CN102663092A - Method for mining and recommending style elements based on clothing graph group - Google Patents
Method for mining and recommending style elements based on clothing graph group Download PDFInfo
- Publication number
- CN102663092A CN102663092A CN2012101040115A CN201210104011A CN102663092A CN 102663092 A CN102663092 A CN 102663092A CN 2012101040115 A CN2012101040115 A CN 2012101040115A CN 201210104011 A CN201210104011 A CN 201210104011A CN 102663092 A CN102663092 A CN 102663092A
- Authority
- CN
- China
- Prior art keywords
- clothing
- image
- style
- proportion
- recommended
- 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
Abstract
The invention provides a method for mining and recommending style elements based on a clothing graph group. The method includes that a user selects or inputs clothing sample images and recommendation requirements for a recommending system; the recommending system preprocesses the inputted clothing sample images, and the specified clothing graph group is built; the recommending system extracts specific features of the clothing images in the specified clothing graph group according to the recommendation requirements; the recommending system performs specific clothing visual element mining for the clothing images; and styles of the clothing images in a database are analyzed according to required and mined clothing visual elements, and various types of clothing are rated. The clothing visual elements of the clothing images are mined by the aid of a feature expression and statistic analysis method, accordingly, design styles of the clothing images are summarized, and the clothing images are retrieved and recommended according to the clothing styles. By the aid of the method, the requirement on novelty when consumers purchase clothing can be met.
Description
Technical field
The present invention relates to a kind of style element and excavate and recommend method, belong to image of clothing and recommend and the analytical approach technical field based on the clothes picture group.
Background technology
Clothes are the effect ever more important in the people life in modern times, and clothes also accounted for larger specific gravity during ecommerce was sold, and the image of clothing commending system can help people effectively to search out required clothes.Existing technology has only adopts the retrieval based on key word; The label for labelling that need take time and effort; And adopt clothes attribute selected in the technology of CBIR all too simple; Ignored important tendency in the clothes attribute, and most be purpose to retrieve identical or similar clothes, can't satisfy the novelty that the clothes consumer pursues.In addition, also will produce bigger help to dress designing, processing and sale industry for the quantitative test of extensive image of clothing, and also not similarly work at present.
Summary of the invention
The objective of the invention is only to adopt retrieval based on key word for what solve that prior art has; The label for labelling that need take time and effort; And adopt clothes attribute selected in the technology of CBIR all too simple, ignored important tendency in the clothes attribute, and majority is a purpose to retrieve identical or similar clothes; Can't satisfy the novelty problem that the clothes consumer pursues, and then provide a kind of style element to excavate and recommend method based on the clothes picture group.
The objective of the invention is to realize through following technical scheme:
A kind of style element based on the clothes picture group is excavated and recommend method, may further comprise the steps:
Steps A, said commending system are set up clothes and are specified the clothes picture group;
Step B, said commending system carries out feature extraction to specifying the image of clothing in the clothes picture group, obtains the style and features code book according to said training feature vector;
Step C, said commending system carry out proportion to the style and features code book with respect to every width of cloth specify image and calculate, and obtain style proportion matrix;
Step D, said commending system obtain the livery visual element according to the expressive force function of each style and features of style proportion matrix computations according to this expressive force function;
Step e, said commending system carries out feature extraction to the image of clothing in the storehouse to be recommended, obtains eigenmatrix to be recommended, calculates the to be recommended proportion vector of said livery visual element in this image to be recommended;
Step F, said commending system are calculated the similarity of above-mentioned style proportion matrix and proportion to be recommended vector, the ranking results that output is recommended.
Beneficial effect of the present invention: the present invention extracts through feature representation and the statistical analysis technique clothes visual element to image of clothing, and then sums up the design style of image of clothing, thereby according to garment language image of clothing is retrieved and recommended.The present invention can satisfy the novelty that the consumer pursues when buying clothes, for the user recommends clothes, and help in the clothes industry extensive image of clothing to be carried out quantitative test.The present invention is through the style element of extraction clothes, and then the design style of summary image of clothing.
Embodiment
To do further detailed description to the present invention below: present embodiment provided detailed embodiment, but protection scope of the present invention is not limited to following embodiment being to implement under the prerequisite with technical scheme of the present invention.
The related a kind of style element based on the clothes picture group of present embodiment is excavated and recommend method, comprising:
Steps A: said commending system is set up and is specified the clothes picture group.
Steps A specifically comprises:
A1: said commending system is selected the image of clothing of the fashionable dress news conference in a certain brand and season, or the image of clothing of different-style, sets up and specifies the clothes picture group, be respectively with this clothes result who recommends out to meet corresponding stylistic category; Specify the interior image of clothes picture group to comprise some images, preferably, the image number is five.
A2: carry out the normalization operation to specifying image and the image in the storehouse to be recommended in the clothes picture group, with image size normalizing to M*N.
Step B: said commending system carries out feature extraction to specifying the image of clothing in the clothes picture group, obtains the style and features code book according to said training feature vector;
Said step B specifically comprises:
B1: extract characteristics of image to specifying each width of cloth image of clothing in the clothes picture group, obtain set of eigenvectors;
B2: utilize K-means that above-mentioned set of eigenvectors is carried out cluster operation, obtain the style and features code book.Preferably, classification is counted K and is made as 50;
Step C: said commending system carries out proportion to the style and features code book with respect to every width of cloth specify image and calculates, and obtains style proportion matrix;
Said step C specifically comprises:
C1: specify the image I in the clothes picture group to each width of cloth clothes, calculate i code book
C i Number, obtain
Count (I, C i ), i=1,2 ... K;
C2: calculate the proportion vector of each code book, obtain style proportion matrix:
Step D: said commending system obtains the livery visual element according to the expressive force function of each style and features of style proportion matrix computations according to this expressive force function;
Said step 4 comprises:
D1: the expressive force function that calculates each style and features
Wherein, avg (.) is the expectation formula, and var (.) is a formula of variance.
D2: choose preceding N
sThe individual style and features that makes that the expressive force function is maximum, with this as the livery visual element
Step e: said commending system carries out feature extraction to the image of clothing in the storehouse to be recommended, obtains eigenmatrix to be recommended, calculates the to be recommended proportion vector of said livery visual element in this image to be recommended.
To the proportion vector that sorts of waiting of the image calculation livery visual element of waiting to sort,
Step F: said commending system calculates the similarity of above-mentioned style proportion matrix and proportion to be recommended vector, the output recommendation results.
Calculate the similarity of above-mentioned style proportion matrix and proportion to be recommended vector,
Wherein, R (.) is a similarity function.Preferably, adopt the cosine similarity function,
Step G: said commending system calculates distribution and the ratio of livery visual element in image to be recommended according to above-mentioned style proportion matrix and eigenmatrix to be recommended, carries out the style analysis.
Step G specifically comprises:
G1: train a plurality of appointment clothes picture groups,, calculate and its most similar style appointment clothes picture group for each proper vector in the image to be recommended, with δ as activation value;
G2: add up in this image in all proper vectors, the ratio of above-mentioned activation and distribution are with this analysis and scoring as style.
Claims (7)
1. the style element based on the clothes picture group is excavated and recommend method, it is characterized in that, may further comprise the steps:
A: said commending system is specified the image of clothing collection arbitrarily, finds wherein general character and personality elements as target in order to study;
B: said commending system carries out feature extraction to the image of clothing of target tightening, obtains the style and features code book according to said training feature vector;
C: said commending system carries out proportion to the style and features code book with respect to every width of cloth specify image and calculates, and obtains style proportion matrix;
D: said commending system obtains the garment language element according to the expressive force function of each style and features of style proportion matrix computations according to this expressive force function;
E: said commending system carries out feature extraction to the image of clothing in the storehouse to be recommended, obtains eigenmatrix to be recommended, calculates the to be recommended proportion vector of said livery visual element in this image to be recommended;
F: said commending system calculates the similarity of above-mentioned style proportion matrix and proportion to be recommended vector, the output recommendation results.
2. the method for claim 1 is characterized in that, said steps A specifically comprises step:
A1: said commending system is selected the image of clothing of the fashionable dress news conference in a certain brand and season, or the image of clothing of different-style, sets up and specifies the clothes picture group, be respectively with this clothes result who recommends out to meet corresponding stylistic category; Specify the interior image of clothes picture group to comprise some images;
A2: carry out the normalization operation to specifying image and the image in the storehouse to be recommended in the clothes picture group, with image size normalizing to M*N.
3. the method for claim 1 is characterized in that, said step B specifically comprises step:
B1: extract characteristics of image to specifying each width of cloth image of clothing in the clothes picture group, obtain set of eigenvectors;
B2: utilize K-means that above-mentioned set of eigenvectors is carried out cluster operation, obtain the style and features code book.
4. the method for claim 1 is characterized in that, said step C specifically comprises step:
C1: specify the image I in the clothes picture group to each width of cloth clothes, calculate i code book
C i Number, obtain
Count (I, C i ), i=1,2 ... K;
C2: calculate the proportion vector of each code book, obtain style proportion matrix:
5. the method for claim 1 is characterized in that, said step D specifically comprises step:
D1: the expressive force function that calculates each style and features
Wherein, avg (.) is the expectation formula, and var (.) is a formula of variance;
D2: choose preceding N
sThe individual style and features that makes that the expressive force function is maximum, with this as the livery visual element
6. the method for claim 1 is characterized in that, said step e is to the proportion vector that sorts of waiting of the image calculation livery visual element of waiting to sort,
7. the method for claim 1 is characterized in that, calculates above-mentioned style proportion matrix and the vectorial similarity of proportion to be recommended in said step F,
Wherein, R (.) is a similarity function, preferably, adopts the cosine similarity function,
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210104011.5A CN102663092B (en) | 2012-04-11 | 2012-04-11 | Method for mining and recommending style elements based on clothing graph group |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210104011.5A CN102663092B (en) | 2012-04-11 | 2012-04-11 | Method for mining and recommending style elements based on clothing graph group |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102663092A true CN102663092A (en) | 2012-09-12 |
| CN102663092B CN102663092B (en) | 2015-01-28 |
Family
ID=46772583
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210104011.5A Active CN102663092B (en) | 2012-04-11 | 2012-04-11 | Method for mining and recommending style elements based on clothing graph group |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102663092B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103337027A (en) * | 2013-06-20 | 2013-10-02 | 咸晓梦 | Clothing information pushing method and system |
| CN103559504A (en) * | 2013-11-04 | 2014-02-05 | 北京京东尚科信息技术有限公司 | Image target category identification method and device |
| WO2014106213A1 (en) * | 2012-12-31 | 2014-07-03 | Agrawal Vandana | Style recommendation engine and method |
| CN107122548A (en) * | 2017-04-27 | 2017-09-01 | 贵州大学 | Towards the user cognition and the modeling method of quantitative model of Modeling Style |
| CN107391599A (en) * | 2017-06-30 | 2017-11-24 | 中原智慧城市设计研究院有限公司 | Image search method based on style and features |
| CN108171568A (en) * | 2017-12-11 | 2018-06-15 | 武汉纺织大学 | A kind of clothes in knowledge based library recommend method and system |
| CN108287857A (en) * | 2017-02-13 | 2018-07-17 | 腾讯科技（深圳）有限公司 | Expression picture recommends method and device |
| CN109117779A (en) * | 2018-08-06 | 2019-01-01 | 百度在线网络技术（北京）有限公司 | One kind, which is worn, takes recommended method, device and electronic equipment |
| CN110795582A (en) * | 2019-10-31 | 2020-02-14 | 京东方科技集团股份有限公司 | Image recommendation method, system, terminal device and server |
| US11321887B2 (en) * | 2018-12-24 | 2022-05-03 | Accenture Global Solutions Limited | Article design |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101842790A (en) * | 2006-09-14 | 2010-09-22 | 麦歇浦股份有限公司 | Matching the fit of garments to consumers |
| CN102254043A (en) * | 2011-08-17 | 2011-11-23 | 电子科技大学 | Semantic mapping-based clothing image retrieving method |
| WO2011152844A1 (en) * | 2010-06-01 | 2011-12-08 | Hewlett-Packard Development Company, L.P. | Image clustering using a personal clothing model |
| CN102360365A (en) * | 2011-10-08 | 2012-02-22 | 广东工业大学 | Method for generating extension strategy for screening clothing on internet |
-
2012
- 2012-04-11 CN CN201210104011.5A patent/CN102663092B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101842790A (en) * | 2006-09-14 | 2010-09-22 | 麦歇浦股份有限公司 | Matching the fit of garments to consumers |
| WO2011152844A1 (en) * | 2010-06-01 | 2011-12-08 | Hewlett-Packard Development Company, L.P. | Image clustering using a personal clothing model |
| CN102254043A (en) * | 2011-08-17 | 2011-11-23 | 电子科技大学 | Semantic mapping-based clothing image retrieving method |
| CN102360365A (en) * | 2011-10-08 | 2012-02-22 | 广东工业大学 | Method for generating extension strategy for screening clothing on internet |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2014106213A1 (en) * | 2012-12-31 | 2014-07-03 | Agrawal Vandana | Style recommendation engine and method |
| CN103337027A (en) * | 2013-06-20 | 2013-10-02 | 咸晓梦 | Clothing information pushing method and system |
| CN103559504A (en) * | 2013-11-04 | 2014-02-05 | 北京京东尚科信息技术有限公司 | Image target category identification method and device |
| CN103559504B (en) * | 2013-11-04 | 2016-08-31 | 北京京东尚科信息技术有限公司 | Image target category identification method and device |
| US10824669B2 (en) | 2017-02-13 | 2020-11-03 | Tencent Technology (Shenzhen) Company Limited | Sticker recommendation method and apparatus, server cluster, and storage medium |
| CN108287857A (en) * | 2017-02-13 | 2018-07-17 | 腾讯科技（深圳）有限公司 | Expression picture recommends method and device |
| CN107122548A (en) * | 2017-04-27 | 2017-09-01 | 贵州大学 | Towards the user cognition and the modeling method of quantitative model of Modeling Style |
| CN107391599A (en) * | 2017-06-30 | 2017-11-24 | 中原智慧城市设计研究院有限公司 | Image search method based on style and features |
| CN107391599B (en) * | 2017-06-30 | 2021-01-12 | 中原智慧城市设计研究院有限公司 | Image retrieval method based on style characteristics |
| CN108171568A (en) * | 2017-12-11 | 2018-06-15 | 武汉纺织大学 | A kind of clothes in knowledge based library recommend method and system |
| CN109117779A (en) * | 2018-08-06 | 2019-01-01 | 百度在线网络技术（北京）有限公司 | One kind, which is worn, takes recommended method, device and electronic equipment |
| US11321887B2 (en) * | 2018-12-24 | 2022-05-03 | Accenture Global Solutions Limited | Article design |
| CN110795582A (en) * | 2019-10-31 | 2020-02-14 | 京东方科技集团股份有限公司 | Image recommendation method, system, terminal device and server |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102663092B (en) | 2015-01-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102663092A (en) | Method for mining and recommending style elements based on clothing graph group | |
| Corbiere et al. | Leveraging weakly annotated data for fashion image retrieval and label prediction | |
| US10747826B2 (en) | Interactive clothes searching in online stores | |
| CN104331417B (en) | A kind of matching method of individual subscriber dress ornament | |
| CN108665333B (en) | Commodity recommendation method and device, electronic equipment and storage medium | |
| CN102609523B (en) | The collaborative filtering recommending method classified based on taxonomy of goods and user | |
| CN104424230B (en) | A kind of cyber recommended method and device | |
| US9916613B1 (en) | Automatic color palette based recommendations for affiliated colors | |
| CN107122980B (en) | Method and device for identifying categories to which commodities belong | |
| CN107679960B (en) | Personalized clothing recommendation method based on clothing image and label text bimodal content analysis | |
| CN103927309B (en) | A kind of method and device to business object markup information label | |
| US20140108200A1 (en) | Method and system for recommending search phrases | |
| US10776417B1 (en) | Parts-based visual similarity search | |
| CN106547864A (en) | A kind of Personalized search based on query expansion | |
| CN105045909B (en) | The method and apparatus that trade name is identified from text | |
| CN108959453B (en) | Information extraction method and device based on text clustering and readable storage medium | |
| CN103049513A (en) | Multi-visual-feature fusion method of commodity images of clothing, shoes and bags | |
| CN102819566A (en) | Cross-catalogue indexing method for business images | |
| CN106649276A (en) | Identification method and device for core product word in title | |
| Rubio et al. | Multi-modal joint embedding for fashion product retrieval | |
| CN102831161A (en) | Semi-supervision sequencing study method for image searching based on manifold regularization | |
| CN113034237A (en) | Dress suit recommendation system and method | |
| CN112508114A (en) | Intelligent clothing recommendation system and method | |
| CN109840336B (en) | Clothing design sample recommendation method and device | |
| Li et al. | Sparse representation based visual element analysis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20210409 Address after: Room 206-10, building 16, 1616 Chuangxin Road, Songbei District, Harbin City, Heilongjiang Province Patentee after: Harbin jizuo technology partnership (L.P.) Patentee after: Harbin Institute of Technology Asset Management Co.,Ltd. Address before: 150000 No. 92, West Da Zhi street, Nangang District, Harbin, Heilongjiang. Patentee before: HARBIN INSTITUTE OF TECHNOLOGY |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20210611 Address after: Room 206-12, building 16, 1616 Chuangxin Road, Songbei District, Harbin City, Heilongjiang Province Patentee after: Harbin Institute of Technology Institute of artificial intelligence Co.,Ltd. Address before: Room 206-10, building 16, 1616 Chuangxin Road, Songbei District, Harbin City, Heilongjiang Province Patentee before: Harbin jizuo technology partnership (L.P.) Patentee before: Harbin Institute of Technology Asset Management Co.,Ltd. |
|
| TR01 | Transfer of patent right |