CN104142920A - Online image retrieval system - Google Patents
Online image retrieval system Download PDFInfo
- Publication number
- CN104142920A CN104142920A CN201310161562.XA CN201310161562A CN104142920A CN 104142920 A CN104142920 A CN 104142920A CN 201310161562 A CN201310161562 A CN 201310161562A CN 104142920 A CN104142920 A CN 104142920A
- Authority
- CN
- China
- Prior art keywords
- image
- retrieval system
- cluster
- image retrieval
- ray image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
Abstract
The invention relates to an online image retrieval system and belongs to the field of a software product. The system is characterized by being used for combining the features of movable picture search based on full study on mobile internet, reinforcement learning, and picture search and mining theory; researching an image feature combinatorial optimization method based on parallel reinforcement learning, combining image feature cluster index parallel construction methods of image physical features and semantic features, and designing a practical online image retrieval system based on data updating of logic reinforcement learning and actively discovering strategy of new data, so as to greatly improve the retrieval efficiency, and gain a great advance in the field of the software product.
Description
Technical field
At a ray image retrieval system, belong to software product field.
Background technology
The E-commerce market season Surveillance > > of < < 2012Q3 China Mobile according to China's authoritative mobile Internet third party data mining and integrated marketing mechanism Chinese mugwort matchmaker's consulting (iiMediaResearch) issue shows, within 2011, China Mobile's ecommerce userbase has reached 0.92 hundred million people, to the end of the year 2012, China Mobile's ecommerce userbase will reach 1.46 hundred million people, increase by 58.7% on a year-on-year basis.Expected for the end of the year 2015, mobile e-business userbase will reach 3.48 hundred million people.Within 2011, China Mobile's E-commerce market scale has reached 156.7 hundred million yuan, and to the end of the year 2012, China Mobile's E-commerce market scale will reach 251.5 hundred million yuan, increase on year-on-year basis 60.5%.Expected for the end of the year 2015, mobile e-business market scale will reach 1046.7 hundred million yuan.Mobile e-business is in fast-developing period.
Summary of the invention
Object of the present invention is achieved through the following technical solutions:
At a ray image retrieval system, belong to software product.Be characterized in: fully study on the basis of mobile Internet, intensified learning, picture search and excavation theory, feature in conjunction with moving images search, the characteristics of image cluster index parallel constructing method of image characteristic combination optimization method, combining image physical features and semantic feature and the Data Update of logic-based intensified learning and the new data active discovery strategy of research based on parallel intensified learning, design one actual in ray image retrieval system.
Further, above-mentioned is a kind of at ray image retrieval system, it is characterized in that: described image indexing system is CBIR, groundwork concentrates in identification and the color of Description Image, texture, shape, spatial relationship, and on the measuring similarity of Image Feature Matching, after image characteristics extraction and method for measuring similarity are determined, retrieval has just become the process of searching the image the most similar to given image in image data base.
Further, above-mentioned is a kind of at ray image retrieval system, it is characterized in that: described image is that image is comprised of by certain semantic association relation (cluster index) all image blocks for the overall situation, similar image has the identical cluster index that quantity does not wait.
Further, above-mentioned is a kind of at ray image retrieval system, it is characterized in that: described cluster is on the basis that image is cut apart and assemblage characteristic extracts, and adopts the k-means clustering algorithm based on Map/Reduce to implement cluster to image block.
Further, above-mentioned is a kind of at ray image retrieval system, it is characterized in that: described cluster is that image block to be clustered is stored on different back end dispersedly in its bottom layer realization, each piece can also copy some parts, be stored on different back end, to reach fault-tolerant object, the back end in cluster is in charge of the storage on its place node.
Further, above-mentioned is a kind of at ray image retrieval system, it is characterized in that: described image retrieval is on the basis of image block cluster, regard image block cluster index as vision keyword, all vision keywords have formed vision keyword dictionary, and image is by the feature clustering vector representation consisting of a series of vision keyword.
Embodiment
At a ray image retrieval system, belong to software product.Be characterized in: fully study on the basis of mobile Internet, intensified learning, picture search and excavation theory, feature in conjunction with moving images search, the characteristics of image cluster index parallel constructing method of image characteristic combination optimization method, combining image physical features and semantic feature and the Data Update of logic-based intensified learning and the new data active discovery strategy of research based on parallel intensified learning, design one actual in ray image retrieval system.
One of this project main research is for cutting apart as data research image for Large Scale Graphs and assemblage characteristic extraction problem.This project is intended starting with from cutting apart of image, based on to the considering of local characteristics of image and combination of multiple features, adopts cutting apart and assemblage characteristic extraction problem of intensified learning method research image.And adopt D-S evidence theory to merge learning outcome.
The core pixel of the target and background in the image that user uses mobile terminal to take pictures to obtain has higher determinacy, and pixel is large for the degree of membership of target; And marginal portion is owing to being subject to the interference of other classification pixels, there is higher uncertainty, in addition, user also often shows as uncertainty to the subjectivity of image understanding and the demand of search, the uncertain transformation model of qualitative, quantitative can be expressed the uncertainty of concept, the uncertainty of reduction Concept Hierarchies well, can effectively study the uncertain problem of image in cutting apart, this project adopts the image segmentation algorithm of the uncertain transformation model of qualitative, quantitative to solving the ignorance of traditional images partitioning algorithm to uncertain information.
The final D-S evidence theory that adopts merges the learning outcome of each agent.In supposing the system, there is n agent, after all agent have completed the study of one-period, their Q table is merged.Q (s in normalization Q table, a) value, utilize D-S evidence to merge, first the learning outcome of agent1 and agent2 is merged, then fusion results and agent3 are merged, the like, carry out altogether (n-1) inferior fusion, each fusion is all two fusions between evidence.
Due to the feature extraction of the view data of magnanimity and exponential Feature Combination, need parallelization and work to raise the efficiency.MapReduce can not only be for the treatment of large-scale data, and possesses the characteristics such as the standby management of automatically parallelizing, load balancing and calamity; And the retractility of MapReduce is very good, be better than most of distributed treatment frameworks in the past, therefore, native system has adopted this distributed treatment framework of MapReduce to learn the Feature Combination of moving images classification.
Early stage investigation according to project team to market, does not also have in conjunction with the company of the moving images search of mobile Internet, intensified learning, picture search and digging technology, does not domesticly have other direct competitions opponent at present, so we have larger advantage.Project team possesses stronger technical foundation at aspects such as mobile Internet, intensified learning, picture search and excavations, can guarantee the technical service that is in time to utilize first-strike advantage, and this project will have good industrialization prospect.
Claims (6)
1. at a ray image retrieval system, belong to software product field.Be characterized in: fully study on the basis of mobile Internet, intensified learning, picture search and excavation theory, feature in conjunction with moving images search, the characteristics of image cluster index parallel constructing method of image characteristic combination optimization method, combining image physical features and semantic feature and the Data Update of logic-based intensified learning and the new data active discovery strategy of research based on parallel intensified learning, design one actual in ray image retrieval system.
2. according to claim 1 a kind of at ray image retrieval system, it is characterized in that: described image indexing system is CBIR, groundwork concentrates in identification and the color of Description Image, texture, shape, spatial relationship, and on the measuring similarity of Image Feature Matching, after image characteristics extraction and method for measuring similarity are determined, retrieval has just become the process of searching the image the most similar to given image in image data base.
3. according to claim 2 a kind of at ray image retrieval system, it is characterized in that: described image is for the overall situation, image is comprised of by certain semantic association relation (cluster index) all image blocks, and similar image has the identical cluster index that quantity does not wait.
4. according to claim 3 a kind of at ray image retrieval system, it is characterized in that: described cluster is on the basis that image is cut apart and assemblage characteristic extracts, adopt the k-means clustering algorithm based on Map/Reduce to implement cluster to image block.
5. according to claim 3 a kind of at ray image retrieval system, it is characterized in that: described cluster is that image block to be clustered is stored on different back end dispersedly in its bottom layer realization, each piece can also copy some parts, be stored on different back end, to reach fault-tolerant object, the back end in cluster is in charge of the storage on its place node.
6. according to claim 3 a kind of at ray image retrieval system, it is characterized in that: described image retrieval is on the basis of image block cluster, regard image block cluster index as vision keyword, all vision keywords have formed vision keyword dictionary, and image is by the feature clustering vector representation consisting of a series of vision keyword.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310161562.XA CN104142920A (en) | 2013-05-06 | 2013-05-06 | Online image retrieval system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310161562.XA CN104142920A (en) | 2013-05-06 | 2013-05-06 | Online image retrieval system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104142920A true CN104142920A (en) | 2014-11-12 |
Family
ID=51852096
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310161562.XA Pending CN104142920A (en) | 2013-05-06 | 2013-05-06 | Online image retrieval system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104142920A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104778281A (en) * | 2015-05-06 | 2015-07-15 | 苏州搜客信息技术有限公司 | Image index parallel construction method based on community analysis |
| CN109074659A (en) * | 2016-05-04 | 2018-12-21 | 皇家飞利浦有限公司 | Medical image resources registration |
| CN110297935A (en) * | 2019-06-28 | 2019-10-01 | 京东数字科技控股有限公司 | Image search method, device, medium and electronic equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050004897A1 (en) * | 1997-10-27 | 2005-01-06 | Lipson Pamela R. | Information search and retrieval system |
| CN102508909A (en) * | 2011-11-11 | 2012-06-20 | 苏州大学 | Image retrieval method based on multiple intelligent algorithms and image fusion technology |
| CN103064991A (en) * | 2013-02-05 | 2013-04-24 | 杭州易和网络有限公司 | Mass data clustering method |
| CN103077253A (en) * | 2013-01-25 | 2013-05-01 | 西安电子科技大学 | High-dimensional mass data GMM (Gaussian Mixture Model) clustering method under Hadoop framework |
-
2013
- 2013-05-06 CN CN201310161562.XA patent/CN104142920A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050004897A1 (en) * | 1997-10-27 | 2005-01-06 | Lipson Pamela R. | Information search and retrieval system |
| CN102508909A (en) * | 2011-11-11 | 2012-06-20 | 苏州大学 | Image retrieval method based on multiple intelligent algorithms and image fusion technology |
| CN103077253A (en) * | 2013-01-25 | 2013-05-01 | 西安电子科技大学 | High-dimensional mass data GMM (Gaussian Mixture Model) clustering method under Hadoop framework |
| CN103064991A (en) * | 2013-02-05 | 2013-04-24 | 杭州易和网络有限公司 | Mass data clustering method |
Non-Patent Citations (1)
| Title |
|---|
| 杜晨阳: "分布式聚类算法研究与应用", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104778281A (en) * | 2015-05-06 | 2015-07-15 | 苏州搜客信息技术有限公司 | Image index parallel construction method based on community analysis |
| CN109074659A (en) * | 2016-05-04 | 2018-12-21 | 皇家飞利浦有限公司 | Medical image resources registration |
| CN109074659B (en) * | 2016-05-04 | 2023-04-04 | 皇家飞利浦有限公司 | Medical atlas registration |
| CN110297935A (en) * | 2019-06-28 | 2019-10-01 | 京东数字科技控股有限公司 | Image search method, device, medium and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Gao et al. | 3D model retrieval using weighted bipartite graph matching | |
| CN103186568B (en) | Method and system for accessing tree-structured data | |
| CN103281351A (en) | Cloud server platform capable of efficiently processing and analyzing remote sensing data | |
| CN103235825A (en) | Method used for designing large-quantity face recognition search engine and based on Hadoop cloud computing frame | |
| CN106933833A (en) | A kind of positional information method for quickly querying based on Spatial Data Index Technology | |
| CN105718555A (en) | Hierarchical semantic description based image retrieving method | |
| CN102595102A (en) | Video structurally storing method | |
| CN106294738A (en) | A kind of Intelligent household scene collocation method | |
| CN104142920A (en) | Online image retrieval system | |
| CN116522272A (en) | Multi-source space-time data transparent fusion method based on urban information unit | |
| Xu et al. | Individual stem detection in residential environments with MLS data | |
| CN104376116A (en) | Search method and device for figure information | |
| Choi et al. | Conceptual data modeling for realizing context-aware services | |
| Peifeng et al. | Smart wardrobe system based on android platform | |
| CN106203449A (en) | The approximation space clustering system of mobile cloud environment | |
| CN104142921A (en) | Image feature optimal-combination method based on cloud computing | |
| CN109543087A (en) | A kind of construction method and application method of international three pole data interoperation automotive engine system | |
| CN113139110B (en) | Regional characteristic processing method, regional characteristic processing device, regional characteristic processing equipment, storage medium and program product | |
| CN102831216B (en) | Image retrieval system and method for maintaining geometrical shape information of object | |
| CN104199950A (en) | Method of searching for academic papers on basis of fast matching of image similarities | |
| CN103164504A (en) | Smartphone refined picture searching system and method | |
| CN104850541B (en) | A kind of semantic locations conversion method based on position semantic net | |
| CN108197201B (en) | Mobile cloud data mining method based on public security event | |
| Zhang et al. | Semi-Supervised Cross-Modal Hashing Based on Label Prediction and Distance Preserving | |
| CN103699906A (en) | Mobile face recognition system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20141112 |
|
| DD01 | Delivery of document by public notice | ||
| DD01 | Delivery of document by public notice |
Addressee: Liu Yu Document name: Notification that Application Deemed to be Withdrawn |