CN116229089A - Appearance geometric analysis method and system - Google Patents
Appearance geometric analysis method and system Download PDFInfo
- Publication number
- CN116229089A CN116229089A CN202310517795.2A CN202310517795A CN116229089A CN 116229089 A CN116229089 A CN 116229089A CN 202310517795 A CN202310517795 A CN 202310517795A CN 116229089 A CN116229089 A CN 116229089A
- Authority
- CN
- China
- Prior art keywords
- mapping
- subspace
- probability
- manifold
- space
- 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
- 238000004458 analytical method Methods 0.000 title claims abstract description 58
- 238000013507 mapping Methods 0.000 claims abstract description 120
- 238000012545 processing Methods 0.000 claims abstract description 51
- 238000000034 method Methods 0.000 claims abstract description 40
- 238000004364 calculation method Methods 0.000 claims abstract description 30
- 230000006870 function Effects 0.000 claims description 57
- 238000001514 detection method Methods 0.000 claims description 19
- 238000011156 evaluation Methods 0.000 claims description 15
- 230000007547 defect Effects 0.000 claims description 13
- 239000002131 composite material Substances 0.000 claims description 12
- 238000010276 construction Methods 0.000 claims description 10
- 238000005259 measurement Methods 0.000 claims description 10
- 239000013598 vector Substances 0.000 claims description 10
- 238000009499 grossing Methods 0.000 claims description 8
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000012512 characterization method Methods 0.000 claims description 5
- 230000007787 long-term memory Effects 0.000 claims description 4
- 230000035772 mutation Effects 0.000 claims description 4
- 238000006467 substitution reaction Methods 0.000 claims description 4
- 230000000007 visual effect Effects 0.000 claims 1
- 238000004422 calculation algorithm Methods 0.000 abstract description 6
- 238000010586 diagram Methods 0.000 description 14
- 230000008569 process Effects 0.000 description 9
- 238000004590 computer program Methods 0.000 description 7
- 230000014509 gene expression Effects 0.000 description 6
- 230000008901 benefit Effects 0.000 description 4
- 238000003860 storage Methods 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 230000001419 dependent effect Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 230000010354 integration Effects 0.000 description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 210000001161 mammalian embryo Anatomy 0.000 description 2
- 230000015654 memory Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000000638 solvent extraction Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001364 causal effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000002787 reinforcement Effects 0.000 description 1
- 238000012502 risk assessment Methods 0.000 description 1
- 238000007711 solidification Methods 0.000 description 1
- 230000008023 solidification Effects 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
Images
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
- G06V10/40—Extraction of image or video features
- G06V10/42—Global feature extraction by analysis of the whole pattern, e.g. using frequency domain transformations or autocorrelation
- G06V10/422—Global feature extraction by analysis of the whole pattern, e.g. using frequency domain transformations or autocorrelation for representing the structure of the pattern or shape of an object therefor
-
- 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
- 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
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
-
- 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)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an appearance geometric analysis method and system, comprising the following steps: according to the determined target problem to be analyzed, obtaining a characteristic factor of a corresponding image, and dividing a topological space corresponding to the target problem into a plurality of subspaces according to the characteristic factor; each subspace containing one or more feature factors; the characteristic factors of each subspace are subjected to mapping treatment of corresponding manifold to obtain corresponding mapping results, and the mapping results are fused to obtain a multidimensional real space; obtaining a probability analysis table for representing the relation between geometric features according to a probability map constructed by the multidimensional real space and a preset logic processing sequence; obtaining a corresponding probability result according to the probability analysis table, and processing the corresponding target problem according to the probability result; the method can overcome the difference caused by the uncertainty algorithm and the local calculation of various micro features, and has stronger applicability.
Description
Technical Field
The invention relates to the technical field of machine vision, in particular to an appearance geometric analysis method and system.
Background
In the industrial detection direction, the coarse granularity is gradually changed into the fine granularity, and the correlation theory is established on the traditional probability statistics and integral calculation, so that the characteristic analysis can be effectively performed, but the micro characteristics or detection can be difficult to different degrees. First, the prior art often relies on tags for fine-grained detection or analysis, but this is not beneficial for determining that a situation is highly likely but not yet present, and is not sensitive to potential risks; in addition, different micro features also have differences, if the same feature extraction method is adopted for a plurality of features with larger differences, the detection and analysis precision of fine granularity can be reduced, so that the performance of the whole analysis method is reduced; meanwhile, the prior art is low in generalization capability based on scene dispersion in machine vision, so that applicability is caused, if different scene types are required to be applied, or various uncertain algorithms for potential risk analysis are analyzed and different performance characteristics are synthesized, the problems are solved by constantly repairing leaks or changing calculation modes, a large number of dynamic calculation processes are caused, and a calculated framework always needs to be adjusted on the application scene, so that quick application is not facilitated.
Disclosure of Invention
The invention aims at overcoming the defects of the prior art, and provides an appearance geometric analysis method and system, which can overcome the difference caused by an uncertain algorithm and local calculation of various micro features and have stronger applicability.
In a first aspect, the present invention provides a geometric analysis method for appearance, comprising:
acquiring characteristic factors of corresponding images according to the determined target problems to be analyzed, and dividing a topological space corresponding to the target problems into a plurality of subspaces according to the characteristic factors; wherein the objective problem includes: measurement based on machine vision, defect detection or template search; each subspace containing one or more feature factors;
the characteristic factors of each subspace are subjected to mapping treatment of corresponding manifold to obtain corresponding mapping results, and the mapping results are fused to obtain a multidimensional real space;
obtaining a probability analysis table for representing the relation between geometric features according to a probability map constructed by the multidimensional real space and a preset logic processing sequence;
and obtaining a corresponding probability result according to the probability analysis table, and processing the corresponding target problem according to the probability result.
The invention divides the topological space of the target problem to be analyzed into a plurality of subspaces, uses corresponding manifold for different subspaces, can solve the target problem of different dimensions, further can synthesize different expression features, maps and fuses the feature factors of different dimensions through a mapping function, and can describe the feature of the relevance of each subspace from the local concept of the subspace instead of describing the feature of the local subspace by applying a single mathematical formula; moreover, by calculating the probability of representing the relation between geometric features, the relation is possibly closely related to the requisite event, so that the accuracy of appearance geometric analysis and the completeness of objective facts can be ensured, and therefore, the accuracy of measurement, defect detection or template search based on machine vision can be improved, and the method has higher reliability and higher applicability.
Further, the feature factors of each subspace are subjected to mapping processing of corresponding manifold to obtain corresponding mapping results, and the mapping results are fused to obtain a multidimensional real space, which comprises the following steps:
sequentially carrying out composite mapping on characteristic factors of the corresponding subspaces and alignment matrixes corresponding to the characteristic factors according to manifold shapes of the corresponding subspaces to respectively obtain corresponding mapping results, and fusing the obtained mapping results to obtain a multidimensional real space; the composite mapping is obtained according to a mapping function of the smooth mapping after feedback calculation.
According to the invention, the correlation characteristic description in the region is set up by adopting the differential angle of the manifold corresponding to the subspace, the corresponding manifold is used for different subspaces, the target problem of different dimensions can be solved, further, different expression characteristics can be synthesized, and the characteristic factors of different dimensions are mapped and fused through the mapping function, so that the characteristic of the correlation of each subspace can be described from the local concept of the subspace, and the characteristic of the local subspace is described by adopting a single mathematical formula.
Further, after dividing the topological space corresponding to the target problem into a plurality of subspaces according to the feature factor, the method further comprises: manifold and smooth mapping are respectively established for each subspace, specifically:
sequentially establishing manifolds with space structures, which are formed by corresponding neighborhoods and open sets of the neighborhoods in real space, for each subspace, and sequentially constructing smooth functions, the smoothness of which is independent of the shapes of the corresponding subspaces, according to each manifold;
and according to the smoothing function, sequentially constructing a smoothing mapping from each subspace to a real space, and respectively obtaining data of the subspace under a coordinate system corresponding to the real space.
According to the invention, a local coordinate system is introduced for each subspace, so that the description of the geometrical characteristics and the correlation and corresponding calculation among the geometrical characteristics are facilitated, and the correlation detection and analysis are facilitated according to specific target problems.
Further, the method sequentially establishes a manifold with a space structure for each subspace, wherein the manifold is composed of a corresponding neighborhood and an open set of the neighborhood in a real space, and specifically comprises the following steps:
when the subspace is the heterogeneous space, selecting a lie algebra aspect group to obtain a corresponding manifold; wherein the feature factors acquired in the heterogeneous space are obtained by the steps of: substitution, crossover, mutation or high frequency grabbing;
when the subspace is the isomorphic space, the corresponding manifold is obtained according to the Mask matrix, the convolution network, the long-term memory network or the time sequence.
Further, the obtaining a probability analysis table for characterizing the relation between geometric features according to the probability map constructed by the multidimensional real space and the preset logic processing sequence comprises the following steps:
according to the relation between the multidimensional real number spaces and a preset logic processing sequence, taking the multidimensional real number spaces as the input of probability functions, and obtaining a probability analysis table for representing the relation between geometric features according to the obtained probability map; wherein the probability function comprises: bayesian networks or markov random fields.
The invention adopts the probability of calculating the relation between the characteristic geometric features to be possibly closely related with the requisite event, thereby ensuring the accuracy of the appearance geometric analysis and the completeness of objective facts, and having higher reliability and stronger applicability.
Further, the processing of the feature factors of each subspace through the mapping of the corresponding manifold further comprises: feedback is carried out for mapping of the corresponding manifold, specifically:
and carrying out feedback calculation on the mapping of the corresponding manifold according to the selected evaluation function so as to enable the constructed corresponding manifold to conform to the characteristics of the target problem data.
Further, after feeding back the mapping for the corresponding manifold, it further includes:
the probability map is logically adjusted according to the evaluation function.
Further, the acquiring the feature factor of the corresponding image includes:
if the characteristic factors are used for calculating the characterization points, lines or planes, the real space is used for representing;
if the feature factors are used to characterize the multidimensional calculations, a complex space or tensor space is used for the representation.
Further, the data types of the feature factors include: vector, scalar, structural features, and Blob data formats.
The method and the device for analyzing the data to be analyzed have the advantages that the data to be analyzed are abstracted, different data types can be processed, the data to be analyzed in vector, scalar, structural characteristics and Blob data formats can be processed, so that the method and the device for analyzing the data to be analyzed can be suitable for modeling and analyzing an image system and a signal system, and different target problems can be processed by adopting the method for analyzing the data to be analyzed, so that the method for analyzing the data to be analyzed is high in adaptability and high in generalization capability.
In a second aspect, the present invention provides an appearance geometry analysis system comprising:
the subspace dividing module is used for acquiring characteristic factors of corresponding images according to the determined target problems to be analyzed, and dividing a topological space corresponding to the target problems into a plurality of subspaces according to the characteristic factors; wherein the objective problem includes: measurement based on machine vision, defect detection or template search; each subspace containing one or more feature factors;
the manifold and mapping processing module is used for obtaining corresponding mapping results by mapping the characteristic factors of each subspace through corresponding manifold, and fusing the mapping results to obtain a multidimensional real space;
the probability map construction module is used for obtaining a probability analysis table for representing the relation between geometric features according to the probability map constructed by the multidimensional real space and the preset logic processing sequence;
and the processing module is used for obtaining a corresponding probability result according to the probability analysis table and processing the corresponding target problem according to the probability result.
Drawings
FIG. 1 is a flow chart of an appearance geometry analyzing method according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of mapping functions provided by an embodiment of the present invention;
FIG. 3 is a schematic diagram of mapping manifold elements into two-dimensional space provided by an embodiment of the present invention;
FIG. 4 is a schematic diagram of a construction probability map provided by an embodiment of the present invention;
fig. 5 is a geometric analysis system for appearance according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1, a flow chart of an appearance geometry analysis method provided by an embodiment of the present invention includes steps S11 to S14, specifically:
step S11, according to a determined target problem to be analyzed, obtaining a characteristic factor of a corresponding image, and dividing a topological space corresponding to the target problem into a plurality of subspaces according to the characteristic factor; wherein the objective problem includes: measurement based on machine vision, defect detection or template search; each subspace contains one or more feature factors.
It should be noted that, since the target problem to be analyzed includes measurement based on machine vision, defect detection or template search, according to a specific target problem, it is necessary to divide subspaces according to corresponding topological spaces in order to create adaptive manifolds for subspaces, including using Mask, homoembryo mapping or using X-dependent variable matrix according to a specific subspace, and therefore, it is necessary to implement specific steps from the framework level. After determining a specific target problem to be analyzed, selecting and constructing a data dimension of the target problem, including: real euclidean space may be used in the case of pure point, line, plane calculations, and complex or tensor space in the case of multi-dimensional fusion calculations.
After the topological space corresponding to the target problem is divided into a plurality of subspaces according to the characteristic factors, the method further comprises the following steps: manifold and smooth mapping are respectively established for each subspace, specifically: sequentially establishing manifolds with space structures, which are formed by corresponding neighborhoods and open sets of the neighborhoods in real space, for each subspace, and sequentially constructing smooth functions, the smoothness of which is independent of the shapes of the corresponding subspaces, according to each manifold; and according to the smoothing function, sequentially constructing a smoothing mapping from each subspace to a real space, and respectively obtaining data of the subspace under a coordinate system corresponding to the real space.
It is worth noting that, assuming a topological space, there is a Set of Open sets (Open sets) on the topological space. If there is an open coverage in the topological space, and the following conditions are met: for any one of the open covers, there is a homoembryo mapping from the open set to a certain open subset on the topological space; wherein, the topology structure of the topology space takes the general topology (user topology) when no special statement is made, namely, the topology structure can be expressed as a set of subsets of the pool of the balls; if the intersection set of the open set and the other open set is not empty, the infinite order of the composite mapping of the open set corresponding to the open set and the open set corresponding to the other open set is continuous; where any order derivative function exists and is continuous, then M is called an n-dimensional Manifold (manifield), which can be expressed as:
wherein, the liquid crystal display device comprises a liquid crystal display device,
is an open set in topological space, +.>
Is an open set->
Can be expressed as a homoembryo map of
，/>
Is in topology space->
Is a subset of the opening group; />
Is another open set, add->
Is an open set->
Is mapped to the embryo of->
Is continuous in infinite order.
Using a neighborhood U corresponding to the subspace and algebra in which there is real space or open set in complex space on the neighborhood U, i.e. mapping function
To represent the corresponding manifold (U, -)>
），/>
The mapping function with smoothness independent of the selection of the graph can be constructed by being guided everywhere in a real space or analyzed everywhere in a complex space; the complex space is used for establishing true covariant tensor associated features, and the real space is used for establishing common vectors and scalar features; the mapping function has algebra of spatial structure, which combines various characteristics or defects in multiple dimensions, including: homoembryo, isomorphic, or Lie algebra to handle different transformations.
Referring to fig. 2, a mapping diagram of a mapping function according to an embodiment of the present invention is shown. In the figure, there is a mapping relation F: N
M and mapping relation G.M->
P is smooth mapping, and the composite mapping G is F to N ∈F>
P is also a smooth map; mapping relation F.N->
M is a smooth map, inverse of which +.>
Also smooth mapping, then F is homoembryo mapping. Through the processing of the mapping function, the cooperative transformation can be carried out among different manifolds and different features. Because of the variability of space and mapping, many problems of solidification by conventional algorithms are reasonably solved, the mapping function is not only a functional, but also an algebra with a space structure, various characteristics or defects are combined in a multidimensional manner, and corresponding demand characteristics are found.
Will open up the collection
And its corresponding mapping->
Called a coordinate system (Coordinate system), or Chart (Chart), denoted (O, ψ).
Illustratively, in two dimensions
In, give->
One topological space m= = -on>
，
(1) Find a simplest open cover:
=/>
；
(2) Find a simplest map of the embryo, such as an identity map:
:/>
；
m is a 2-dimensional manifold whose coordinate system can be expressed as {
,/>
) }。/>
It should be noted that the same spatial map G is generally taken
For boundary analysis, two break boundaries constitute dependent variables, one composite boundary constitutes a target variable, and Lie algebra theory may be adopted for corresponding analysis. Thus, an overall analysis framework from point to line, line to face, and face to body can be derived, and conversion from line integration to double integration can be accomplished, illustratively using Stokes (Stokes) formulas, to complete contour to Region (Region) correlation and connectivity calculations.
Illustratively, on a smooth surface, P, Q and R functions are derived from successive first order partial derivatives over the smooth surface and its boundaries, the Stokes formula can be expressed as:
according to the invention, a local coordinate system is introduced into each subspace, and each subspace is described by using the corresponding coordinate system according to the corresponding manifold, so that the description of the geometrical characteristics and the correlation and the corresponding calculation among the geometrical characteristics are facilitated, and the correlation detection and analysis according to specific target problems are facilitated.
A manifold with a space structure, which is formed by a corresponding neighborhood and an open set of the neighborhood in a real space, is sequentially established for each subspace, and specifically comprises the following steps: when the subspace is the heterogeneous space, selecting a lie algebra aspect group to obtain a corresponding manifold; wherein the feature factors acquired in the heterogeneous space are obtained by the steps of: substitution, crossover, mutation or high frequency grabbing; when the subspace is the isomorphic space, the corresponding manifold is obtained according to the Mask matrix, the convolution network, the long-term memory network or the time sequence.
According to the invention, the correlation characteristic description in the region is set up by adopting the differential angle of the manifold corresponding to the subspace, the corresponding manifold is used for different subspaces, the target problem of different dimensions can be solved, further, different expression characteristics can be synthesized, and the characteristic factors of different dimensions are mapped and fused through the mapping function, so that the characteristic of the correlation of each subspace can be described from the local concept of the subspace, and the characteristic of the local subspace is described by adopting a single mathematical formula.
Notably, acquiring the feature factors of the corresponding images includes: if the characteristic factors are used for calculating the characterization points, lines or planes, the real space is used for representing; if the feature factors are used to characterize the multidimensional calculations, a complex space or tensor space is used for the representation. The data types of the feature factors include: vector, scalar, structural features, and Blob data formats.
The method and the device for analyzing the data to be analyzed have the advantages that the data to be analyzed are abstracted, different data types can be processed, the data to be analyzed in vector, scalar, structural characteristics and Blob data formats can be processed, so that the method and the device for analyzing the data to be analyzed can be suitable for modeling and analyzing an image system and a signal system, and different target problems can be processed by adopting the method for analyzing the data to be analyzed, so that the method for analyzing the data to be analyzed is high in adaptability and high in generalization capability.
Illustratively, waveforms are considered as one topological space and mapped with a two-dimensional space.
Illustratively, when three-dimensional curved surface reconstruction is performed, a normal vector set is used as a topological space.
It is worth to say that, the subspace obtained by dividing the topological space is not limited to the specific points in the image, and the key point is to construct the topological space; the mapping is to use topological space elements as original values and real space R elements as target values.
Referring to FIG. 3, a mapping diagram of manifold elements mapped to a two-dimensional space is provided in an embodiment of the present invention, in which a manifold is a three-dimensional structure, an element p is taken from the manifold, and a neighborhood U is taken from the manifold, and a mapping function is constructed
Establishing a corresponding coordinate system for the element p in a two-dimensional space; where element p is a describable feature, which may be a vector, scalar, structural feature, blob, etc., abstracting the entire analysis element.
And step S12, the feature factors of all subspaces are subjected to corresponding manifold mapping to obtain corresponding mapping results, and all the mapping results are fused to obtain a multidimensional real space.
Specifically, sequentially carrying out composite mapping on characteristic factors of the corresponding subspaces and alignment matrixes corresponding to the characteristic factors according to manifolds of the corresponding subspaces to respectively obtain corresponding mapping results, and fusing the obtained mapping results to obtain a multidimensional real space; the composite mapping is obtained according to a mapping function of the smooth mapping after feedback calculation.
It should be noted that, the feature factors of the target problem can be obtained according to a preset rule, various feature factors are generated into a topology space, the topology space is analyzed by a differential structure to obtain the key factors and the alignment matrix to be fused, and the subspace is disassembled according to a composite mapping mode to obtain the fusion space, which comprises: a two-dimensional fused space, a three-dimensional fused space, or a higher-dimensional space. And, by the mapping process of the mapping function, a real space description is obtained, at which time no feature already exists, but a Result (Result). This is because in machine vision, real space is the final result data of interest. If the target problem is complex space, the complex space needs to be converted into real space, because the same manifold can only select real space or complex Euclidean space, although mapping between different manifolds can be properly selected according to functional, the mapping must be converted into real space finally, so that unified calculation and subsequent probability map calculation are facilitated. Specifically, the complex number is split into a real part and an imaginary part at the source end, and then synthesized into a Tensor (Tensor), and then a mapping function is performed with a real space.
Illustratively, tensors of each combination of characteristic factors may be mapped to yield a multidimensional real space according to convolution and pooling operations in the neural network.
And step S13, obtaining a probability analysis table for representing the relation between geometric features according to a probability map constructed by the multidimensional real space and a preset logic processing sequence.
Specifically, according to the relation between the multidimensional real number spaces and a preset logic processing sequence, taking the multidimensional real number spaces as the input of probability functions, and according to the obtained probability diagrams, obtaining a probability analysis table for representing the relation between geometric features; wherein the probability function comprises: bayesian networks or markov random fields.
Illustratively, the preset logic processing order is a filtering operation, a normalizing operation, and an encoding operation.
Illustratively, in the target detection network, the target frame and the target class need to be calculated, and the relationship between the multidimensional real space is the correspondence relationship between the target frame and the target class.
It is worth to say that, according to the preset logic processing sequence, the multidimensional real space is used as the input of a Bayesian network or a Markov random field to obtain a probability analysis table containing characteristic characterization.
Illustratively, in reinforcement learning, a Markov random field contains an evaluation of the characteristics of the input, activity, feedback, and maximum benefit made to the various paths.
The invention adopts the probability of calculating the relation between the characteristic geometric features to be possibly closely related with the requisite event, thereby ensuring the accuracy of the appearance geometric analysis and the completeness of objective facts, and having higher reliability and stronger applicability.
And step S14, obtaining a corresponding probability result according to the probability analysis table, and processing the corresponding target problem according to the probability result.
Wherein, the feature factors of each subspace are mapped by the corresponding manifold, and the method further comprises the following steps: feedback is carried out for mapping of the corresponding manifold, specifically: and carrying out feedback calculation on the mapping of the corresponding manifold according to the selected evaluation function so as to enable the constructed corresponding manifold to conform to the characteristics of the target problem data. After feeding back the mapping for the corresponding manifold, the method further comprises: the probability map is logically adjusted according to the evaluation function.
It should be noted that, because a plurality of expressions of possibility appear in the reasoning process and corresponding causal relationships exist, a probability map needs to be constructed, and a conditional probability is generated between two nodes, see fig. 4, which is a schematic diagram of the probability map constructed according to the embodiment of the present invention, so that
Represents a Directed Acyclic Graph (DAG), wherein +.>
Represents the set of all nodes in the graph, and E represents the set of directed connecting segments, and let +.>
For a random variable represented by a node in its directed acyclic graph, node +.>
The joint probability of (2) can be expressed as:
thus, the entire inference probability map is completed in order to build the probability analysis table of the algorithm.
Specifically, according to the shape and the position of the subspace region, different evaluation functions are selected for carrying out. The whole target problem processing may be evaluated, or the target problem after the subspace is divided may be evaluated, because the calculation process itself has decomposition and combination, which is not limited herein. In addition, due to the diversity of data processing, the evaluation of machine learning or mapping functions is counted by using the original region of feature calculation as much as possible, so that the rationality of data and manifold construction is preferentially inspected, the final result of calculation needs to be evaluated with the accuracy actually, if a certain logic processing process is abnormal, the accuracy is reduced, and the algorithm improvement is needed to be carried out on the process, so that the probability map is modified.
The invention also provides an embodiment for analyzing the two-dimensional point array data, which comprises the following steps of firstly preprocessing the data: dividing the point array into a plurality of blocks in an unsupervised manner; secondly, manifold and mapping functions are defined for different blocks, specifically: dividing the multiple areas into different tasks according to the characteristic factors, wherein the tasks comprise: straight line fitting, frequency feature extraction and skeleton analysis.
According to different block definitions, different manifold and mapping functions, the first block is illustratively processed by using a least square method and Huber loss minimum chemistry, the second block is processed by using a Haar wavelet basis to extract frequency characteristics of each period of time; wherein n is a positive integer.
After corresponding manifold processing, obtaining a relation between multidimensional real space, and constructing a probability map and an evaluation function; and (3) performing accuracy evaluation so as to perform feedback calculation and adjustment of a probability map, wherein the accuracy evaluation comprises the following steps: and (3) carrying out accuracy rate assessment on the calculated data and the actual data, and if the data of a certain area is abnormal, re-assessing the manifold corresponding to the area and modifying the probability map model to carry out logic change. Wherein from a manifold point of view, the computation data is the result of an operation of some elements in the topology space, which is an element in a group (or ring or domain), and may even comprise a functional. Because of the multiplicity of passes, a value or function may not meet the needs of a particular analysis, some strategic management of these results is done, and the object of this management may be called a cluster.
Referring to fig. 5, an appearance geometry analyzing system according to an embodiment of the present invention includes: a subspace partitioning module 51, a manifold and mapping processing module 52, a probability map construction module 53 and a processing module 54.
The subspace dividing module 51 is configured to obtain feature factors of corresponding images according to a determined target problem to be analyzed, and divide a topological space corresponding to the target problem into a plurality of subspaces according to the feature factors; wherein the objective problem includes: measurement based on machine vision, defect detection or template search; each subspace contains one or more feature factors.
After the topological space corresponding to the target problem is divided into a plurality of subspaces according to the characteristic factors, the method further comprises the following steps: manifold and smooth mapping are respectively established for each subspace, specifically: sequentially establishing manifolds with space structures, which are formed by corresponding neighborhoods and open sets of the neighborhoods in real space, for each subspace, and sequentially constructing smooth functions, the smoothness of which is independent of the shapes of the corresponding subspaces, according to each manifold; and according to the smoothing function, sequentially constructing a smoothing mapping from each subspace to a real space, and respectively obtaining data of the subspace under a coordinate system corresponding to the real space.
According to the invention, a local coordinate system is introduced for each subspace, so that the description of the geometrical characteristics and the correlation and corresponding calculation among the geometrical characteristics are facilitated, and the correlation detection and analysis are facilitated according to specific target problems.
A manifold with a space structure, which is formed by a corresponding neighborhood and an open set of the neighborhood in a real space, is sequentially established for each subspace, and specifically comprises the following steps: when the subspace is the heterogeneous space, selecting a lie algebra aspect group to obtain a corresponding manifold; wherein the feature factors acquired in the heterogeneous space are obtained by the steps of: substitution, crossover, mutation or high frequency grabbing; when the subspace is the isomorphic space, the corresponding manifold is obtained according to the Mask matrix, the convolution network, the long-term memory network or the time sequence.
According to the invention, the correlation characteristic description in the region is set up by adopting the differential angle of the manifold corresponding to the subspace, the corresponding manifold is used for different subspaces, the target problem of different dimensions can be solved, further, different expression characteristics can be synthesized, and the characteristic factors of different dimensions are mapped and fused through the mapping function, so that the characteristic of the correlation of each subspace can be described from the local concept of the subspace, and the characteristic of the local subspace is described by adopting a single mathematical formula.
Notably, acquiring the feature factors of the corresponding images includes: if the characteristic factors are used for calculating the characterization points, lines or planes, the real space is used for representing; if the feature factors are used to characterize the multidimensional calculations, a complex space or tensor space is used for the representation. The data types of the feature factors include: vector, scalar, structural features, and Blob data formats.
It should be noted that, the subspace dividing module 51 is mainly configured to determine a specific target problem, select and construct a data dimension of the target problem, and divide a corresponding topological space according to a feature factor according to the dimension of the target data, so that the subspace obtained by division is given to the manifold and the mapping processing module 52 for constructing a corresponding manifold and mapping function.
The method and the device for analyzing the data to be analyzed have the advantages that the data to be analyzed are abstracted, different data types can be processed, the data to be analyzed in vector, scalar, structural characteristics and Blob data formats can be processed, so that the method and the device for analyzing the data to be analyzed can be suitable for modeling and analyzing an image system and a signal system, and different target problems can be processed by adopting the method for analyzing the data to be analyzed, so that the method for analyzing the data to be analyzed is high in adaptability and high in generalization capability.
And the manifold and mapping processing module 52 is used for processing the feature factors of each subspace through the mapping of the corresponding manifold to obtain the corresponding mapping results, and fusing the mapping results to obtain the multidimensional real space.
Specifically, sequentially carrying out composite mapping on characteristic factors of the corresponding subspaces and alignment matrixes corresponding to the characteristic factors according to manifolds of the corresponding subspaces to respectively obtain corresponding mapping results, and fusing the obtained mapping results to obtain a multidimensional real space; the composite mapping is obtained according to a mapping function of the smooth mapping after feedback calculation.
It should be noted that, the manifold and mapping processing module 52 is mainly configured to receive the subspaces processed by the subspace partitioning module 51, establish corresponding manifolds and mapping functions for each subspace, and fuse and integrate feature factors of the subspaces to represent a real space through manifold processing, so that the probability map construction module 53 constructs a probability map according to the multidimensional real space.
The probability map construction module 53 is configured to obtain a probability likelihood analysis table for characterizing the relationship between geometric features according to the probability map constructed by the multidimensional real space and the preset logic processing sequence.
Specifically, according to the relation between the multidimensional real number spaces and a preset logic processing sequence, taking the multidimensional real number spaces as the input of probability functions, and according to the obtained probability diagrams, obtaining a probability analysis table for representing the relation between geometric features; wherein the probability function comprises: bayesian networks or markov random fields.
The invention adopts the probability of calculating the relation between the characteristic geometric features to be possibly closely related with the requisite event, thereby ensuring the accuracy of the appearance geometric analysis and the completeness of objective facts, and having higher reliability and stronger applicability.
It should be noted that, the probability map construction module 53 receives the manifold and the multidimensional real space output by the mapping processing module 52, constructs a probability map corresponding to the whole target problem through a bayesian network or a markov random field, so as to obtain a connection between the possible event and the necessary event, and transmits the obtained probability analysis table to the processing module 54 for corresponding processing.
And the processing module 54 is configured to obtain a corresponding probability result according to the probability likelihood analysis table, and process a corresponding target problem according to the probability result.
Wherein, the feature factors of each subspace are mapped by the corresponding manifold, and the method further comprises the following steps: feedback is carried out for mapping of the corresponding manifold, specifically: and carrying out feedback calculation on the mapping of the corresponding manifold according to the selected evaluation function so as to enable the constructed corresponding manifold to conform to the characteristics of the target problem data. After feeding back the mapping for the corresponding manifold, the method further comprises: the probability map is logically adjusted according to the evaluation function.
It should be noted that, the processing module 54 processes specific target problems according to the probability likelihood analysis table transmitted by the probability map construction module 53, including measurement detection based on machine vision, defect detection or template matching, and selects an evaluation function to perform feedback calculation on the manifold and mapping processing module 52, so as to adjust the probability map construction module 53.
The invention divides the topological space of the target problem to be analyzed into a plurality of subspaces, uses corresponding manifold for different subspaces, can solve the target problem of different dimensions, further can synthesize different expression features, maps and fuses the feature factors of different dimensions through a mapping function, and can describe the feature of the relevance of each subspace from the local concept of the subspace instead of describing the feature of the local subspace by applying a single mathematical formula; moreover, by calculating the probability of representing the relation between geometric features, the relation is possibly closely related to the requisite event, so that the accuracy of appearance geometric analysis and the completeness of objective facts can be ensured, and therefore, the accuracy of measurement, defect detection or template search based on machine vision can be improved, and the method has higher reliability and higher applicability.
It will be appreciated by those skilled in the art that embodiments of the present application may also provide a computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application 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 application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. 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 apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, 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.
The foregoing is merely a preferred embodiment of the present invention, and it should be noted that modifications and variations could be made by those skilled in the art without departing from the technical principles of the present invention, and such modifications and variations should also be regarded as being within the scope of the invention.
Claims (10)
1. A method of geometric analysis of an appearance, comprising:
according to the determined target problem to be analyzed, obtaining a characteristic factor of a corresponding image, and dividing a topological space corresponding to the target problem into a plurality of subspaces according to the characteristic factor; wherein the objective problem includes: measurement based on machine vision, defect detection or template search; each subspace containing one or more feature factors;
the characteristic factors of each subspace are subjected to mapping treatment of corresponding manifold to obtain corresponding mapping results, and the mapping results are fused to obtain a multidimensional real space;
obtaining a probability analysis table for representing the relation between geometric features according to a probability map constructed by the multidimensional real space and a preset logic processing sequence;
and obtaining a corresponding probability result according to the probability analysis table, and processing the corresponding target problem according to the probability result.
2. The geometric analysis method of claim 1, wherein the processing of the feature factors of each subspace through the mapping of the corresponding manifold to obtain the corresponding mapping results, and fusing the mapping results to obtain the multidimensional real space comprises:
sequentially carrying out composite mapping on characteristic factors of the corresponding subspaces and alignment matrixes corresponding to the characteristic factors according to manifold shapes of the corresponding subspaces to respectively obtain corresponding mapping results, and fusing the obtained mapping results to obtain a multidimensional real space; the composite mapping is obtained according to a mapping function of the smooth mapping after feedback calculation.
3. The apparent geometry analyzing method according to claim 1, further comprising, after dividing a topological space corresponding to the target problem into a plurality of subspaces according to the feature factor: manifold and smooth mapping are respectively established for each subspace, specifically:
sequentially establishing manifolds with space structures, which are formed by corresponding neighborhoods and open sets of the neighborhoods in real space, for each subspace, and sequentially constructing smooth functions, the smoothness of which is independent of the shapes of the corresponding subspaces, according to each manifold;
and according to the smoothing function, sequentially constructing a smoothing mapping from each subspace to a real space, and respectively obtaining data of the subspace under a coordinate system corresponding to the real space.
4. A geometric analysis method according to claim 3, wherein the creating, for each subspace, a manifold having a spatial structure comprising a corresponding neighborhood and an open set of the neighborhood in real space, specifically:
when the subspace is the heterogeneous space, selecting a lie algebra aspect group to obtain a corresponding manifold; wherein the feature factors acquired in the heterogeneous space are obtained by the steps of: substitution, crossover, mutation or high frequency grabbing;
when the subspace is the isomorphic space, the corresponding manifold is obtained according to the Mask matrix, the convolution network, the long-term memory network or the time sequence.
5. The method of claim 1, wherein the obtaining a probability likelihood analysis table characterizing a relationship between geometric features from a probability map constructed from the multi-dimensional real space and a predetermined logic processing order comprises:
according to the relation between the multidimensional real number spaces and a preset logic processing sequence, taking the multidimensional real number spaces as the input of probability functions, and obtaining a probability analysis table for representing the relation between geometric features according to the obtained probability map; wherein the probability function comprises: bayesian networks or markov random fields.
6. The geometric analysis method of claim 1, wherein the processing of the feature factors of each subspace through the mapping of the corresponding manifold further comprises: feedback is carried out for mapping of the corresponding manifold, specifically:
and carrying out feedback calculation on the mapping of the corresponding manifold according to the selected evaluation function so as to enable the constructed corresponding manifold to conform to the characteristics of the target problem data.
7. The method of claim 6, further comprising, after feeding back the mapping for the corresponding manifold:
the probability map is logically adjusted according to the evaluation function.
8. The geometric analysis method of claim 1, wherein the obtaining feature factors of the corresponding image includes:
if the characteristic factors are used for calculating the characterization points, lines or planes, the real space is used for representing;
if the feature factors are used to characterize the multidimensional calculations, a complex space or tensor space is used for the representation.
9. The method of geometric analysis of a visual appearance of claim 8, wherein the data type of the feature factor comprises: vector, scalar, structural features, and Blob data formats.
10. A geometric analysis system, comprising:
the subspace dividing module is used for acquiring characteristic factors of corresponding images according to the determined target problems to be analyzed, and dividing a topological space corresponding to the target problems into a plurality of subspaces according to the characteristic factors; wherein the objective problem includes: measurement based on machine vision, defect detection or template search; each subspace containing one or more feature factors;
the manifold and mapping processing module is used for obtaining corresponding mapping results by mapping the characteristic factors of each subspace through corresponding manifold, and fusing the mapping results to obtain a multidimensional real space;
the probability map construction module is used for obtaining a probability analysis table for representing the relation between geometric features according to the probability map constructed by the multidimensional real space and the preset logic processing sequence;
and the processing module is used for obtaining a corresponding probability result according to the probability analysis table and processing the corresponding target problem according to the probability result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310517795.2A CN116229089B (en) | 2023-05-10 | 2023-05-10 | Appearance geometric analysis method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310517795.2A CN116229089B (en) | 2023-05-10 | 2023-05-10 | Appearance geometric analysis method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116229089A true CN116229089A (en) | 2023-06-06 |
| CN116229089B CN116229089B (en) | 2023-07-14 |
Family
ID=86575424
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310517795.2A Active CN116229089B (en) | 2023-05-10 | 2023-05-10 | Appearance geometric analysis method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116229089B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040091152A1 (en) * | 2002-11-12 | 2004-05-13 | Brand Matthew E. | Method for mapping high-dimensional samples to reduced-dimensional manifolds |
| CN110781766A (en) * | 2019-09-30 | 2020-02-11 | 广州大学 | Grassmann manifold discriminant analysis image recognition method based on characteristic spectrum regularization |
| CN113901679A (en) * | 2021-12-13 | 2022-01-07 | 中国南方电网有限责任公司超高压输电公司广州局 | Reliability analysis method and device for power system and computer equipment |
| WO2022051546A1 (en) * | 2020-09-02 | 2022-03-10 | The General Hospital Corporation | Methods for identifying cross-modal features from spatially resolved data sets |
| WO2022167774A1 (en) * | 2021-02-04 | 2022-08-11 | Benevolentai Technology Limited | Graph embedding systems and apparatus |
| CN115048983A (en) * | 2022-05-17 | 2022-09-13 | 北京理工大学 | Counterforce sample defense method of artificial intelligence system based on data manifold topology perception |
-
2023
- 2023-05-10 CN CN202310517795.2A patent/CN116229089B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040091152A1 (en) * | 2002-11-12 | 2004-05-13 | Brand Matthew E. | Method for mapping high-dimensional samples to reduced-dimensional manifolds |
| CN110781766A (en) * | 2019-09-30 | 2020-02-11 | 广州大学 | Grassmann manifold discriminant analysis image recognition method based on characteristic spectrum regularization |
| WO2022051546A1 (en) * | 2020-09-02 | 2022-03-10 | The General Hospital Corporation | Methods for identifying cross-modal features from spatially resolved data sets |
| WO2022167774A1 (en) * | 2021-02-04 | 2022-08-11 | Benevolentai Technology Limited | Graph embedding systems and apparatus |
| CN113901679A (en) * | 2021-12-13 | 2022-01-07 | 中国南方电网有限责任公司超高压输电公司广州局 | Reliability analysis method and device for power system and computer equipment |
| CN115048983A (en) * | 2022-05-17 | 2022-09-13 | 北京理工大学 | Counterforce sample defense method of artificial intelligence system based on data manifold topology perception |
Non-Patent Citations (2)
| Title |
|---|
| SERGUEI BARANNIKOV 等: "Manifold Topology Divergence: a Framework for Comparing Data Manifolds", ARXIV:2106.04024V2, pages 1 - 22 * |
| 冀治航 等: "基于k密集近邻算法的局部Fisher向量编码方法", 大连理工大学学报, vol. 60, no. 4, pages 411 - 419 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116229089B (en) | 2023-07-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Dalca et al. | Learning conditional deformable templates with convolutional networks | |
| Liu et al. | Graph-based method for face identification from a single 2D line drawing | |
| Furukawa | SOM of SOMs | |
| Zhou et al. | Point cloud denoising review: from classical to deep learning-based approaches | |
| Makantasis et al. | Rank-r fnn: A tensor-based learning model for high-order data classification | |
| WO2023164929A1 (en) | Multi-source remote sensing image fusion method and apparatus, device and storage medium | |
| Bhattacharjee et al. | Single index Fréchet regression | |
| Phalak et al. | Scan2plan: Efficient floorplan generation from 3d scans of indoor scenes | |
| Li et al. | Variational diffusion autoencoders with random walk sampling | |
| Cheng et al. | Graph convolution with low-rank learnable local filters | |
| Miranda et al. | Geometric data analysis based on manifold learning with applications for image understanding | |
| US20210150078A1 (en) | Reconstructing an object | |
| CN116229089B (en) | Appearance geometric analysis method and system | |
| Wittmann et al. | Link Prediction for Flow-Driven Spatial Networks | |
| Gaillard et al. | Learning topology of a labeled data set with the supervised generative gaussian graph | |
| Muñoz et al. | Manifold learning for coherent design interpolation based on geometrical and topological descriptors | |
| Han et al. | Large-scale traffic data imputation using matrix completion on graphs | |
| Kalaie et al. | A Geometric Deep Learning Framework for Generation of Virtual Left Ventricles as Graphs | |
| Chowdhury et al. | A State‐of‐the‐Art Computer Vision Adopting Non‐Euclidean Deep‐Learning Models | |
| Cai et al. | Ske2Grid: skeleton-to-grid representation learning for action recognition | |
| Nguyen et al. | Fast temporal wavelet graph neural networks | |
| Wang et al. | On the statistical analysis of complex tree-shaped 3D objects | |
| Jeong et al. | Progressive tree-like curvilinear structure reconstruction with structured ranking learning and graph algorithm | |
| Zheng et al. | Manifold learning | |
| Falih et al. | Collaborative multi-view attributed networks mining |
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 | ||
| CP03 | Change of name, title or address | ||
| CP03 | Change of name, title or address |
Address after: 510000 3, building 9, two South Road, Dashi street, Dashi street, Guangzhou, Guangdong, Panyu District, China. Patentee after: Guangzhou Yihong Intelligent Equipment Co.,Ltd. Address before: 101-301, building 3, No.9, Huijiang Shinan 2nd Road, Dashi street, Panyu District, Guangzhou, Guangdong 511400 Patentee before: GUANGZHOU EHOLLY INTELLIGENT EQUIPMENT Co.,Ltd. |