A node network consistency matching method
By adopting the combination of global label allocation and local matching patterns in the node network, the problem of inefficient matching in large-scale dynamic node networks is solved, and fast consistency comparison and efficient node matching are achieved, which is suitable for parallel processing of large-scale Graph node.
Patent Information
- Application Number
- CN202410472331.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-04-19
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-04-19
AI Technical Summary
In large-scale dynamic node networks, the traditional node network consistency matching algorithm is inefficient and cannot effectively handle changes in dynamic node names and graph structures, resulting in a decrease in matching efficiency.
Initializing global label allocation and dynamic switching to local matching mode, matching neighbor nodes through local structural information, applying recursively using local matching algorithm, and processing complex nodes in combination with global planning algorithms, reducing the number of nodes in the matching process and improving efficiency.
It improves the efficiency of node network consistency matching and system operation efficiency, can handle large-scale graph node comparison, reduces data conflicts, and enhances the universality and processing capabilities of the algorithm.
Smart Images

Figure CN118353790B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of matching algorithms, and in particular to a node network consistency matching method. Background Art
[0002] Node network consistency matching technology is a key technology in the field of complex network analysis. It is mainly used to compare and analyze the structural and functional similarities between two or more large-scale network graphs.
[0003] In social networks, nodes represent individuals or organizations, and edges represent social connections. Node network consistency matching technology can be used to analyze the similarities between different social networks, such as comparing the social network structures of different countries or regions, or studying the evolution of social networks over time. This is of great significance for understanding social behavior patterns, predicting social trends, and formulating social network strategies. Nodes in a transportation network can be cities, transportation hubs, or intersections, and edges represent roads or routes. Node network consistency matching technology can help analyze the transportation network structure of different cities or regions, thereby optimizing traffic flow, reducing congestion, and improving transportation efficiency. In the field of network security, nodes can represent devices or users in the network, and edges can represent data flows or communications. By comparing the consistency between a normal network and an attacked network, potential security threats can be identified and the security of the network can be improved.
[0004] As the scale of the network continues to increase, the consistency check of the node network becomes very complicated. Traditional algorithms usually use unique node names and graph traversal methods, which require high information before network matching. However, in a dynamic node network, the name in the node is used as node privacy, and a dynamic adjustment strategy will be adopted, making the traditional matching algorithm unusable. A new node matching algorithm needs to be proposed. Only relying on the graph structure for network matching, a single entry node is usually selected for comparison. However, for large-scale network structures, the node matching efficiency is greatly reduced.
[0005] In view of this, in order to improve the efficiency of network matching and comparison, we proposed a node network consistency matching method. Summary of the invention
[0006] The purpose of the present invention is to provide a node network consistency matching method to solve the problems raised in the above background technology.
[0007] In order to solve the above technical problems, one of the purposes of the present invention is to provide a node network consistency matching method, comprising the following steps:
[0008] S1. Initialization and global label assignment: The network nodes are represented in the form of a graph, and an initial label is assigned to each node using a global planning algorithm (the degree of the node is generally selected as the label);
[0009] S2, Dynamically switch to local matching: When a local structure is detected in the network, the node matching algorithm switches to the local matching mode;
[0010] S3, local matching process: using the information of the local structure detected in step S2 to match neighbor nodes, the specific process includes the following:
[0011] S3.1. Starting from a known node (such as an input node), determine the nodes that must be matched through simple reasoning based on the structure and properties of the circuit; the structure of the circuit includes power supply, load, wire, switch; the properties of the circuit include current, voltage, resistance, electric power, and electric energy;
[0012] S3.2, assign new, unique labels to these matching nodes instead of relying on the initial label generation algorithm;
[0013] S3.3, using a local matching algorithm to search for possible matching nodes in the circuit, which helps to reduce the number of nodes that need to be processed during the matching process;
[0014] S3.4. When some matching nodes are found, these nodes are deleted from the list of nodes to be matched, and the matching information is recorded in the node equivalence file;
[0015] S3.5, use the node equivalence file as input to tell the local matching algorithm which nodes should be pre-matched;
[0016] S3.6, the local matching algorithm will continue to search for matching nodes among the remaining nodes until all nodes have been checked or a suitable matching node is found;
[0017] S3.7. In this process, if you encounter a matching problem that cannot be solved, you can call the global planning algorithm to help solve the problem;
[0018] S3.8, Recursive application: The local matching algorithm is recursively applied to the newly matched nodes, gradually expanding the matching area and refining the matching results.
[0019] As a further improvement of the technical solution, in step S1, initialization and global label allocation specifically include:
[0020] Initialization: When the node matching algorithm starts, the network nodes are represented in the form of a graph, where node units and connections exist as different types of modules. The node unit module is used to carry specific data and operations, and the connection module is responsible for the association and information transmission between different node units.
[0021] Global label assignment: Use a global planning algorithm to assign an initial label to each node; these labels are set based on the type of node and the number of connections between the nodes. This step uses node invariants as labels to ensure that the labels remain unchanged under graph isomorphism mapping.
[0022] As a further improvement of the technical solution, in step S2, dynamically switching to local matching specifically includes:
[0023] Detecting local structure: Node matching algorithms assess whether there is local structure in the network, such as symmetry or repeated patterns;
[0024] Local matching: Once the local structure is identified, the node matching algorithm switches to local matching mode; in this mode, the node matching algorithm focuses on matching the neighborhood of nodes rather than the entire network.
[0025] As a further improvement of the technical solution, in step S3, the local matching process includes:
[0026] Neighbor node partitioning: For each pair of matched nodes, the node matching algorithm only partitions their neighbor nodes instead of the entire graph;
[0027] Matching based on local information: Node matching algorithms use local information to match neighbor nodes, which usually involves matching neighbor nodes based on their labels and the connection relationships between them.
[0028] As a further improvement of the present technical solution, the local matching algorithm at least includes a region matching algorithm (mainly a grayscale matching algorithm), a feature matching algorithm and a phase matching algorithm.
[0029] As a further improvement of the present technical solution, in the local matching algorithm, the methods used to make the matching cost include the following: SAD (sum of absolute deviations), SSD (sum of squares), ZSAD (zero mean sum of squares), BT (absolute value), NCC (normalization), ZNCC (zero mean normalization), Census (stereo matching statistics), StarCensus (binocular stereo matching), rank (ranking function), etc.
[0030] As a further improvement of this technical solution, it also includes:
[0031] S4. Error isolation and handling; specifically including:
[0032] Error detection: During the matching process, if inconsistent labels or other signs of errors are found, the node matching algorithm will mark these nodes as potential errors;
[0033] Error location: Through local matching and recursive processes, the node matching algorithm attempts to isolate and locate errors. Specifically, the node matching algorithm first performs matching operations in a local range, and preliminarily screens out areas where errors may exist by continuously comparing and analyzing the characteristics and attributes of the nodes. Subsequently, a recursive process is used to dig deeper into these areas, gradually narrowing the scope of the error, and ultimately achieving accurate isolation and location of the error. This may involve multiple iterations, each of which is based on the previous matching result.
[0034] As a further improvement of this technical solution, it also includes:
[0035] S5. Special processing of symmetric nodes; specifically including:
[0036] Symmetry identification: For highly symmetrical node units, the node matching algorithm uses a local matching algorithm to quickly determine the result of any matching of two nodes, avoiding unnecessary global re-labeling;
[0037] Optimized runtime: With this optimization, the node matching algorithm is significantly more efficient in handling symmetric nodes.
[0038] As a further improvement of this technical solution, it also includes:
[0039] S6. Result output and analysis; namely:
[0040] Generate matching reports: The node matching algorithm outputs a node equivalence and error report indicating the differences between the two networks.
[0041] As a further improvement of this technical solution, it also includes:
[0042] S7, end or iteration; specifically including:
[0043] Successful match: If the node matching algorithm successfully matches all nodes and no errors are found, the verification process ends;
[0044] Iterative Improvement: If there are errors, users can iterate based on the output to improve the matching results by pre-matching more nodes or adjusting the input.
[0045] The second object of the present invention is to provide a node network consistency matching system platform device, including a processor, a memory, and a computer program stored in the memory and running on the processor, and the processor is used to implement the steps of the above-mentioned node network consistency matching method when executing the computer program.
[0046] A third object of the present invention is to provide a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned node network consistency matching method are implemented.
[0047] Compared with the prior art, the present invention has the following beneficial effects:
[0048] 1. In this node network consistency matching method, a solution of regional node fusion is proposed. By fusing separated nodes, parallel network node matching is promoted to achieve fast consistency comparison, thereby improving system operation efficiency. It can be mainly used to solve the comparison algorithm of large-scale node graphs, compare the consistency of graph nodes, and analyze inconsistent node information.
[0049] 2. In the node network consistency matching method, the proposed node matching algorithm does not require nodes to have unique tags, which can improve the versatility of the algorithm; improve the characteristics of the nodes, use a hierarchical method to merge some nodes into one node, and reduce the comparison scale;
[0050] 3. In this node network consistency matching method, a strategy suitable for parallelization is proposed, which can quickly process large-scale graph nodes; in the parallelization process, queues and middleware are used to improve the efficiency of processing task nodes and reduce data conflicts. BRIEF DESCRIPTION OF THE DRAWINGS
[0051] Figure 1 It is an exemplary overall method principle flow chart of the present invention;
[0052] Figure 2 It is a structural diagram of an exemplary electronic computer platform device in the present invention. DETAILED DESCRIPTION
[0053] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0054] Example 1
[0055] like Figure 1 As shown, this embodiment provides a node network consistency matching method, comprising the following steps:
[0056] S1. Initialization and global label assignment: The network nodes are represented in the form of a graph, and an initial label is assigned to each node using a global planning algorithm (the degree of the node is generally selected as the label);
[0057] In this step, initialization and global label allocation specifically include:
[0058] Initialization: When the node matching algorithm starts, the network nodes are represented in the form of a graph, where node units and connections exist as different types of modules. The node unit module is used to carry specific data and operations, and the connection module is responsible for the association and information transmission between different node units.
[0059] Global label assignment: Use a global planning algorithm to assign an initial label to each node; these labels are set based on the type of node and the number of connections between the nodes. This step uses node invariants as labels to ensure that the labels remain unchanged under graph isomorphism mapping.
[0060] Specifically, in real-world scenarios, nodes are often associated with useful information, which is considered as labels and used to describe the characteristics of nodes. However, it is difficult to obtain a complete set of labels for all nodes. Only a portion of nodes have labels, and those unlabeled nodes need to be obtained through prediction, so a node classification task is required.
[0061] ;
[0062] use Represents a graph, is a node set, is the edge set. The subset of nodes with labels in The goal of node classification is to use graph (all nodes) and The label information of θ is used to learn a mapping to predict the labels of unlabeled nodes.
[0063] Furthermore, in many data sets, the graph is not complete and some connections between nodes are missing, either because they are not observed or because of the natural evolution of the graph. Represents a graph, is a node set, For edge sets, are all possible pairs of nodes. The complement of an edge is defined as The purpose of link prediction is Assign a value to each edge in , indicating the possibility of this edge appearing.
[0064] S2, Dynamically switch to local matching: When a local structure is detected in the network, the node matching algorithm switches to the local matching mode;
[0065] In this step, dynamically switching to local matching specifically includes:
[0066] Detecting local structure: Node matching algorithms assess whether there is local structure in the network, such as symmetry or repeated patterns;
[0067] Local matching: Once the local structure is identified, the node matching algorithm switches to local matching mode; in this mode, the node matching algorithm focuses on matching the neighborhood of nodes rather than the entire network.
[0068] S3, local matching process: using the information of the local structure detected in step S2 to match neighbor nodes; in this step, the local matching process includes:
[0069] Neighbor node partitioning: For each pair of matched nodes, the node matching algorithm only partitions their neighbor nodes instead of the entire graph;
[0070] Matching based on local information: Node matching algorithms use local information to match neighbor nodes, which usually involves matching neighbor nodes based on their labels and the connection relationships between them.
[0071] Specifically, the statistical features at the node level usually include node degree, node centrality, clustering coefficient, motifs, etc. Among them, node degree and centrality distinguish nodes from the connection relationship of a single node; the clustering coefficient measures the proportion of closed triangles in the local neighborhood of a node from a structural perspective. It measures the tightness of the clustering of the node's neighbor nodes (the degree of mutual connection between neighbor nodes).
[0072] Common methods for calculating local variables of clustering coefficients are as follows:
[0073] ;
[0074] In the formula, For Node The tightness of the neighbor node clustering, the numerator represents the node The number of edges between neighbor nodes, the denominator represents the node The number of edges generated by connecting all neighbor nodes to each other, Is a node degree; For edge sets, For Node The edge between two neighbor nodes, Representation Node The node set of .
[0075] Furthermore, the simplest way to extract graph-level features is to aggregate node-level features as a graph-level representation. However, this method is completely based on local node information and may miss important overall characteristics of the graph. Therefore, it is preferred to use the Weisfeiler-Lehman kernel algorithm for processing. Among them, the WL kernel is an iterative neighborhood aggregation strategy that aggregates node-level features containing more information into a graph-level representation. The basic idea of the WL algorithm is to first assign an initial label to each node. , generally choose the degree of the node as the label. Then, by hashing the multiset of the current labels in the neighborhood of the node, iteratively assign a new label to each node. After iterative K re-labeling, the label of each node aggregates the K-hop neighborhood structure, and these labels are used to calculate histograms or other summary statistics as the feature representation of the graph. The WL algorithm can effectively determine whether the graph is isomorphic. The specific algorithm formula is:
[0076] ;
[0077] In the formula, For Node Iteration The next label, represents a hash function, For Node Iteration The next label, Representation Node The node set of .
[0078] Furthermore, the specific process of the local matching process includes the following:
[0079] S3.1. Starting from a known node (such as an input node), determine the nodes that must be matched through simple reasoning based on the structure and properties of the circuit; the structure of the circuit includes power supply, load, wire, switch; the properties of the circuit include current, voltage, resistance, electric power, and electric energy;
[0080] S3.2, assign new, unique labels to these matching nodes instead of relying on the initial label generation algorithm;
[0081] S3.3, using a local matching algorithm to search for possible matching nodes in the circuit, which helps to reduce the number of nodes that need to be processed during the matching process;
[0082] S3.4. When some matching nodes are found, these nodes are deleted from the list of nodes to be matched, and the matching information is recorded in the node equivalence file;
[0083] S3.5, use the node equivalence file as input to tell the local matching algorithm which nodes should be pre-matched;
[0084] S3.6, the local matching algorithm will continue to search for matching nodes among the remaining nodes until all nodes have been checked or a suitable matching node is found;
[0085] S3.7. In this process, if you encounter a matching problem that cannot be solved, you can call the global planning algorithm to help solve the problem;
[0086] S3.8, Recursive application: The local matching algorithm is recursively applied to the newly matched nodes, gradually expanding the matching area and refining the matching results.
[0087] Furthermore, the local matching algorithm at least includes a region matching algorithm (mainly a grayscale matching algorithm), a feature matching algorithm and a phase matching algorithm.
[0088] Among them, in the region matching algorithm, in essence, the region matching algorithm based on the principle of photometric invariance often uses the grayscale of the neighborhood window as the matching primitive and the correlation degree as the basis for judgment, which can obtain a denser disparity map. It is classified from two aspects: grayscale correlation and non-parametric transformation. The research on grayscale correlation algorithm mainly focuses on the window construction of matching cost aggregation, mainly including: fixed window algorithm, multi-window association algorithm and adaptive window algorithm.
[0089] The performance of the algorithm depends on two properties of the window: first, the pixels in the window should have the same depth, and the number of pixels with different depths should be as small as possible; second, the window should contain enough grayscale changes that are easy to identify and calculate. The idea of the adaptive algorithm is to construct a scale function with grayscale value and depth value as independent variables to select the window.
[0090] A prominent disadvantage of the region matching algorithm is that it is often difficult to obtain accurate matching results for textureless areas due to the insufficient sharpness of the correlation function changes and the difficulty in retaining the depth discontinuity characteristics.
[0091] Grayscale matching algorithm: After epipolar correction, select one from the left image. [2m+1,2m+1] For a region of the same size, select a region of the same size on the same row on the right, traverse all the pixels in this row on the right image from left to right, use the calculation of the correlation value as the matching criterion, and find the corresponding matching point. Different correlation functions have different properties. For example, the commonly used correlation formulas are:
[0092] Grayscale difference square formula, normalized grayscale difference square formula, grayscale cross-correlation formula, normalized grayscale cross-correlation formula, grayscale cross-correlation coefficient formula, normalized grayscale cross-correlation coefficient formula.
[0093] The difference between the above formulas is that different correlation values will result in different calculated correlation peaks.
[0094] Furthermore, in the local matching algorithm, the following methods can be used to make the matching cost: SAD, SSD, ZSAD, BT, NCC, ZNC, Census, StarCensus, rank, etc.
[0095] Among them, SAD is the absolute value of the corresponding pixel difference in the image sequence, that is, the absolute value of the pixels of the two matching blocks is subtracted, and then all the values in the matching window are added up. The smaller this value is, the higher the matching degree is; SSD is the sum of squares of the corresponding pixel differences in the image sequence; ZSAD is the zero mean sum of squares; BT is the absolute value of the pixel gray value difference;
[0096] NCC is normalization, which is used to calculate and represent the correlation of images; ZNCC is zero mean normalization;
[0097] Specifically, NCC and ZNCC are both common cross-correlation calculation methods in template matching. ZNCC is more robust than NCC. It subtracts the mean value in the window from the formula and is more resistant to changes in illumination. The specific formula is as follows:
[0098] ;
[0099] ;
[0100] In the formula, represents the original image, represents the template image, Indicates the number of pixels (elements) in the template, , Represent the standard deviation of the original image and the template image respectively, , Represent the mean of the original image and template image respectively;
[0101] Furthermore, ZNCC can be regarded as the cosine value of the angle between two vectors, that is, when the angle between two vectors tends to 0, that is, when the cosine value tends to 1, the two vectors are more consistent, and vice versa, they are less correlated;
[0102] Census is stereo matching statistics; Census transformation is to convert pixel grayscale into a bit string using the local grayscale difference in the neighborhood of pixel Q. The idea is very simple. By comparing the grayscale value of the pixel in the neighborhood window (the window size is nxm, n and m are both odd numbers) with the grayscale value of the central pixel of the window, the Boolean value Q obtained by comparison is mapped to a bit string, and finally the value of the bit string is used as the Census transformation value Cs of the central pixel; StarCensus is binocular stereo matching;
[0103] Rank is the sorting function.
[0104] S4. Error isolation and handling; specifically including:
[0105] Error detection: During the matching process, if inconsistent labels or other signs of errors are found, the node matching algorithm will mark these nodes as potential errors;
[0106] Error location: Through local matching and recursive processes, the node matching algorithm attempts to isolate and locate errors. Specifically, the node matching algorithm first performs matching operations in a local range, and preliminarily screens out areas where errors may exist by continuously comparing and analyzing the characteristics and attributes of the nodes. Subsequently, a recursive process is used to dig deeper into these areas, gradually narrowing the scope of the error, and ultimately achieving accurate isolation and location of the error. This may involve multiple iterations, each of which is based on the previous matching result.
[0107] S5. Special processing of symmetric nodes; specifically including:
[0108] Symmetry identification: For highly symmetrical node units, the node matching algorithm uses a local matching algorithm to quickly determine the result of any matching of two nodes, avoiding unnecessary global re-labeling;
[0109] Optimized runtime: With this optimization, the node matching algorithm is significantly more efficient in handling symmetric nodes.
[0110] S6. Result output and analysis; namely:
[0111] Generate matching reports: The node matching algorithm outputs a node equivalence and error report indicating the differences between the two networks.
[0112] S7, end or iteration; specifically including:
[0113] Successful match: If the node matching algorithm successfully matches all nodes and no errors are found, the verification process ends;
[0114] Iterative Improvement: If there are errors, users can iterate based on the output to improve the matching results by pre-matching more nodes or adjusting the input.
[0115] like Figure 2 As shown, this embodiment also provides a node network consistency matching system platform device, which includes a processor, a memory, and a computer program stored in the memory and running on the processor.
[0116] The processor includes one or more processing cores. The processor is connected to the memory via a bus. The memory is used to store program instructions. When the processor executes the program instructions in the memory, the steps of the above-mentioned node network consistency matching method are implemented.
[0117] Optionally, the memory can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.
[0118] In addition, the present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the steps of the above-mentioned node network consistency matching method are implemented.
[0119] Optionally, the present invention also provides a computer program product comprising instructions, which, when executed on a computer, enables the computer to execute the steps of the node network consistency matching method in each aspect described above.
[0120] A person of ordinary skill in the art can understand that the process of implementing all or part of the steps of the above-mentioned embodiments can be completed by hardware, or can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, and the above-mentioned storage medium can be a read-only memory, a disk or an optical disk, etc.
[0121] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.
Claims
1. A node network consistency matching method, characterized in that: The steps include: S1. Initialization and global label assignment: The network nodes are represented in the form of a graph, and an initial label is assigned to each node using a global planning algorithm; S2, Dynamically switch to local matching: When a local structure is detected in the network, the node matching algorithm switches to the local matching mode; S3, local matching process: using the information of the local structure detected in step S2 to match neighbor nodes, the specific process includes the following: S3.
1. Starting from the known nodes, determine the nodes that must be matched through simple reasoning based on the structure and properties of the circuit; the structure of the circuit includes power supply, load, wire, switch; the properties of the circuit include current, voltage, resistance, electric power, and electric energy; S3.2, assign new, unique labels to these matching nodes; S3.3, using a local matching algorithm to search for possible matching nodes in the circuit; S3.
4. When some matching nodes are found, these nodes are deleted from the list of nodes to be matched, and the matching information is recorded in the node equivalence file; S3.5, use the node equivalence file as input to tell the local matching algorithm which nodes should be pre-matched; S3.6, the local matching algorithm will continue to search for matching nodes among the remaining nodes until all nodes have been checked or a suitable matching node is found; S3.
7. In this process, if you encounter a matching problem that cannot be solved, you can call the global planning algorithm to help solve the problem; S3.8, Recursive application: The local matching algorithm is recursively applied to the newly matched nodes, gradually expanding the matching area and refining the matching results.
2. The node network consistency matching method according to claim 1, characterized in that: In step S1, initialization and global label allocation specifically include: Initialization: When the node matching algorithm starts, the network nodes are represented in the form of a graph, where node units and connections exist as different types of modules. The node unit module is used to carry specific data and operations, and the connection module is responsible for the association and information transmission between different node units. Global label assignment: An initial label is assigned to each node using a global planning algorithm; these labels are set based on the type of node and the number of connections to the node.
3. The node network consistency matching method according to claim 1, characterized in that: In step S2, dynamically switching to local matching specifically includes: Detecting local structure: The node matching algorithm evaluates whether there is local structure in the network; Local matching: Once the local structure is identified, the node matching algorithm switches to local matching mode.
4. The node network consistency matching method according to claim 1, characterized in that: In step S3, the local matching process includes: Neighbor node partitioning: For each pair of matched nodes, the node matching algorithm only partitions their neighbor nodes instead of the entire graph; Matching based on local information: Node matching algorithms use local information to match neighbor nodes, which usually involves matching neighbor nodes based on their labels and the connection relationships between them.
5. The node network consistency matching method according to claim 1, characterized in that: The local matching algorithm at least includes a region matching algorithm, a feature matching algorithm and a phase matching algorithm.
6. The node network consistency matching method according to claim 1, characterized in that: In the local matching algorithm, the methods used to make the matching cost include the following: absolute error and SAD, square sum SSD, zero mean absolute error and ZSAD, absolute value BT, normalized NCC, zero mean normalized ZNC, stereo matching statistics Census, binocular stereo matching StarCensus, and ranking function rank.
7. The node network consistency matching method according to claim 1, characterized in that: Also includes: S4, Error isolation and handling; Specifically include: Error detection: During the matching process, if inconsistent labels or other signs of errors are found, the node matching algorithm will mark these nodes as potential errors; Error location: Through local matching and recursive processes, the node matching algorithm attempts to isolate and locate errors. Specifically, the node matching algorithm first performs matching operations in a local range, and preliminarily screens out areas where errors may exist by continuously comparing and analyzing the characteristics and attributes of the nodes. Subsequently, a recursive process is used to dig deeper into these areas, gradually narrowing the scope of the error, and ultimately achieving accurate isolation and location of the error. This may involve multiple iterations, each of which is based on the previous matching result.
8. The node network consistency matching method according to claim 1, characterized in that: Also includes: S5. Special processing of symmetric nodes; specifically including: Symmetry identification: For highly symmetrical node units, the node matching algorithm uses a local matching algorithm to quickly determine the result of any matching of two nodes, avoiding unnecessary global re-labeling; Optimized runtime: With this optimization, the node matching algorithm is significantly more efficient in handling symmetric nodes.
9. The node network consistency matching method according to claim 1, characterized in that: Also includes: S6. Result output and analysis; Right now: Generate matching reports: The node matching algorithm outputs a node equivalence and error report indicating the differences between the two networks.
10. The node network consistency matching method according to claim 1, characterized in that: Also includes: S7, end or iteration; Specifically include: Successful match: If the node matching algorithm successfully matches all nodes and no errors are found, the verification process ends; Iterative Improvement: If there are errors, users can iterate based on the output to improve the matching results by pre-matching more nodes or adjusting the input.
