CN108052680B - Image data target identification Enhancement Method based on data map, Information Atlas and knowledge mapping - Google Patents
Image data target identification Enhancement Method based on data map, Information Atlas and knowledge mapping Download PDFInfo
- Publication number
- CN108052680B CN108052680B CN201810023920.3A CN201810023920A CN108052680B CN 108052680 B CN108052680 B CN 108052680B CN 201810023920 A CN201810023920 A CN 201810023920A CN 108052680 B CN108052680 B CN 108052680B
- Authority
- CN
- China
- Prior art keywords
- image
- entity
- data map
- knowledge mapping
- relation
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
Abstract
The present invention is a kind of image data target identification Enhancement Method based on data map, Information Atlas, knowledge mapping framework.Be mainly used for solving conventional images recognition methods can not recognition training concentrate and do not mark the problem of image recognition of classification, belong to distributed computing and Software Engineering technology crossing domain.Key is from the existing image type recognition result based on deep learning method, three layers of map are constructed according to existing image resource, unidentified image category is subjected to characteristic matching in data map and obtains initial matching result, identified image category is carried out to relationship match on Information Atlas and obtains intermediate match result, indirect interaction relationship match is finally carried out in knowledge mapping, the confidence level of intermediate match result and sequence are calculated, the highest matching image classification of User reliability is recommended.
Description
Technical field
The present invention is a kind of image data target identification enhancing based on data map, Information Atlas and knowledge mapping framework
Method.Be mainly used for solving conventional images recognition methods can not recognition training concentrate and do not mark the problem of image recognition of classification, belong to
In distributed computing and Software Engineering technology crossing domain.
Background technique
Knowledge mapping has become with the form with markd digraph the strong tools for indicating knowledge, and can assign
Text information is semantic.Knowledge mapping is that project, entity or user are showed by way of node, will by way of side
The node of interaction between each other is chained up the figure of construction, and the side between node can indicate any semantic relation.Knowledge graph
The building of spectrum is divided into information extraction, knowledge fusion and knowledge according to the process of knowledge acquisition and processes three levels, defines knowledge graph
Spectrum is the entity with attribute by netted knowledge base made of relational links, and researching value is can be with the smallest generation
The information accumulated in internet is organized into the knowledge that can be utilized by valence, to know by implementation of inference conceptual retrieval and graphically
Know and shows.Data are obtained by observation number or other basic individual items.The back that information is combined by data and data
Scape is conveyed, and is suitable for analysis and is explained.Knowledge is the general understanding and experience obtained from the information of accumulation, can be pushed away according to knowledge
Measure new background.
Most traditional image recognition technology is manual identified, and having passed through can be identified by the method for deep learning and marked
The image of knowledge, machine gradually simulate an equation in identifying same class object repeatedly, this equation, which approaches, to be identified pair
The common trait of elephant finally reaches the purpose of identification, but the method for deep learning not can solve the identification for not identifying image
Problem.The present invention mainly goes correctly to identify as far as possible by the framework by data map, Information Atlas, knowledge mapping
More objects not being identified.
Summary of the invention
Technical problem: the present invention is a kind of image data mesh based on data map, Information Atlas and knowledge mapping framework
Identify other Enhancement Method.It introduces data map, Information Atlas, knowledge mapping framework and carries out knowledge reasoning to image is not identified, from
And realize the image that in picture or shot by camera arrives of identifying of automated intelligent.The present invention is by the structure of image, feature,
Frequency and interactive relation, semantic relation are individually placed in data map, Information Atlas and knowledge mapping, by means of this framework
Provide the highest recognition result of confidence level.
Technical solution: the present invention is a kind of tactic method, and the image object that can be applied to picture or video camera is known
Not, facilitate to solve the problems, such as not identifying in current machine study and do not identify image.The present invention is by the structure of image, feature,
Frequency and interactive relation, semantic relation are individually placed in data map, Information Atlas and knowledge mapping, then will not identify figure
As being compared with the data in data map, initial matching result is obtained；Again with the friendship in identified image and Information Atlas
Mutual relation is compared, and intermediate match result is obtained by calculation；Finally knowledge reasoning is carried out in knowledge mapping to obtain indirectly
Interactive relation, it is exactly final result that we are identified that wherein confidence level is highest.
Architecture: this patent is first against having identified that image establishes data map, Information Atlas, knowledge graph music stand
Structure obtains the Data Matching degree R(Ai with Ai for unidentified image A ergodic data map), from big to small according to R (Ai)
Sequence select initial matching result A1, A2 ... An；By initial matching result and identified image set { B } in Information Atlas into
Row relationship match obtains relationship match degree Y(Ai) and X (Ai), according to the sequence of X (Ai) from big to small select A1, A2 ... Ak,
Ai ' and { B } are subjected to computational reasoning in knowledge mapping, obtain the highest result of confidence level.Currently, we can use
Deep learning identifies identified image, and the present invention on this basis, introduces data map, Information Atlas, knowledge graph music stand
Structure carries out knowledge reasoning to image is not identified, thus realize automated intelligent identify in picture or shot by camera arrives
Image.Building data map, Information Atlas, knowledge mapping is given below to illustrate:
Data map: data map can record the essential attribute in image entities, including color phenotypes etc., but not have
In the case where context of co-text, themselves is nonsensical.Data map can use the number such as array, chained list, queue, tree, stack, figure
It is expressed according to structure.Data map also can recorde structure included in image entities appearance frequency, including structure, the time and
The frequency of three levels in space.Data map can describe the tightness degree of different images entity associated, and we term it density, can
To reflect which entity relationship is close, which entity relationship is sparse.But data map can only carry out static analysis, Wu Fabiao
Up to the interactive relation between entity.Also the accuracy of data is not analyzed on data map simultaneously, in fact it could happen that different
Entity but indicate same meaning, such as tomato and tomato, attribute frequency possessed by both image entities be all it is identical,
This generates data redundancies；
Information Atlas: information is conveyed by the context after data and data combination, by concept mapping and
The information of suitable analysis and explanation after correlativity combination.Information Atlas can be expressed according to relational database.Information
Data cleansing is carried out on map, eliminates redundant data.Information Atlas can recorde the interactive relation between entity；
Knowledge mapping: knowledge mapping is closed according to semantic between the further perfect entity of data map and Information Atlas
System, improves the side density and node density of knowledge mapping by information inference and entity link, knowledge mapping without architectural characteristic
So that itself can be with seamless link.Information inference needs the support of dependency rule, the new relation obtained by reasoning it is correct
Degree Cr can be obtained by following equation.New relation between entity 1 and entity 2 can be expressed as Cr(E1, E2), z indicates all
New indirect interaction relationship,Indicate new relation matching degree,Between one between presentation-entity 1 and entity 2
Relationship is connect, is to think that this new relation is set up when correctness is more than a threshold value.
The utility model has the advantages that the present invention is a kind of image data mesh based on data map, Information Atlas and knowledge mapping framework
Other Enhancement Method is identified, there is the following remarkable advantage:
(1) present invention constructs data map, Information Atlas and knowledge mapping three-tier architecture, progressive, realizes efficient
Identification；
(2) currently based on the method for deep learning can not recognition training concentrate and do not mark the image category of classification, the present invention
It can solve the image category problem for not marking classification in training set, recommend the recognition result of the high confidence level of user；
(3) image and map are carried out Data Matching and relationship match by the present invention, constantly diminution identification range, pass through knowledge
Reasoning obtains most accurate result.
Detailed description of the invention
Fig. 1 is to data map, the formal definitions of Information Atlas and knowledge mapping.
Fig. 2 is the specific flow chart of the image recognition processes based on data map, Information Atlas and knowledge mapping framework.
Specific embodiment
It is a kind of based on data map, Information Atlas, knowledge mapping framework image-recognizing method detailed process it is as follows:
Step 1) corresponds to 001 in Fig. 2, according to existing image resource, establishes based on data map, Information Atlas, knows
Know the frame of map；
Step 2 corresponds to 002 in Fig. 2, obtains images to be recognized；
Step 3) corresponds to 003 in Fig. 2, and images to be recognized is divided into and has identified image set { Bi } and unidentified image
Two modules of A；
Step 4) corresponds to 004 in Fig. 2, by the entity A i of the unidentified image A obtained based on step 3 and data map
Data Matching is carried out, matching degree R is obtained by formula 1, shown in data map as shown in figure 1.Assuming that entity attribute in data map
(including structure, color, feature, frequency, partial structurtes etc.) and Ai matching value are μ (α), and α indicates that Ai can be with unidentified image
A matched attribute, x expression is all can be with matched attribute:
(1)；
Step 5) corresponds to 005 in Fig. 2, rule of thumb sets an initial value n, according to R (Ai) from big to small suitable
Sequence select A1, A2 ... An；
Step 6) correspond to Fig. 2 in 006, using A1, A2 ... An is as initial matching result；
Step 7) corresponds to 007 in Fig. 2, and the initial matching obtained based on step 6 in conjunction with step 3 as a result, obtain not
Identification image set { B } traverses Information Atlas, finds corresponding interactive relation, by formula 2,3 obtain A1, A2 ...
Relationship match degree Y (A1), Y (A2) ... the Y (An) of An and { B }, shown in Information Atlas as shown in figure 1, Ai and B2, B3, B4 have directly
Connect interactive relation.Assuming that the interactive relation p matching degree in Information Atlas between Bi and Ai is θ (π), π wherein indicates an interaction
Relationship, q indicate all relationships:
(2)；
(3)；
Step 8) corresponds to 008 in Fig. 2, obtains intermediate match degree X(A1 by formula 4), X (A2) ... X (An), wherein R
(Ai) it is independent from each other with Y (Ai), and r1+r2=1,
(4)；
Step 9) correspond to Fig. 2 in 009, rule of thumb set an initial value k (k≤n), according to X (Ai) from greatly to
Small sequence select A1, A2 ... Ak；
Step 10) correspond to Fig. 2 in 010, based on step 8 obtain intermediate match result A1, A2 ... Ak, in knowledge
In map by reasoning available { B } and A1, A2 ... the indirect interaction relationship of Ak, as shown in figure 1 shown in knowledge mapping, Ai '
There is indirect interaction relationship with B5, the correctness Cr of obtained new relation can be obtained by formula 5.It is new between entity 1 and entity 2
Relationship can be expressed as Cr (E1, E2), and z indicates all new indirect interaction relationships,Indicate new relation matching degree,An indirect relation between presentation-entity 1 and entity 2 is to think the relationship when correctness is more than a threshold value
It sets up:
(5)；
Step 11) correspond to Fig. 2 in 011, if relationship set up, calculate confidence level (Confidence), wherein Cr (E1,
E2 the interactive relation) and in Information Atlas is independent from each other, and y1+y2=1, is obtained by formula 6:
(6)；
Step 12) corresponds to 012 in Fig. 2, compares the size of confidence level, the maximum image entities of confidence level are exactly to identify
Final result.
Claims (1)
1. a kind of image data target identification Enhancement Method based on data map, Information Atlas and knowledge mapping, feature exist
In introducing data map, Information Atlas, knowledge mapping framework and make inferences to not identifying image, realize the identification of automated intelligent
Out in picture or image that shot by camera arrives, by the structure of image, feature, frequency and interactive relation, semantic relation
It is placed in data map, Information Atlas and knowledge mapping, provides the highest recognition result of confidence level, the party by means of this framework
Method includes:
Step 1) according to existing image resource, establish based on data map, Information Atlas, knowledge mapping frame；
Step 2 obtains images to be recognized；
Images to be recognized is divided by step 3) has identified image set { BiAnd two modules of unidentified image A；
Step 4) is by the entity A of unidentified the image A and data map that are obtained based on step 3)iData Matching is carried out, by formula 1
Obtain matching degree R (Ai)；
Assuming that entity attribute and Ai matching value are μ (α) in data map, α indicates AiIt can be with a category of unidentified images match
Property, x expression is all can be with matched attribute, and wherein entity attribute includes structure, color, feature, frequency, partial structurtes；
(1)；
Step 5) rule of thumb sets an initial value n, according to R (Ai) sequence from big to small selects A1、A2、… An；
Step 6) is by A1、A2、… AnAs initial matching result；
Initial matching that step 7) is obtained based on step 6) as a result, the unidentified image set { B } obtained in conjunction with step 3) to information
Map is traversed, and is found corresponding interactive relation, is obtained A by formula 2, formula 31、A2、… AnThe relationship match of { B }
Spend Y (A1)、Y(A2)…Y(An), Y (Ai,Bi) it is intermediate variable；
Assuming that B in Information AtlasiWith AiBetween interactive relation p matching degree be θ (π), wherein π indicate an interactive relation, q
Indicate all relationships:
(2)；
(3)；
Step 8) obtains intermediate match degree X(A by formula 41), X (A2)…X(An), wherein R (Ai) and Y (Ai) be independent from each other,
And r1+r2=1,
(4)；
Step 9) rule of thumb sets an initial value k, wherein k≤n, according to X (Ai) sequence from big to small selects A1、A2、…
Ak；
The intermediate match result A that step 10) is obtained based on step 8)1、A2、… Ak, can be obtained in knowledge mapping by reasoning
To { B } and A1、A2、… AkIndirect interaction relationship, the correctness Cr of obtained new relation can obtain by formula 5；
New relation between entity 1 and entity 2 can be expressed as Cr (E1,E2), z indicates all new indirect interaction relationships,
Indicate new relation matching degree, E1→E2An indirect relation between presentation-entity 1 and entity 2, when correctness is more than a threshold value
When think that new relation between entity 1 and entity 2 is set up:
(5)；
If step 11) relationship is set up, confidence level Confidence is calculated, wherein Cr (E1,E2) and Information Atlas in interactive relation
It is independent from each other, and y1+y2=1, it is obtained by formula 6:
(6)；
Step 12) compares the size of confidence level, and the maximum image entities of confidence level are exactly the final result identified.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810023920.3A CN108052680B (en) | 2018-01-10 | 2018-01-10 | Image data target identification Enhancement Method based on data map, Information Atlas and knowledge mapping |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810023920.3A CN108052680B (en) | 2018-01-10 | 2018-01-10 | Image data target identification Enhancement Method based on data map, Information Atlas and knowledge mapping |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108052680A CN108052680A (en) | 2018-05-18 |
| CN108052680B true CN108052680B (en) | 2019-08-23 |
Family
ID=62127004
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810023920.3A Active CN108052680B (en) | 2018-01-10 | 2018-01-10 | Image data target identification Enhancement Method based on data map, Information Atlas and knowledge mapping |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108052680B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108804950A (en) * | 2018-06-09 | 2018-11-13 | 海南大学 | Based on data collection of illustrative plates, modeling and the data-privacy guard method of Information Atlas and knowledge mapping |
| CN108875414A (en) * | 2018-06-09 | 2018-11-23 | 海南大学 | Based on data map, the modeling of Information Atlas and knowledge mapping and data security protection method |
| CN108804945A (en) * | 2018-06-09 | 2018-11-13 | 海南大学 | Based on data collection of illustrative plates, the information privacy protection method of Information Atlas and knowledge mapping |
| CN112528993A (en) * | 2020-12-17 | 2021-03-19 | 济南浪潮高新科技投资发展有限公司 | Method for improving image recognition accuracy rate based on knowledge graph |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106990973A (en) * | 2017-05-25 | 2017-07-28 | 海南大学 | A kind of service software development approach of the value driving based on data collection of illustrative plates, Information Atlas and knowledge mapping framework |
| CN106997399A (en) * | 2017-05-24 | 2017-08-01 | 海南大学 | A kind of classification question answering system design method that framework is associated based on data collection of illustrative plates, Information Atlas, knowledge mapping and wisdom collection of illustrative plates |
| CN107038262A (en) * | 2017-05-30 | 2017-08-11 | 海南大学 | It is a kind of associate frequency calculate based on data collection of illustrative plates, the semantic modeling of Information Atlas and knowledge mapping framework and abstract Enhancement Method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2014089776A1 (en) * | 2012-12-12 | 2014-06-19 | Google Inc. | Ranking search results based on entity metrics |
-
2018
- 2018-01-10 CN CN201810023920.3A patent/CN108052680B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106997399A (en) * | 2017-05-24 | 2017-08-01 | 海南大学 | A kind of classification question answering system design method that framework is associated based on data collection of illustrative plates, Information Atlas, knowledge mapping and wisdom collection of illustrative plates |
| CN106990973A (en) * | 2017-05-25 | 2017-07-28 | 海南大学 | A kind of service software development approach of the value driving based on data collection of illustrative plates, Information Atlas and knowledge mapping framework |
| CN107038262A (en) * | 2017-05-30 | 2017-08-11 | 海南大学 | It is a kind of associate frequency calculate based on data collection of illustrative plates, the semantic modeling of Information Atlas and knowledge mapping framework and abstract Enhancement Method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108052680A (en) | 2018-05-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108052680B (en) | Image data target identification Enhancement Method based on data map, Information Atlas and knowledge mapping | |
| CN103559504B (en) | Image target category identification method and device | |
| CN110555050B (en) | Heterogeneous network node representation learning method based on meta-path | |
| CN104966104B (en) | A kind of video classification methods based on Three dimensional convolution neutral net | |
| CN108920544A (en) | A kind of personalized position recommended method of knowledge based map | |
| CN109816032B (en) | Unbiased mapping zero sample classification method and device based on generative countermeasure network | |
| CN108229578B (en) | Image data target identification method based on three layers of data, information and knowledge map framework | |
| Balakrishnan | Significance of classification techniques in prediction of learning disabilities | |
| CN110321291A (en) | Test cases intelligent extraction system and method | |
| CN113190688B (en) | Complex network link prediction method and system based on logical reasoning and graph convolution | |
| CN103824115A (en) | Open-network-knowledge-base-oriented between-entity relationship deduction method and system | |
| CN113140254B (en) | Meta-learning drug-target interaction prediction system and prediction method | |
| CN111931505A (en) | Cross-language entity alignment method based on subgraph embedding | |
| CN110033097B (en) | Method and device for determining association relation between user and article based on multiple data fields | |
| CN111709244A (en) | Deep learning method for identifying causal relationship of contradictory dispute events | |
| CN111950708A (en) | Neural network structure and method for discovering daily life habits of college students | |
| CN110309300A (en) | A method of identification natural sciences knowledge-ID | |
| CN109800300A (en) | A kind of learning Content recommended method and system | |
| CN108319705A (en) | A kind of individualized teaching process automatic generation method and system based on teaching resource | |
| Liu et al. | Fine-grained machine teaching with attention modeling | |
| CN115687760A (en) | User learning interest label prediction method based on graph neural network | |
| CN105787045B (en) | A kind of precision Enhancement Method for visual media semantic indexing | |
| Foody | Rating crowdsourced annotations: evaluating contributions of variable quality and completeness | |
| CN114254093A (en) | Multi-space knowledge enhanced knowledge graph question-answering method and system | |
| CN114817552A (en) | Exercise association relation processing method, exercise association relation processing device, exercise association relation processing equipment and storage medium |
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 |