CN106844785A - Saliency segmentation-based content-based image retrieval method - Google Patents
Saliency segmentation-based content-based image retrieval method Download PDFInfo
- Publication number
- CN106844785A CN106844785A CN201710151860.9A CN201710151860A CN106844785A CN 106844785 A CN106844785 A CN 106844785A CN 201710151860 A CN201710151860 A CN 201710151860A CN 106844785 A CN106844785 A CN 106844785A
- Authority
- CN
- China
- Prior art keywords
- image
- feature
- training
- foreground target
- visual vocabulary
- 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
- 238000000034 method Methods 0.000 title claims abstract description 73
- 230000011218 segmentation Effects 0.000 title claims abstract description 16
- 230000000007 visual effect Effects 0.000 claims abstract description 34
- 238000012549 training Methods 0.000 claims abstract description 33
- 238000012360 testing method Methods 0.000 claims abstract description 31
- 239000013598 vector Substances 0.000 claims description 28
- 238000000605 extraction Methods 0.000 claims description 9
- 239000000284 extract Substances 0.000 claims description 8
- 238000011017 operating method Methods 0.000 claims description 4
- 238000000638 solvent extraction Methods 0.000 claims description 4
- 238000013459 approach Methods 0.000 claims description 3
- 230000000052 comparative effect Effects 0.000 claims description 3
- 238000000205 computational method Methods 0.000 claims description 3
- 238000003709 image segmentation Methods 0.000 claims description 3
- 238000003064 k means clustering Methods 0.000 claims description 3
- 239000000203 mixture Substances 0.000 claims description 3
- 238000005520 cutting process Methods 0.000 claims description 2
- 238000007689 inspection Methods 0.000 claims 1
- 238000005516 engineering process Methods 0.000 description 6
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 239000008186 active pharmaceutical agent Substances 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000009432 framing Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
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
- G06F16/5838—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content using colour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/237—Lexical tools
- G06F40/242—Dictionaries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration using histogram techniques
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Library & Information Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Image Analysis (AREA)
Abstract
A saliency segmentation based content-based image retrieval method comprises a training process and a testing process; the training process comprises the steps of: step one, establishing a visual vocabulary dictionary: sequentially analyzing each image in the training image set, and establishing a visual vocabulary dictionary for subsequent retrieval; secondly, segmenting the image to obtain a foreground target image and a background area image of the image by using the visual saliency characteristics of the image; respectively extracting color features and texture features of the image from the foreground target image and the background area image; step four, on the basis of the operation of the step one, counting a visual vocabulary distribution histogram of each picture in a database; the test process comprises a fifth step, and the search of the test image can be realized on the basis of the operations from the first step to the fourth step. The invention provides the image retrieval method based on the content and based on the significance segmentation, which effectively reduces the semantic gap and has higher accuracy.
Description
Technical field
The present invention relates to a kind of image search method, more particularly to a kind of image retrieval side of view-based access control model conspicuousness segmentation
Method, belongs to CBIR field.
Background technology
With the development of multimedia technology, the quantity of digital picture presents the growth of geometry level.How in vast and numerous figure
As the resource that user's request is fast and accurately found out in resource is also put in face of people.
CBIR technology allows the pictures of user input one to search with same or like content
Other pictures, it represents the mainstream development trend of image retrieval technologies.
Josef Sivic et al. proposed bag of words in 2006.Its core concept is that one is detected in entire image
A little key points, then extract the local feature of these key points, and these local features are quantified as " visual word " afterwards.It is so every
Width image can be expressed as the characteristic vector on " visual word ".The mature technology of text retrieval can be just used for reference afterwards, should
Retrieval is completed with expansible index and Fast search technique.
The features such as bag of words have simply and can utilize text retrieval mature technology, but there is certain lacking simultaneously
Fall into.Because may including foreground target and background two parts in piece image, if carrying out feature extraction in entire image
Easily cause subject goal not clear and definite enough, by-end characteristic remarkable can so largely effect on image in results such as foreground targets
The result of retrieval.
The content of the invention
In order to the presence semantic gap, the accuracy that overcome the shortcomings of existing image search method are poor, the invention provides
A kind of effective reduction semantic gap, accuracy the CBIR method based on conspicuousness segmentation higher.
The technical solution adopted for the present invention to solve the technical problems is：
A kind of CBIR method based on conspicuousness segmentation, described image search method includes training
Journey and test process；
The training process is comprised the following steps：
Step one, set up visual vocabulary dictionary：Treatment is analyzed to every image that training image is concentrated successively, after being
A visual vocabulary dictionary is set up in continuous retrieval；
Step 2, the vision significance feature using image, segmentation obtain the foreground target figure and Background regional image of image；
Step 3, the color character and textural characteristics that extract image respectively to foreground target figure and Background regional image；
Step 4, on the basis of the operation of step one, in staqtistical data base per pictures visual vocabulary distribution histogram；
The test process includes step 5, on the basis of step one to step 4 operation, test image is examined
Rope, process is as follows：
Step 5.1：Foreground target figure and Background regional image are obtained to the segmentation of test image step 2 methods described；
Step 5.2：Test image foreground target figure and Background regional image are extracted with step 3 methods described to test image
Color character and textural characteristics；
Step 5.3：The visual word of foreground target figure and Background regional image is counted with step 4 methods described to test image
Remittance statistic histogram；
Step 5.4：Each feature that test image is obtained according to step 5.1, step 5.2 and step 5.3 successively with instruction
The character pair for practicing every image in image set carries out distance metric；
Step 5.5：Distance metric to every kind of feature of step 5.4 gained adds up, and obtains total distance, uses D1,
D2... DiRepresent, wherein i represents the amount of images of training image concentration；
Step 5.6：For every image that training image is concentrated, according to D1,D2... DiOrder from small to large is to training
Image sorts and exports.
By the operation of above-mentioned steps, you can realize the retrieval to test image.
Further, it is in the step one, the step of set up visual vocabulary dictionary：
Step 1.1：Color image into gray level image；
Step 1.2：The characteristic vector of every image that training image is concentrated is obtained on the gray level image that step 1.1 is obtained
Represent collection, represented with Fm, wherein F1, F2 ... Fm represents that the characteristic vector of an image represents collection respectively, and m represents training figure
The number of image in image set；The characteristic vector is multidimensional, and its dimension is represented with N, N>2；
The characteristic vector for obtaining every image that training image is concentrated represents that the method for integrating is SIFT algorithms；
Step 1.3：Clustered and obtained each feature for the characteristic vector that the character representation that step 1.2 is obtained is concentrated
The representative feature vector of classification.K classification is obtained, K representative feature is vectorial, and K is positive integer；
The clustering method is K-means clustering algorithms；
The representative feature vector of each feature classification is the barycenter of whole feature vector clusters in this feature classification；
Step 1.4：Visual vocabulary dictionary needed for obtaining later retrieval, the visual vocabulary dictionary is by step 1.4 gained
Each feature classification representative feature vector composition.
Further, in the step 2, the cutting operation step for single image is：
Step 2.1：The Saliency maps of image are calculated, the computational methods of Saliency maps are used based on the straight of regional correlation
Square figure comparative approach；
Step 2.2：The Saliency maps obtained using step 2.1, iteration operation image partitioning algorithm, respectively obtain image
Foreground target figure and Background regional image, image segmentation algorithm use GrabCut.
In the step 3, the extraction operating procedure for single image is：
Step 3.1：The color space of foreground target figure and Background regional image is transformed into HSV color spaces；
Step 3.2 extracts color character in the H passages and channel S of foreground target figure, extracting method using calculate H values with
The algorithm of the statistic histogram of S values；
Step 3.3 extracts color character in the H passages and channel S of Background regional image, extracting method using calculate H values with
The algorithm of the statistic histogram of S values；
Step 3.4 texture feature extraction on the V passages of foreground target figure, extracting method is calculated using local binary patterns
Method；
Step 3.5 texture feature extraction on the V passages of Background regional image, extracting method is calculated using local binary patterns
Method.
In the step 4, for single image, the operating procedure for setting up character representation is：
Step 4.1：The foreground target figure and Background regional image that step 2 is obtained are converted into gray level image；
Step 4.2：The visual vocabulary dictionary that step 1 is obtained is combined in the gray level image that step 4.1 is obtained, is counted respectively
The statistic histogram of the visual vocabulary of foreground target figure and Background regional image.
In the step 5.4, the method that distance metric is used is COS distance.
The present invention is foreground target figure and Background regional image by by means of vision significance model, dividing the image into, so
The local features such as the global characteristics such as color, texture and SIFT are extracted respectively on this two classes image afterwards, finally use these features
In image retrieval.This search method can more accurately represent the semantic content of retrieving image, be language in a kind of reduction retrieval
Effective trial of adopted wide gap.
Beneficial effects of the present invention are mainly manifested in：The method combination saliency partitioning algorithm, by a picture in its entirety
The characteristics of image of foreground target and background, fusion foreground target and background is partitioned into, is completed to retrieve by means of bag of words.We
The more traditional bag of words searching algorithm of method can more accurately represent the semantic content of retrieving image, be that one kind can be reduced effectively
The search method of semantic gap.
Brief description of the drawings
Fig. 1 is the flow chart of the CBIR method based on conspicuousness segmentation.
Specific embodiment
The invention will be further described below in conjunction with the accompanying drawings.
Reference picture 1, a kind of CBIR method based on conspicuousness segmentation, the picture in the implementation case
It is divided into 100 classes, there are 100 per class picture.Random 20 pictures of selecting for training, use by remaining 80 pictures in every class picture
In test.Training image is trained using the CBIR method split based on conspicuousness and to test
Image is retrieved, and its structural framing is as shown in figure 1, concrete operation step includes training process and test process.
The training process includes step one to step 4, specially：
Step one, set up visual vocabulary dictionary.
Treatment is analyzed to every image that training image is concentrated successively, is that later retrieval sets up a visual vocabulary word
Allusion quotation.Set up concretely comprising the following steps for visual vocabulary dictionary：
Step 1.1：Color image into gray level image.
Step 1.2：The characteristic vector of every image that training image is concentrated is obtained on the gray level image that step 1.1 is obtained
Represent collection, represented with Fm, wherein F1, F2 ... Fm represents that the characteristic vector of an image represents collection respectively, and m=2000 is represented
Training image concentrates the number of image；The characteristic vector is 128 dimensions.
The method that the character representation for obtaining every image that training image is concentrated integrates is SIFT (Scale
Invariant Feature Transform) algorithm.
Step 1.3：Clustered and obtained each feature for the characteristic vector that the character representation that step 1.2 is obtained is concentrated
The representative feature vector of classification.Obtain 30 classifications, 30 representative feature vectors.
The clustering method is K-means clustering algorithms.
The representative feature vector of each feature classification is the barycenter of whole feature vector clusters in this feature classification.
Step 1.4：Visual vocabulary dictionary needed for obtaining later retrieval.The visual vocabulary dictionary is by step 1.4 gained
Each feature classification representative feature vector composition.
Step 2, the vision significance feature using image, segmentation obtain the foreground target figure and Background regional image of image.
Concrete operation step for single image is：
Step 2.1：Calculate the Saliency maps of image.The computational methods of Saliency maps are used based on the straight of regional correlation
Square figure comparative approach.
Step 2.2：The Saliency maps obtained using step 2.1, iteration operation image partitioning algorithm, respectively obtain image
Foreground target figure and Background regional image.Image segmentation algorithm uses GrabCut.
Step 3, the color character and textural characteristics that extract image respectively to foreground target figure and Background regional image.For
The concrete operation step of single image is：
Step 3.1：The color space of foreground target figure and Background regional image is transformed into HSV color spaces.
Step 3.2：Color character is extracted in the H passages and channel S of foreground target figure.Extracting method is using calculating H values
With the algorithm of the statistic histogram of S values.H channel histograms have 64 dimensions, and channel S histogram has 16 dimensions.
Step 3.3：Color character is extracted in the H passages and channel S of Background regional image.Extracting method is using calculating H values
With the algorithm of the statistic histogram of S values.H channel histograms have 64 dimensions, and channel S histogram has 16 dimensions.
Step 3.4：The texture feature extraction on the V passages of foreground target figure.Extracting method uses local binary patterns
(Local Binary Pattern) algorithm.The textural characteristics have 256 dimensions.
Step 3.5 texture feature extraction on the V passages of Background regional image.Extracting method uses local binary patterns
(Local Binary Pattern) algorithm.The textural characteristics have 256 dimensions.
Step 4, on the basis of the operation of step one, in staqtistical data base per pictures visual vocabulary distribution histogram.
For single image, the concrete operation step for setting up character representation is：
Step 4.1：The foreground target figure and Background regional image that step 2 is obtained are converted into gray level image.
Step 4.2：The visual vocabulary dictionary that step 1 is obtained is combined in the gray level image that step 4.1 is obtained, is counted respectively
The statistic histogram that visual vocabulary occurs in the middle visual vocabulary dictionary of foreground target figure and Background regional image.
The test process includes step 5.
Step 5, step one to step 4 operation on the basis of, test image is retrieved.
Step 5.1：Foreground target figure and Background regional image are obtained to the segmentation of test image step 2 methods described.
Step 5.2：Test image foreground target figure and Background regional image are extracted with step 3 methods described to test image
Color character and textural characteristics.
Step 5.3：The visual word of foreground target figure and Background regional image is counted with step 4 methods described to test image
Remittance statistic histogram.
Step 5.4：Each feature that test image is obtained according to step 5.1, step 5.2 and step 5.3 successively with instruction
The character pair for practicing every image in image set carries out distance metric, and the distance for obtaining uses D respectivelyS,DHC, DSCAnd DLBPRepresent.
The method that distance metric is used is COS distance.
Step 5.5：Distance metric to every kind of feature of step 5.4 gained is weighted cumulative, obtains total distance, uses
D1,D2... DiRepresent, wherein i represents the amount of images of training image concentration, Di=DS+DHC+DSC+3*DLBP
Step 5.6：For every image that training image is concentrated, according to D1,D2... DiOrder from small to large is to training
Image sorts and exports.
By the operation of above-mentioned steps, you can realize the retrieval to test image.
Above-described specific descriptions, purpose, technical scheme and beneficial effect to inventing have been carried out further specifically
It is bright, should be understood that and the foregoing is only specific embodiment of the invention, for explaining the present invention, it is not used to limit this
The protection domain of invention, all any modification, equivalent substitution and improvements within the spirit and principles in the present invention, done etc., all should
It is included within protection scope of the present invention.
Claims (6)
1. it is a kind of based on conspicuousness segmentation CBIR method, it is characterised in that：Described image search method
Including training process and test process；
The training process is comprised the following steps：
Step one, set up visual vocabulary dictionary：Treatment is analyzed to every image that training image is concentrated successively, is follow-up inspection
Mono- visual vocabulary dictionary of Suo Jianli；
Step 2, the vision significance feature using image, segmentation obtain the foreground target figure and Background regional image of image；
Step 3, the color character and textural characteristics that extract image respectively to foreground target figure and Background regional image；
Step 4, on the basis of the operation of step one, in staqtistical data base per pictures visual vocabulary distribution histogram；
The test process includes step 5, on the basis of step one to step 4 operation, test image is retrieved, mistake
Journey is as follows：
Step 5.1：Foreground target figure and Background regional image are obtained to the segmentation of test image step 2 methods described；
Step 5.2：The color of test image foreground target figure and Background regional image is extracted with step 3 methods described to test image
Color feature and textural characteristics；
Step 5.3：Test image is united with the visual vocabulary that step 4 methods described counts foreground target figure and Background regional image
Meter histogram；
Step 5.4：Each feature that test image is obtained according to step 5.1, step 5.2 and step 5.3 successively with training figure
The character pair of every image in image set carries out distance metric；
Step 5.5：Distance metric to every kind of feature of step 5.4 gained adds up, and obtains total distance, uses D1,D2... Di
Represent, wherein i represents the amount of images of training image concentration；
Step 5.6：For every image that training image is concentrated, according to D1,D2... DiOrder from small to large is to training image
Sort and export.
It is 2. as claimed in claim 1 to be based on the CBIR method that conspicuousness is split, it is characterised in that：It is described
It is in step one, the step of set up visual vocabulary dictionary：
Step 1.1：Color image into gray level image；
Step 1.2：The characteristic vector that every image that training image is concentrated is obtained on the gray level image that step 1.1 is obtained is represented
Collection, represented with Fm, wherein F1, F2 ... Fm represents that the characteristic vector of an image represents collection respectively, and m represents training image collection
The number of middle image；The characteristic vector is multidimensional, and its dimension is represented with N, N>2；
The characteristic vector for obtaining every image that training image is concentrated represents that the method for integrating is SIFT algorithms；
Step 1.3：Clustered and obtained each feature classification for the characteristic vector that the character representation that step 1.2 is obtained is concentrated
Representative feature vector.K classification is obtained, K representative feature is vectorial, and K is positive integer；
The clustering method is K-means clustering algorithms；
The representative feature vector of each feature classification is the barycenter of whole feature vector clusters in this feature classification；
Step 1.4：Visual vocabulary dictionary needed for obtaining later retrieval, the visual vocabulary dictionary is every by step 1.4 gained
The representative feature vector composition of individual feature classification.
It is 3. as claimed in claim 1 or 2 to be based on the CBIR method that conspicuousness is split, it is characterised in that：
In the step 2, the cutting operation step for single image is：
Step 2.1：The Saliency maps of image are calculated, the computational methods of Saliency maps use the histogram based on regional correlation
Comparative approach；
Step 2.2：The Saliency maps obtained using step 2.1, iteration operation image partitioning algorithm, respectively obtain the prospect of image
Target figure and Background regional image, image segmentation algorithm use GrabCut.
It is 4. as claimed in claim 1 or 2 to be based on the CBIR method that conspicuousness is split, it is characterised in that：
In the step 3, the extraction operating procedure for single image is：
Step 3.1：The color space of foreground target figure and Background regional image is transformed into HSV color spaces；
Step 3.2 extracts color character in the H passages and channel S of foreground target figure, and extracting method is using calculating H values and S values
Statistic histogram algorithm；
Step 3.3 extracts color character in the H passages and channel S of Background regional image, and extracting method is using calculating H values and S values
Statistic histogram algorithm；
Step 3.4 texture feature extraction on the V passages of foreground target figure, extracting method uses local binary patterns algorithm；
Step 3.5 texture feature extraction on the V passages of Background regional image, extracting method uses local binary patterns algorithm.
It is 5. as claimed in claim 1 or 2 to be based on the CBIR method that conspicuousness is split, it is characterised in that：
In the step 4, for single image, the operating procedure for setting up character representation is：
Step 4.1：The foreground target figure and Background regional image that step 2 is obtained are converted into gray level image；
Step 4.2：The visual vocabulary dictionary that step 1 is obtained is combined in the gray level image that step 4.1 is obtained, prospect is counted respectively
The statistic histogram of the visual vocabulary of target figure and Background regional image.
It is 6. as claimed in claim 1 or 2 to be based on the CBIR method that conspicuousness is split, it is characterised in that：
In the step 5.4, the method that distance metric is used is COS distance.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710151860.9A CN106844785A (en) | 2017-03-15 | 2017-03-15 | Saliency segmentation-based content-based image retrieval method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710151860.9A CN106844785A (en) | 2017-03-15 | 2017-03-15 | Saliency segmentation-based content-based image retrieval method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106844785A true CN106844785A (en) | 2017-06-13 |
Family
ID=59144602
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710151860.9A Pending CN106844785A (en) | 2017-03-15 | 2017-03-15 | Saliency segmentation-based content-based image retrieval method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106844785A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107357834A (en) * | 2017-06-22 | 2017-11-17 | 浙江工业大学 | Image retrieval method based on visual saliency fusion |
| CN108334627A (en) * | 2018-02-12 | 2018-07-27 | 北京百度网讯科技有限公司 | Searching method, device and the computer equipment of new media content |
| CN110276281A (en) * | 2019-06-10 | 2019-09-24 | 浙江工业大学 | A kind of screenshotss picture and text identification extracting method and system towards mobile terminal |
| CN111415357A (en) * | 2020-03-19 | 2020-07-14 | 长光卫星技术有限公司 | Portable shadow extraction method based on color image |
| CN115131589A (en) * | 2022-08-31 | 2022-09-30 | 天津艺点意创科技有限公司 | Image generation method for intelligent design of Internet literary works |
| CN117874278A (en) * | 2024-03-11 | 2024-04-12 | 盛视科技股份有限公司 | Image retrieval method and system based on multi-region feature combination |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020265A (en) * | 2012-12-25 | 2013-04-03 | 深圳先进技术研究院 | Image retrieval method and system |
| CN103064985A (en) * | 2013-01-28 | 2013-04-24 | 北京理工大学 | Priori knowledge based image retrieval method |
| CN105469096A (en) * | 2015-11-18 | 2016-04-06 | 南京大学 | Feature bag image retrieval method based on Hash binary code |
| CN106126572A (en) * | 2016-06-17 | 2016-11-16 | 中国科学院自动化研究所 | Image search method based on area validation |
-
2017
- 2017-03-15 CN CN201710151860.9A patent/CN106844785A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020265A (en) * | 2012-12-25 | 2013-04-03 | 深圳先进技术研究院 | Image retrieval method and system |
| CN103064985A (en) * | 2013-01-28 | 2013-04-24 | 北京理工大学 | Priori knowledge based image retrieval method |
| CN105469096A (en) * | 2015-11-18 | 2016-04-06 | 南京大学 | Feature bag image retrieval method based on Hash binary code |
| CN106126572A (en) * | 2016-06-17 | 2016-11-16 | 中国科学院自动化研究所 | Image search method based on area validation |
Non-Patent Citations (1)
| Title |
|---|
| 汪启伟: "图像直方图特征及其应用研究", 《中国博士学位论文全文数据库信息科技辑》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107357834A (en) * | 2017-06-22 | 2017-11-17 | 浙江工业大学 | Image retrieval method based on visual saliency fusion |
| CN108334627A (en) * | 2018-02-12 | 2018-07-27 | 北京百度网讯科技有限公司 | Searching method, device and the computer equipment of new media content |
| CN110276281A (en) * | 2019-06-10 | 2019-09-24 | 浙江工业大学 | A kind of screenshotss picture and text identification extracting method and system towards mobile terminal |
| CN111415357A (en) * | 2020-03-19 | 2020-07-14 | 长光卫星技术有限公司 | Portable shadow extraction method based on color image |
| CN111415357B (en) * | 2020-03-19 | 2023-04-07 | 长光卫星技术股份有限公司 | Portable shadow extraction method based on color image |
| CN115131589A (en) * | 2022-08-31 | 2022-09-30 | 天津艺点意创科技有限公司 | Image generation method for intelligent design of Internet literary works |
| CN115131589B (en) * | 2022-08-31 | 2022-11-22 | 天津艺点意创科技有限公司 | Image generation method for intelligent design of Internet literary works |
| CN117874278A (en) * | 2024-03-11 | 2024-04-12 | 盛视科技股份有限公司 | Image retrieval method and system based on multi-region feature combination |
| CN117874278B (en) * | 2024-03-11 | 2024-05-28 | 盛视科技股份有限公司 | Image retrieval method and system based on multi-region feature combination |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106844785A (en) | Saliency segmentation-based content-based image retrieval method | |
| CN104199931B (en) | A kind of consistent semantic extracting method of trademark image and trade-mark searching method | |
| CN107357834A (en) | Image retrieval method based on visual saliency fusion | |
| CN109409384A (en) | Image-recognizing method, device, medium and equipment based on fine granularity image | |
| CN107103326A (en) | The collaboration conspicuousness detection method clustered based on super-pixel | |
| CN106021603A (en) | Garment image retrieval method based on segmentation and feature matching | |
| Wang et al. | Improving human action recognition by non-action classification | |
| CN103279738B (en) | Automatic identification method and system for vehicle logo | |
| Wang et al. | Cattle face recognition method based on parameter transfer and deep learning | |
| Tian et al. | Natural scene text detection with MC–MR candidate extraction and coarse-to-fine filtering | |
| CN107330027A (en) | A kind of Weakly supervised depth station caption detection method | |
| CN105678244B (en) | A kind of near video search method based on improved edit-distance | |
| CN113112498B (en) | Grape leaf spot identification method based on fine-grained countermeasure generation network | |
| Wang et al. | S 3 D: Scalable pedestrian detection via score scale surface discrimination | |
| CN104156690A (en) | Gesture recognition method based on image space pyramid bag of features | |
| CN111507413A (en) | City management case image recognition method based on dictionary learning | |
| CN103985130A (en) | Image significance analysis method for complex texture images | |
| CN108268875A (en) | A kind of image meaning automatic marking method and device based on data smoothing | |
| CN114842507A (en) | Reinforced pedestrian attribute identification method based on group optimization reward | |
| Xu et al. | A robust hierarchical detection method for scene text based on convolutional neural networks | |
| CN108805183A (en) | A kind of image classification method of fusion partial polymerization descriptor and local uniform enconding | |
| CN109145140A (en) | One kind being based on the matched image search method of hand-drawn outline figure and system | |
| CN106066887B (en) | A kind of sequence of advertisements image quick-searching and analysis method | |
| CN104537392A (en) | Object detection method based on distinguishing semantic component learning | |
| CN110363164A (en) | Unified method based on LSTM time consistency video analysis |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170613 |