud pairs through an SVD algorithm according to a plurality of dimensions to obtain a rotation matrix and a translation vector;
constructing a rigid transformation relation between a source point cloud and a target point cloud according to the rotation matrix and the translation vector;
and rotating and/or translating each point of the source point cloud according to the rigid transformation relation to obtain a registration result from the source point cloud to the target point cloud.
In addition, in order to achieve the above object, the present invention also proposes a registration device based on probability distribution distance feature description, the registration device based on probability distribution distance feature description comprising:
the generation module is used for generating a source point cloud mixing probability distribution map according to the source point cloud and generating a target point cloud mixing probability distribution map according to the target point cloud;
The calculation module is used for respectively carrying out nonlinear calculation on the source point cloud mixing probability distribution diagram and the target point cloud mixing probability distribution diagram, and taking calculation results as source point cloud characteristics and target point cloud characteristics;
the generation module is further used for generating a point cloud corresponding relation matrix according to the source point cloud characteristics and the target point cloud characteristics;
and the determining module is used for determining a registration result from the source point cloud to the target point cloud according to the point cloud corresponding relation matrix.
In addition, to achieve the above object, the present invention also proposes a registration apparatus based on probability distribution distance feature description, including: a memory, a processor, and a probability distribution distance feature description based registration program stored on the memory and executable on the processor, the probability distribution distance feature description based registration program configured to implement the probability distribution distance feature description based registration method as described above.
In addition, in order to achieve the above object, the present invention also proposes a storage medium having stored thereon a registration program based on a probability distribution distance feature description, which when executed by a processor, implements a registration method based on a probability distribution distance feature description as described above.
According to the registration method based on the probability distribution distance feature description, a source point cloud mixed probability distribution map is generated according to the source point cloud, and a target point cloud mixed probability distribution map is generated according to the target point cloud; respectively carrying out nonlinear calculation on the source point cloud mixing probability distribution map and the target point cloud mixing probability distribution map, and taking calculation results as source point cloud characteristics and target point cloud characteristics; generating a point cloud corresponding relation matrix according to the source point cloud characteristics and the target point cloud characteristics; determining a registration result from the source point cloud to the target point cloud according to the point cloud corresponding relation matrix; by the method, the point cloud corresponding relation matrix is generated according to the source point cloud characteristics and the target point cloud characteristics, and then the registration result from the source point cloud to the target point cloud is determined according to the point cloud corresponding relation matrix, so that the accuracy, the stability and the generalization capability of the registration point cloud can be effectively improved.
Drawings
FIG. 1 is a schematic structural diagram of a registration device described based on probability distribution distance features of a hardware operating environment according to an embodiment of the present invention;
FIG. 2 is a flow chart of a first embodiment of a registration method described based on probability distribution distance features of the present invention;
FIG. 3 is a flow chart of a second embodiment of the registration method described based on probability distribution distance features of the present invention;
fig. 4 is a functional block diagram of a first embodiment of the registration device described based on probability distribution distance features of the present invention.
The achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Referring to fig. 1, fig. 1 is a schematic structural diagram of a registration device described based on probability distribution distance features of a hardware running environment according to an embodiment of the present invention.
As shown in fig. 1, the registration apparatus described based on probability distribution distance features may include: a processor 1001, such as a central processing unit (Central Processing Unit, CPU), a communication bus 1002, a user interface 1003, a network interface 1004, a memory 1005. Wherein the communication bus 1002 is used to enable connected communication between these components. The user interface 1003 may include a Display, an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may further include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a Wireless interface (e.g., a Wireless-Fidelity (Wi-Fi) interface). The Memory 1005 may be a high-speed random access Memory (Random Access Memory, RAM) Memory or a stable nonvolatile Memory (NVM), such as a disk Memory. The memory 1005 may also optionally be a storage device separate from the processor 1001 described above.
It will be appreciated by those skilled in the art that the structure shown in fig. 1 does not constitute a limitation of the registration apparatus described based on probability distribution distance features, and may include more or fewer components than illustrated, or certain components in combination, or a different arrangement of components.
As shown in fig. 1, an operating system, a network communication module, a user interface module, and a registration program described based on probability distribution distance features may be included in the memory 1005 as one storage medium.
In the registration apparatus described based on probability distribution distance features shown in fig. 1, the network interface 1004 is mainly used for data communication with a network integration platform workstation; the user interface 1003 is mainly used for data interaction with a user; the processor 1001 and the memory 1005 in the registration device based on the probability distribution distance feature description may be disposed in the registration device based on the probability distribution distance feature description, where the registration device based on the probability distribution distance feature description calls a registration program based on the probability distribution distance feature description stored in the memory 1005 through the processor 1001, and executes the registration method based on the probability distribution distance feature description provided by the embodiment of the invention.
Based on the hardware structure, the registration method embodiment based on the probability distribution distance feature description is provided.
Referring to fig. 2, fig. 2 is a schematic flow chart of a first embodiment of the registration method described based on probability distribution distance features of the present invention.
In a first embodiment, the registration method described based on probability distribution distance features comprises the steps of:
step S10, a source point cloud mixing probability distribution map is generated according to the source point cloud, and a target point cloud mixing probability distribution map is generated according to the target point cloud.
It should be noted that, the execution body of the embodiment is a registration device described based on probability distribution distance features, and may be other devices that can implement the same or similar functions, for example, a point cloud registration controller, which is not limited in this embodiment, and in this embodiment, a point cloud registration controller is taken as an example for description.
It should be understood that the source point cloud mixed probability distribution map refers to a map data structure, each vertex of the map is a source point cloud target point neighborhood probability distribution and a target point belonging segment probability distribution, and likewise, the target point cloud mixed probability distribution map refers to a map data structure, each vertex of the map is a target point cloud target point neighborhood probability distribution and a target point belonging segment probability distribution.
And S20, respectively carrying out nonlinear calculation on the source point cloud mixing probability distribution map and the target point cloud mixing probability distribution map, and taking the calculation results as a source point cloud characteristic and a target point cloud characteristic.
Further, step S20 includes: calculating a first probability distribution distance between a target vertex in the source point cloud mixed probability distribution diagram and each vertex in the target point cloud mixed probability distribution diagram; generating a source vertex vector according to the first probability distribution distance and the number of source vertices; calculating a second probability distribution distance between a target vertex and a source point cloud segmentation part in the source point cloud mixed probability distribution map; generating a first vertex matrix according to the second probability distribution distance and the number of source vertices; calculating a third probability distribution distance between a target vertex and a target point cloud segmentation part in the source point cloud mixed probability distribution map; generating a second vertex matrix according to the third probability distribution distance and the number of source vertices; expanding the target vertex vector to the first vertex matrix and the second vertex matrix respectively; generating a source point cloud feature matrix according to the expanded first vertex matrix, the expanded second vertex matrix and the target vertex vector; calculating a fourth probability distribution distance between a target vertex in the target point cloud mixing probability distribution diagram and each vertex in the source point cloud mixing probability distribution diagram; generating a target vertex vector according to the fourth probability distribution distance and the number of the target vertices; calculating a fifth probability distribution distance between a target vertex and a target point cloud segmentation part in the target point cloud mixed probability distribution map; generating a third vertex matrix according to the fifth probability distribution distance and the number of target vertices; calculating a sixth probability distribution distance between a target vertex and a source point cloud segmentation part in the target point cloud mixed probability distribution map; generating a fourth vertex matrix according to the sixth probability distribution distance and the number of target vertices; expanding the target vertex vector to the third vertex matrix and the fourth vertex matrix respectively; generating a target point cloud feature matrix according to the expanded third vertex matrix and fourth vertex matrix and the target vertex vector; and obtaining source point cloud characteristics and target point cloud characteristics according to the source point cloud characteristic matrix and the target point cloud characteristic matrix.
It should be understood that the number of the source point clouds and the target point clouds may be N, after the source point cloud mixed probability distribution map and the target point cloud mixed probability distribution map are obtained, calculating probability distribution distances between a target vertex in the source point cloud mixed probability distribution map and each vertex in the target point cloud mixed probability distribution map, that is, a first probability distribution distance, and performing SUM operation on a vector of the first probability distribution distance to obtain a one-dimensional feature of the target vertex, and then forming an N x 1 vector from N vertices; calculating second probability distribution distances between the target vertexes and M source point cloud segmentation parts (in order of from small to large labels) in a source point cloud mixed probability distribution map, forming 1*M-dimensional vectors according to the second probability distribution distances and the number of the source vertexes, and forming a first vertex matrix with N x M dimensions by N vertexes; calculating 1*M-dimensional vectors for third probability distribution distances between target vertexes and M target point cloud segmentation parts (in order of small to large labels) in the source point cloud mixed probability distribution diagram, and forming an N-dimension M-dimension second vertex matrix according to the 1*M-dimension vectors and the N vertexes; and copying, splicing and expanding the N1 target vertex vectors to the N first vertex matrix and the N second vertex matrix, splicing the expanded first vertex matrix, the expanded second vertex matrix and the target source vertex vectors to form an N3M-dimensional source point cloud feature matrix, generating a target point cloud feature matrix in the same mode, expanding the generated target vertex vectors to the third vertex matrix and the fourth vertex matrix after the third vertex matrix and the fourth vertex matrix are obtained, and generating the N3M-dimensional target point cloud feature matrix by the expanded third vertex matrix, the expanded fourth vertex matrix and the target vertex vectors.
It should be noted that, the probability distribution Distance also becomes a difference measure between probability distributions, and the EMD (Earth Mover's Distance) Distance is used to measure, where the EMD Distance quantifies the similarity between two probability distributions by minimizing the conversion cost between the two probability distributions, and the gaussian distribution of the three-dimensional point cloud data is as follows:
wherein S represents a source point cloud, T represents a target point cloud, mu 1 Represents the mean, mu, of each dimension variable of the source point cloud 2 Mean value of each dimension variable of target point cloud is represented, and sigma 1 Covariance matrix of each dimension variable of source point cloud, sigma 2 The covariance matrix of each dimension variable of the target point cloud is represented, x represents the points in the point cloud set, and K represents the dimension here as 3.
The EMD distances can then be described as:
EMD KL (S(x|μ 1 ，∑ 1 )||T(x|μ 2 ，∑ 2 ))＝(μ 1 -μ 2 )+Tr(∑ 1 +∑ 2 -2(∑ 1 1/2 ∑ 2 ∑ 1 1/2 ) 1/2 )
wherein KL represents KL divergence, generally referred to as relative entropy.
Further, the obtaining the source point cloud feature and the target point cloud feature according to the source point cloud feature matrix and the target point cloud feature matrix includes: performing multi-layer perceptron calculation on the source point cloud characteristic matrix and the target point cloud characteristic matrix respectively to obtain a source point cloud characteristic description matrix and a target point cloud characteristic description matrix; preprocessing the source point cloud characteristic description matrix and the target point cloud characteristic description matrix respectively to obtain source point cloud characteristic description and target point cloud characteristic description; and obtaining the source point cloud characteristics and the target point cloud characteristics according to the source point cloud characteristics description and the target point cloud characteristics description.
It can be understood that after the source point cloud feature matrix and the target point cloud feature matrix are obtained, the source point cloud feature description matrix and the target point cloud feature description matrix are calculated by the multi-layer perceptron with the same layer number respectively, the dimensions of the source point cloud feature description matrix and the target point cloud feature description matrix are N x 1024, and then the source point cloud feature description matrix and the target point cloud feature description matrix are converged to obtain feature description with strong expression capability, namely, the source point cloud feature description and the target point cloud feature description.
It should be noted that, the feature description matrix aggregation can be implemented in combination with an attention mechanism, and the self-attention mechanism and the cross-attention mechanism perform intra-graph convolution and cross-graph convolution, so as to effectively aggregate the original feature information into a final feature description matrix.
And step S30, generating a point cloud corresponding relation matrix according to the source point cloud characteristics and the target point cloud characteristics.
It should be understood that the point cloud correspondence matrix refers to a relationship matrix between a source point cloud and a target point cloud, where the point cloud correspondence matrix is obtained by alternately performing row and column computation by using a sinkhorn algorithm.
Further, step S30 includes: performing dot product calculation on the source point cloud features and the target point cloud features to obtain a plurality of feature similarities; constructing a feature similarity matrix according to the feature similarities; and performing row-column calculation on the feature similarity matrix through a sinkhorn algorithm to obtain a point cloud corresponding relation matrix.
It can be understood that the feature similarity refers to the similarity between the source point cloud features and the target point cloud features, the number of the feature similarities is a plurality, after the dot product calculates a plurality of feature similarities, a feature similarity matrix is constructed, at this time, in order to improve the accuracy of obtaining the point cloud corresponding relation matrix, the graph matching problem is converted into the linear assignment problem to be processed, that is, the feature similarity matrix is alternately subjected to row and column calculation through a sinkhorn algorithm to obtain the point cloud corresponding relation matrix, in addition, the point cloud corresponding relation matrix is corrected according to the label and the mark of the segmentation module, and when the marks of the respective segmentation parts where any two points of the source point cloud and the target point cloud are located are different, the corresponding weight between the two points is reduced to zero.
And step S40, determining a registration result from the source point cloud to the target point cloud according to the point cloud corresponding relation matrix.
Further, step S40 includes: obtaining a matched point cloud pair according to the point cloud corresponding relation matrix; calculating the matched point cloud pairs through an SVD algorithm according to a plurality of dimensions to obtain a rotation matrix and a translation vector; constructing a rigid transformation relation between a source point cloud and a target point cloud according to the rotation matrix and the translation vector; and rotating and/or translating each point of the source point cloud according to the rigid transformation relation to obtain a registration result from the source point cloud to the target point cloud.
It should be understood that the matching point cloud pair refers to a point cloud pair matched between a source point cloud and a target point cloud, and the number of dimensions refers to dimensions of calculating the matching point cloud pair, wherein the dimensions include, but are not limited to, a rotation dimension and a translation dimension, that is, rotation and translation calculation are performed on the matching point cloud pair, then a rigid transformation relationship between the source point cloud and the target point cloud is constructed according to a rotation matrix and a translation vector, and then each point in the source point cloud is rotated and/or translated according to the rigid transformation relationship, so that a registration result from the source point cloud to the target point cloud is obtained.
In the embodiment, a source point cloud mixing probability distribution map is generated according to a source point cloud, and a target point cloud mixing probability distribution map is generated according to a target point cloud; respectively carrying out nonlinear calculation on the source point cloud mixing probability distribution map and the target point cloud mixing probability distribution map, and taking calculation results as source point cloud characteristics and target point cloud characteristics; generating a point cloud corresponding relation matrix according to the source point cloud characteristics and the target point cloud characteristics; determining a registration result from the source point cloud to the target point cloud according to the point cloud corresponding relation matrix; by the method, the point cloud corresponding relation matrix is generated according to the source point cloud characteristics and the target point cloud characteristics, and then the registration result from the source point cloud to the target point cloud is determined according to the point cloud corresponding relation matrix, so that the accuracy, the stability and the generalization capability of the registration point cloud can be effectively improved.
In an embodiment, as shown in fig. 3, a second embodiment of the registration method described based on probability distribution distance features according to the present invention is proposed based on the first embodiment, and the step S10 includes:
step S101, an initial source point cloud and an initial point cloud are obtained, and calibration is carried out on the initial source point cloud and the initial point cloud.
It should be appreciated that after obtaining the initial source point cloud and the initial point cloud, since the depth sensing device is subject to environmental interference during measurement, the obtained collected data is often accompanied by noise, holes, outliers, and the like, and thus, preprocessing is required before the three-dimensional point cloud (the initial source point cloud and the initial point cloud) formally performs the calculation tasks, for example, the preprocessing includes, but is not limited to, calibration and quality enhancement, that is, calibration is performed on the initial source point cloud and the initial point cloud first, the calibration may be semi-automatic and automatic by implementing semi-automatic calibration by means of external object reference and checkerboard method, and the calibration may be camera calibration.
And step S102, respectively carrying out quality enhancement on the calibrated initial source point cloud and the calibrated initial point cloud to obtain a source point cloud and a target point cloud.
It can be understood that the source point cloud refers to a point cloud after the initial source point cloud is calibrated and the quality of the source point cloud is enhanced, and likewise, the target point cloud refers to a point cloud after the initial point cloud is calibrated and the quality of the source point cloud is enhanced, and the quality of the source point cloud is enhanced through sampling optimization, filtering noise reduction and hole filling.
Step S103, generating a source point cloud model according to the source point cloud, and generating a target point cloud model according to the target point cloud.
And step S104, semantic segmentation is respectively carried out on the source point cloud model and the target point cloud model through a target proximity algorithm, and a source point cloud segmentation result and a target point cloud segmentation result are obtained.
It can be understood that after the source point cloud model and the target point cloud model are obtained, semantic segmentation is performed on the source point cloud model and the target point cloud model through a target proximity algorithm, so that grouping of the interior of point cloud data is realized, the point cloud features can be helped to be embedded into certain global characteristics, and in some point clouds with obvious feature distribution, after the semantic segmentation is completed, a source point cloud segmentation result and a target point cloud segmentation result are obtained.
Step S105, a source point cloud mixing probability distribution map is generated according to the source point cloud segmentation result, and a target point cloud mixing probability distribution map is generated according to the target point cloud segmentation result.
It should be understood that after the source point cloud segmentation result and the target point cloud segmentation result are obtained, the source point cloud segmentation result and the target point cloud segmentation result are respectively marked, the marked matching is performed, the feature of each segmentation part is extracted for similarity solving in the specific matching process, when the solved similarity is larger than a similarity threshold, the segmentation part with the highest similarity in the source point cloud and the target point cloud is considered to be matched with each other and have the same mark, then after the matching is completed, a source point cloud mixing probability distribution map is generated according to the source point cloud segmentation result, and a target point cloud mixing probability distribution map is generated according to the target point cloud segmentation result.
Further, step S105 includes: calculating the source point cloud segmentation result to obtain Gaussian distribution of a source point cloud segmentation part and neighborhood Gaussian distribution of each point of the source point cloud; splicing the neighborhood Gaussian distribution of each point of the source point cloud and the Gaussian distribution of the source point cloud segmentation part to which each point of the source point cloud belongs to obtain the source point cloud mixed Gaussian distribution; generating a source point cloud mixing probability distribution diagram according to the source point cloud mixing Gaussian distribution and a preset diagram structure; calculating the target point cloud segmentation result to obtain Gaussian distribution of a target point cloud segmentation part and neighborhood Gaussian distribution of each point of the target point cloud; splicing the neighborhood Gaussian distribution of each point of the target point cloud and the Gaussian distribution of the target point cloud segmentation part to which each point of the target point cloud belongs to obtain the target point cloud Gaussian mixture distribution; and generating a target point cloud mixing probability distribution map according to the target point cloud mixing Gaussian distribution and a preset map structure.
It can be understood that after the source point cloud segmentation result is obtained, the source point cloud is segmented into m parts according to the source point cloud segmentation result, and the corresponding gaussian distribution is calculated, that is, the source point cloud segmentation part gaussian distribution guide (s_m i ) Then searching adjacent points by setting the value of the neighborhood radius r as each point in the source point cloud and calculating the neighborhood Gaussian distribution as the distribution of the point, namely, neighborhood Gaussian distribution guide (s_p) i ) Also, the target point cloud is divided into m parts according to the target point cloud division result and the corresponding gaussian distribution is calculated, that is, the target point cloud division part gaussian distribution guide (t_m) i ) Then searching adjacent points by setting the value of the neighborhood radius r as each point in the target point cloud and calculating the neighborhood Gaussian distribution as the distribution of the point, namely, neighborhood Gaussian distribution guide (t_p) of each point in the target point cloud i ) Respectively calculating the Gaussian mixture distribution of the source point cloud<guass(s_p i )，guass(s_m i )>Target point cloud mixed Gaussian distribution<guass(t_p i )，guass(t_m i )>And then generating a source point cloud mixing probability distribution map according to the source point cloud mixing Gaussian distribution and a preset graph structure, and generating a target point cloud mixing probability distribution map according to the target point cloud mixing Gaussian distribution and the preset graph structure, namely respectively modeling the source point cloud mixing Gaussian distribution and the target point cloud mixing Gaussian distribution by using the preset graph structure, taking each mixing Gaussian distribution as a vertex in the graph, and establishing one edge for any two vertices in a full-connection mode to obtain the source point cloud mixing probability distribution map and the target point cloud mixing probability distribution map.
In the embodiment, an initial source point cloud and an initial point cloud are obtained, and the initial source point cloud and the initial point cloud are calibrated; respectively carrying out quality enhancement on the calibrated initial source point cloud and initial point cloud to obtain a source point cloud and a target point cloud; generating a source point cloud model according to the source point cloud, and generating a target point cloud model according to the target point cloud; semantic segmentation is respectively carried out on the source point cloud model and the target point cloud model through a target proximity algorithm, so that a source point cloud segmentation result and a target point cloud segmentation result are obtained; generating a source point cloud mixing probability distribution map according to the source point cloud segmentation result, and generating a target point cloud mixing probability distribution map according to the target point cloud segmentation result; by the method, the source point cloud model and the target point cloud model are generated after the initial source point cloud and the initial point cloud are calibrated and the quality of the initial point cloud is enhanced, semantic segmentation is respectively carried out on the source point cloud model and the target point cloud model through a target proximity algorithm, and a source point cloud mixing probability distribution map and a target point cloud mixing probability distribution map are generated according to a source point cloud segmentation result and a target point cloud segmentation result, so that the accuracy of generating the point cloud mixing probability distribution map can be effectively improved.
In addition, the embodiment of the invention also provides a storage medium, wherein the storage medium is stored with a registration program based on the probability distribution distance feature description, and the registration program based on the probability distribution distance feature description realizes the steps of the registration method based on the probability distribution distance feature description when being executed by a processor.
Because the storage medium adopts all the technical schemes of all the embodiments, the storage medium has at least all the beneficial effects brought by the technical schemes of the embodiments, and the description is omitted here.
In addition, referring to fig. 4, an embodiment of the present invention further provides a registration device based on probability distribution distance feature description, where the registration device based on probability distribution distance feature description includes:
the generating module 10 is configured to generate a source point cloud mixing probability distribution map according to the source point cloud, and generate a target point cloud mixing probability distribution map according to the target point cloud.
The calculating module 20 is configured to perform nonlinear calculation on the source point cloud blending probability distribution map and the target point cloud blending probability distribution map, and take the calculation result as a source point cloud feature and a target point cloud feature.
The generating module 10 is further configured to generate a point cloud correspondence matrix according to the source point cloud feature and the target point cloud feature.
The determining module 30 is configured to determine a registration result from the source point cloud to the target point cloud according to the point cloud correspondence matrix.
In the embodiment, a source point cloud mixing probability distribution map is generated according to a source point cloud, and a target point cloud mixing probability distribution map is generated according to a target point cloud; respectively carrying out nonlinear calculation on the source point cloud mixing probability distribution map and the target point cloud mixing probability distribution map, and taking calculation results as source point cloud characteristics and target point cloud characteristics; generating a point cloud corresponding relation matrix according to the source point cloud characteristics and the target point cloud characteristics; determining a registration result from the source point cloud to the target point cloud according to the point cloud corresponding relation matrix; by the method, the point cloud corresponding relation matrix is generated according to the source point cloud characteristics and the target point cloud characteristics, and then the registration result from the source point cloud to the target point cloud is determined according to the point cloud corresponding relation matrix, so that the accuracy, the stability and the generalization capability of the registration point cloud can be effectively improved.
It should be noted that the above-described working procedure is merely illustrative, and does not limit the scope of the present invention, and in practical application, a person skilled in the art may select part or all of them according to actual needs to achieve the purpose of the embodiment, which is not limited herein.
In addition, technical details that are not described in detail in the present embodiment may refer to the registration method described based on probability distribution distance features provided in any embodiment of the present invention, which is not described herein.
Furthermore, it should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The foregoing embodiment numbers of the present invention are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.
From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. Read Only Memory)/RAM, magnetic disk, optical disk) and including several instructions for causing a terminal device (which may be a mobile phone, a computer, an integrated platform workstation, or a network device, etc.) to perform the method according to the embodiments of the present invention.
The foregoing description is only of the preferred embodiments of the present invention, and is not intended to limit the scope of the invention, but rather is intended to cover any equivalents of the structures or equivalent processes disclosed herein or in the alternative, which may be employed directly or indirectly in other related arts.
Claims (10)
1. A registration method based on probability distribution distance feature description, characterized in that the registration method based on probability distribution distance feature description comprises the following steps:
generating a source point cloud mixing probability distribution map according to the source point cloud, and generating a target point cloud mixing probability distribution map according to the target point cloud;
respectively carrying out nonlinear calculation on the source point cloud mixing probability distribution map and the target point cloud mixing probability distribution map, and respectively taking calculation results as a source point cloud characteristic and a target point cloud characteristic;
generating a point cloud corresponding relation matrix according to the source point cloud characteristics and the target point cloud characteristics;
and determining a registration result from the source point cloud to the target point cloud according to the point cloud corresponding relation matrix.
2. The registration method based on probability distribution distance feature description according to claim 1, wherein the generating a source point cloud mixture probability distribution map from source point clouds and generating a target point cloud mixture probability distribution map from target point clouds includes:
Acquiring an initial source point cloud and an initial point cloud, and calibrating the initial source point cloud and the initial point cloud;
respectively carrying out quality enhancement on the calibrated initial source point cloud and initial point cloud to obtain a source point cloud and a target point cloud;
generating a source point cloud model according to the source point cloud, and generating a target point cloud model according to the target point cloud;
semantic segmentation is respectively carried out on the source point cloud model and the target point cloud model through a target proximity algorithm, so that a source point cloud segmentation result and a target point cloud segmentation result are obtained;
and generating a source point cloud mixing probability distribution map according to the source point cloud segmentation result, and generating a target point cloud mixing probability distribution map according to the target point cloud segmentation result.
3. The registration method based on probability distribution distance feature description according to claim 2, wherein the generating a source point cloud mixture probability distribution map according to the source point cloud segmentation result and generating a target point cloud mixture probability distribution map according to the target point cloud segmentation result includes:
calculating the source point cloud segmentation result to obtain Gaussian distribution of a source point cloud segmentation part and neighborhood Gaussian distribution of each point of the source point cloud;
splicing the neighborhood Gaussian distribution of each point of the source point cloud and the Gaussian distribution of the source point cloud segmentation part to which each point of the source point cloud belongs to obtain the source point cloud mixed Gaussian distribution;
Generating a source point cloud mixing probability distribution diagram according to the source point cloud mixing Gaussian distribution and a preset diagram structure;
calculating the target point cloud segmentation result to obtain Gaussian distribution of a target point cloud segmentation part and neighborhood Gaussian distribution of each point of the target point cloud;
splicing the neighborhood Gaussian distribution of each point of the target point cloud and the Gaussian distribution of the target point cloud segmentation part to which each point of the target point cloud belongs to obtain the target point cloud Gaussian mixture distribution;
and generating a target point cloud mixing probability distribution map according to the target point cloud mixing Gaussian distribution and a preset map structure.
4. The registration method based on probability distribution distance feature description according to claim 1, wherein the performing nonlinear calculation on the source point cloud mixture probability distribution map and the target point cloud mixture probability distribution map respectively, and using the calculation result as a source point cloud feature and a target point cloud feature, includes:
calculating a first probability distribution distance between a target vertex in the source point cloud mixed probability distribution diagram and each vertex in the target point cloud mixed probability distribution diagram;
generating a source vertex vector according to the first probability distribution distance and the number of source vertices;
calculating a second probability distribution distance between a target vertex and a source point cloud segmentation part in the source point cloud mixed probability distribution map;
Generating a first vertex matrix according to the second probability distribution distance and the number of source vertices;
calculating a third probability distribution distance between a target vertex and a target point cloud segmentation part in the source point cloud mixed probability distribution map;
generating a second vertex matrix according to the third probability distribution distance and the number of source vertices;
expanding the target vertex vector to the first vertex matrix and the second vertex matrix respectively;
generating a source point cloud feature matrix according to the expanded first vertex matrix, the expanded second vertex matrix and the target vertex vector;
calculating a fourth probability distribution distance between a target vertex in the target point cloud mixing probability distribution diagram and each vertex in the source point cloud mixing probability distribution diagram;
generating a target vertex vector according to the fourth probability distribution distance and the number of the target vertices;
calculating a fifth probability distribution distance between a target vertex and a target point cloud segmentation part in the target point cloud mixed probability distribution map;
generating a third vertex matrix according to the fifth probability distribution distance and the number of target vertices;
calculating a sixth probability distribution distance between a target vertex and a source point cloud segmentation part in the target point cloud mixed probability distribution map;
generating a fourth vertex matrix according to the sixth probability distribution distance and the number of target vertices;
Expanding the target vertex vector to the third vertex matrix and the fourth vertex matrix respectively;
generating a target point cloud feature matrix according to the expanded third vertex matrix and fourth vertex matrix and the target vertex vector;
and obtaining source point cloud characteristics and target point cloud characteristics according to the source point cloud characteristic matrix and the target point cloud characteristic matrix.
5. The registration method based on probability distribution distance feature description according to claim 4, wherein the obtaining the source point cloud feature and the target point cloud feature according to the source point cloud feature matrix and the target point cloud feature matrix includes:
performing multi-layer perceptron calculation on the source point cloud characteristic matrix and the target point cloud characteristic matrix respectively to obtain a source point cloud characteristic description matrix and a target point cloud characteristic description matrix;
preprocessing the source point cloud characteristic description matrix and the target point cloud characteristic description matrix respectively to obtain source point cloud characteristic description and target point cloud characteristic description;
and obtaining the source point cloud characteristics and the target point cloud characteristics according to the source point cloud characteristics description and the target point cloud characteristics description.
6. The registration method based on probability distribution distance feature description according to claim 1, wherein the generating a point cloud correspondence matrix according to the source point cloud feature and the target point cloud feature includes:
Performing dot product calculation on the source point cloud features and the target point cloud features to obtain a plurality of feature similarities;
constructing a feature similarity matrix according to the feature similarities;
and performing row-column calculation on the feature similarity matrix through a sinkhorn algorithm to obtain a point cloud corresponding relation matrix.
7. The registration method based on probability distribution distance feature description according to claim 6, wherein the determining a registration result of the source point cloud to the target point cloud according to the point cloud correspondence matrix includes:
obtaining a matched point cloud pair according to the point cloud corresponding relation matrix;
calculating the matched point cloud pairs through an SVD algorithm according to a plurality of dimensions to obtain a rotation matrix and a translation vector;
constructing a rigid transformation relation between a source point cloud and a target point cloud according to the rotation matrix and the translation vector;
and rotating and/or translating each point of the source point cloud according to the rigid transformation relation to obtain a registration result from the source point cloud to the target point cloud.
8. A registration device based on probability distribution distance feature description, characterized in that the registration device based on probability distribution distance feature description comprises:
The generation module is used for generating a source point cloud mixing probability distribution map according to the source point cloud and generating a target point cloud mixing probability distribution map according to the target point cloud;
the calculation module is used for respectively carrying out nonlinear calculation on the source point cloud mixing probability distribution diagram and the target point cloud mixing probability distribution diagram, and taking calculation results as source point cloud characteristics and target point cloud characteristics;
the generation module is further used for generating a point cloud corresponding relation matrix according to the source point cloud characteristics and the target point cloud characteristics;
and the determining module is used for determining a registration result from the source point cloud to the target point cloud according to the point cloud corresponding relation matrix.
9. A registration device based on probability distribution distance feature description, characterized in that the registration device based on probability distribution distance feature description comprises: a memory, a processor and a probability distribution distance feature description based registration program stored on the memory and executable on the processor, the probability distribution distance feature description based registration program configured to implement the probability distribution distance feature description based registration method as claimed in any one of claims 1 to 7.
10. A storage medium, wherein a registration procedure based on a probability distribution distance feature description is stored on the storage medium, which, when executed by a processor, implements the registration method based on a probability distribution distance feature description as claimed in any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310383322.8A CN116468761A (en) | 2023-04-06 | 2023-04-06 | Registration method, equipment and storage medium based on probability distribution distance feature description |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310383322.8A CN116468761A (en) | 2023-04-06 | 2023-04-06 | Registration method, equipment and storage medium based on probability distribution distance feature description |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116468761A true CN116468761A (en) | 2023-07-21 |
Family
ID=87180112
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310383322.8A Pending CN116468761A (en) | 2023-04-06 | 2023-04-06 | Registration method, equipment and storage medium based on probability distribution distance feature description |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116468761A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117689698A (en) * | 2024-02-04 | 2024-03-12 | 安徽蔚来智驾科技有限公司 | Point cloud registration method, intelligent device and storage medium |
-
2023
- 2023-04-06 CN CN202310383322.8A patent/CN116468761A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117689698A (en) * | 2024-02-04 | 2024-03-12 | 安徽蔚来智驾科技有限公司 | Point cloud registration method, intelligent device and storage medium |
| CN117689698B (en) * | 2024-02-04 | 2024-04-19 | 安徽蔚来智驾科技有限公司 | Point cloud registration method, intelligent device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111902825A (en) | Polygonal object labeling system and method for training object labeling system | |
| US11727584B2 (en) | Shape supplementation device, shape supplementation learning device, method, and program | |
| CN112328715B (en) | Visual positioning method, training method of related model, related device and equipment | |
| CN114155244B (en) | Defect detection method, device, equipment and storage medium | |
| Cicconet et al. | A convolutional approach to reflection symmetry | |
| KR101618996B1 (en) | Sampling method and image processing apparatus for estimating homography | |
| WO2023030163A1 (en) | Method, apparatus and device for converting texture map of three-dimensional model, and medium | |
| CN113177592B (en) | Image segmentation method and device, computer equipment and storage medium | |
| CN111707275B (en) | Positioning method, positioning device, electronic equipment and computer readable storage medium | |
| CN115984662B (en) | Multi-mode data pre-training and identifying method, device, equipment and medium | |
| CN110704652A (en) | Vehicle image fine-grained retrieval method and device based on multiple attention mechanism | |
| CN116468761A (en) | Registration method, equipment and storage medium based on probability distribution distance feature description | |
| CN111179270A (en) | Image co-segmentation method and device based on attention mechanism | |
| CN115457492A (en) | Target detection method and device, computer equipment and storage medium | |
| CN109190662A (en) | A kind of three-dimensional vehicle detection method, system, terminal and storage medium returned based on key point | |
| Yamni et al. | Accurate 2D and 3D images classification using translation and scale invariants of Meixner moments | |
| CN116563096B (en) | Method and device for determining deformation field for image registration and electronic equipment | |
| CN113435479A (en) | Feature point matching method and system based on regional feature expression constraint | |
| CN111914809A (en) | Target object positioning method, image processing method, device and computer equipment | |
| CN117058008A (en) | Remote sensing image geometry and radiation integrated correction method, device, equipment and medium | |
| CN117274072A (en) | Point cloud denoising method and device based on two-dimensional multi-modal range image | |
| CN110717405A (en) | Face feature point positioning method, device, medium and electronic equipment | |
| CN113139617B (en) | Power transmission line autonomous positioning method and device and terminal equipment | |
| CN112561817B (en) | Remote sensing image cloud removing method, device, equipment and storage medium based on AM-GAN | |
| CN114998630A (en) | Ground-to-air image registration method from coarse to fine |
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 |