CN105930499B - A kind of image searching method and system - Google Patents
A kind of image searching method and system Download PDFInfo
- Publication number
- CN105930499B CN105930499B CN201610300183.8A CN201610300183A CN105930499B CN 105930499 B CN105930499 B CN 105930499B CN 201610300183 A CN201610300183 A CN 201610300183A CN 105930499 B CN105930499 B CN 105930499B
- Authority
- CN
- China
- Prior art keywords
- picture
- candidate
- distance
- query sample
- reading
- 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
- 238000000034 method Methods 0.000 title claims abstract description 29
- 239000013598 vector Substances 0.000 claims abstract description 87
- 239000000284 extract Substances 0.000 claims abstract description 6
- 238000004422 calculation algorithm Methods 0.000 claims description 14
- 230000011218 segmentation Effects 0.000 claims description 9
- 238000000605 extraction Methods 0.000 claims description 6
- 238000007619 statistical method Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 abstract description 3
- 230000008569 process Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 2
- 235000013399 edible fruits Nutrition 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- HUTDUHSNJYTCAR-UHFFFAOYSA-N ancymidol Chemical compound C1=CC(OC)=CC=C1C(O)(C=1C=NC=NC=1)C1CC1 HUTDUHSNJYTCAR-UHFFFAOYSA-N 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization 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
Landscapes
- Engineering & Computer Science (AREA)
- Library & Information Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Image Analysis (AREA)
Abstract
The present invention is applicable in field of computer technology, provide a kind of image searching method and system, this method comprises: receiving the picture query sample that user submits, extract the feature vector of the picture query sample, described eigenvector is retrieved in index structure, to obtain associated candidate mark, the approximate characteristic vector that corresponding picture candidate is identified with the candidate is searched in the caching histogram constructed in advance, calculate the feature vector and the distance between the approximate characteristic vector of the picture query sample, when the distance is less than the first preset threshold, the picture candidate is determined as search result and exports the candidate mark, when the distance is greater than the second preset threshold, abandon the associated candidate mark, to substantially increase picture retrieval efficiency, reducing simultaneously needs the caching of searching system It asks.
Description
Technical field
The invention belongs to field of computer technology more particularly to a kind of image searching method and systems.
Background technique
In the similarity search of high dimensional data (for example, picture searching, multimedia retrieval etc.), dimension disaster makes traditional
Tree index algorithm (such as: b+ tree) become very inefficient or even is degenerated to force search (linear scan), therefore, have
Researcher proposes that, using higher-dimension sensitive hash algorithm progress proximity search, this technology has obtained very extensive research.However,
Emphasis is all concentrated on optimization and generates candidate by either tree index algorithm or higher-dimension sensitive hash algorithm, the two
In the process of (candidate generation), the lookup final result (candidate from candidate is had ignored
Refinement expense).Since this process usually requires to execute multiple I/O (input/output) access, to read number
According to the calculating for carrying out final result, so that (such as such as, photographic search engine is based on the application of high dimensional data similarity search
The Multimedia retrieval system etc. of content) in user query need multiple I/O access that can just obtain final result, lead to recall precision
Lowly, the problem of overhead increases.
Summary of the invention
The embodiment of the present invention is designed to provide a kind of image searching method and system, it is intended to solve due to the prior art
A kind of a kind of effective image searching method can not be provided, inquiry in picture search process is caused to need multiple I/O access ability
The problem of obtaining final result, leading to low recall precision, overhead increase.
On the one hand, the present invention provides a kind of image searching method, the method includes the following steps:
The picture query sample that user submits is received, the feature vector of the picture query sample is extracted;
Described eigenvector is retrieved, in index structure to obtain associated candidate mark;
It is searched in the caching histogram constructed in advance and identifies the approximate special of corresponding picture candidate with the candidate
Levy vector;
Calculate the feature vector and the distance between the approximate characteristic vector of the picture query sample;
When the distance is less than the first preset threshold, the picture candidate is determined as described in search result and output
Candidate mark abandons the associated candidate mark when the distance is greater than the second preset threshold.
On the other hand, the present invention provides a kind of image searching systems, which is characterized in that the system comprises:
Vector extraction unit extracts the spy of the picture query sample for receiving the picture query sample of user's submission
Levy vector;
Mark acquiring unit, for retrieving described eigenvector in index structure, to obtain associated candidate mark;
Vector search unit, for searching figure corresponding with candidate mark in the caching histogram constructed in advance
The approximate characteristic vector of piece candidate;
Metrics calculation unit, between the feature vector and the approximate characteristic vector for calculating the picture query sample
Distance;And
Search result processing unit is used for when the distance is less than the first preset threshold, and the picture candidate is true
It is set to search result and exports the candidate mark and is abandoned described associated when the distance is greater than the second preset threshold
Candidate mark.
In embodiments of the present invention, when receiving the picture query sample of user's submission, figure is retrieved in index structure
The feature vector of piece query sample is searched and is waited in the caching histogram constructed in advance to obtain associated candidate mark
The person of choosing identifies the approximate characteristic vector of corresponding picture candidate, calculate picture query sample feature vector and approximation characteristic to
The distance between amount, when distance is less than the first preset threshold, is determined as search result for picture candidate and exports candidate
Mark abandons associated candidate mark, to reduce the I/O in retrieving when distance is greater than the second preset threshold
Access, substantially increases picture retrieval efficiency, while reducing the buffer size to searching system.
Detailed description of the invention
Fig. 1 is the implementation flow chart for the image searching method that the embodiment of the present invention one provides;
Fig. 2 is the implementation flow chart of image searching method provided by Embodiment 2 of the present invention;
Fig. 3 is the structural schematic diagram for the image searching system that the embodiment of the present invention three provides;And
Fig. 4 is the structural schematic diagram for the image searching system that the embodiment of the present invention four provides.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
Specific implementation of the invention is described in detail below in conjunction with specific embodiment:
Embodiment one:
The implementation process that Fig. 1 shows the image searching method of the offer of the embodiment of the present invention one is only shown for ease of description
Go out part related to the embodiment of the present invention, details are as follows:
In step s101, the picture query sample that user submits is received, the feature vector of picture query sample is extracted.
The embodiment of the present invention is suitable for picture or search engine or search system based on picture, for example, picture searching draws
It holds up or multimedia search engine.When user submits picture query sample to carry out picture query, first (i.e. to picture query sample
The picture that user submits) it handles, the feature vector of picture query sample is extracted, to be used for subsequent lookup.
In step s 102, the feature vector of extraction is retrieved, in index structure to obtain associated candidate mark.
In step s 103, picture candidate corresponding with candidate mark is searched in the caching histogram constructed in advance
Approximate characteristic vector.
In embodiments of the present invention, using the approximate characteristic vector of histogram storage picture, all cachings in histogram
The feature vector of picture indicates that characteristic value segmentation each in this way can be indicated by corresponding numerical value, the spy of picture using characteristic value segmentation
It levies vector and is segmented expression by multiple characteristic values, to obtain the approximate characteristic vector of picture, needed so as to greatly reduce in caching
The image data to be cached, under same buffer memory capacity, caching histogram can cache the characteristic of more pictures, reduce
I/O access in retrieving, substantially increases picture retrieval efficiency, while reducing the buffer size to searching system.
In step S104, the distance between feature vector and approximate characteristic vector of picture query sample are calculated.
In step s105, when distance is less than the first preset threshold, picture candidate is determined as search result and defeated
Candidate identifies out, when distance is greater than the second preset threshold, abandons associated candidate mark.
In embodiments of the present invention, picture corresponding with candidate mark is searched in the caching histogram constructed in advance to wait
The approximate characteristic vector of the person of choosing calculates the distance between feature vector and approximate characteristic vector of picture query sample, works as distance
When less than the first preset threshold, show that picture candidate is the search result that user expects, so that picture candidate is determined as
Search result simultaneously exports candidate mark, when distance is greater than the second preset threshold, associated candidate mark is abandoned, to mention
The response speed and retrieval accuracy of high user search.
Embodiment two:
Fig. 2 shows the implementation processes of image searching method provided by Embodiment 2 of the present invention, for ease of description, only show
Go out part related to the embodiment of the present invention, details are as follows:
In step s 201, the picture query sample that user submits is received, the feature vector of picture query sample is extracted.
The embodiment of the present invention is suitable for picture or search engine or search system based on picture, for example, picture searching draws
It holds up or multimedia search engine.When user submits picture query sample to carry out picture query, first (i.e. to picture query sample
The picture that user submits) it handles, the feature vector of picture query sample is extracted, to be used for subsequent lookup.
In step S202, the feature vector of extraction is retrieved in index structure, to obtain associated candidate mark.
In step S203, picture candidate corresponding with candidate mark is searched in the caching histogram constructed in advance
Approximate characteristic vector.
In embodiments of the present invention, using the approximate characteristic vector of histogram storage picture, all cachings in histogram
The feature vector of picture indicates that characteristic value segmentation each in this way can be indicated by corresponding numerical value, the spy of picture using characteristic value segmentation
It levies vector and is segmented expression by multiple characteristic values, to obtain the approximate characteristic vector of picture, needed so as to greatly reduce in caching
The image data to be cached, under same buffer memory capacity, caching histogram can cache the characteristic of more pictures, reduce
I/O access in retrieving, substantially increases picture retrieval efficiency, while reducing the buffer size to searching system.
Preferably, when building caches histogram, firstly, being obtained according to user's history search record in preset period of time
Taking in the preset period of time frequency of occurrence in search result is more than the picture set of preset times, extracts in picture set and owns
The feature vector of picture, it is for statistical analysis to all feature vectors, to obtain each characteristic value in all feature vectors
Frequency of occurrence, all characteristic values are segmented according to obtained frequency of occurrence, the characteristic value region of segmentation are deposited into slow
It deposits in the histogram constructed in advance, which is determined as to cache histogram, it is optimal data cached straight to obtain
Fang Tu, thus under same buffer memory capacity, so that Installed System Memory can cache the characteristic of more pictures, to reduce retrieval
I/O access in the process.
Specifically, in histogram, the corresponding value in the characteristic value region of each segmentation.
Preferably, when being segmented all characteristic values according to obtained frequency of occurrence, most using dynamic programming algorithm
Smallization target error function (that is: makes approximate distance and the difference of actual distance between picture feature vector minimum), will be all
Each characteristic value in feature vector is segmented, to improve the accuracy of segmentation, guarantees picture approximate characteristic vector and figure
The higher similarity of piece real features vector.
In step S204, the distance between feature vector and approximate characteristic vector of picture query sample are calculated.
In step S205, when distance is less than the first preset threshold, picture candidate is determined as search result and defeated
Candidate identifies out, when distance is greater than the second preset threshold, abandons associated candidate mark.
In embodiments of the present invention, picture corresponding with candidate mark is searched in the caching histogram constructed in advance to wait
The approximate characteristic vector of the person of choosing calculates the distance between feature vector and approximate characteristic vector of picture query sample, works as distance
When less than the first preset threshold, show that picture candidate is the search result that user expects, so that picture candidate is determined as
Search result simultaneously exports candidate mark, when distance is greater than the second preset threshold, associated candidate mark is abandoned, to mention
The response speed and retrieval accuracy of high user search.
In step S206, when distance is between the first preset threshold and the second pre- threshold value, multistep k arest neighbors is used
Algorithm reads picture candidate from hard disk, is determined and is read according to the distance between picture query sample and the picture candidate of reading
Whether the picture candidate taken is search result.
In embodiments of the present invention, when the distance between the feature vector of picture query sample and approximate characteristic vector are located at
When between the first preset threshold and the second pre- threshold value, illustrate that picture candidate cannot directly be confirmed as search result, it can not
It is determined as non-search result, at this point, multistep k nearest neighbor algorithm (Multi- can be used in order to further increase retrieval accuracy
Step k-nearest neighbor query processing) picture candidate is read from hard disk, according to picture query
The distance between sample and the picture candidate of reading determine whether the picture candidate read is search result.
In step S207, when not finding approximate characteristic vector in buffer structure, multistep k nearest neighbor algorithm is used
Candidate is read from hard disk and identifies corresponding picture candidate, according between picture query sample and the picture candidate of reading
Distance determine read picture candidate whether be search result.
In embodiments of the present invention, when not finding approximate characteristic vector in buffer structure, show that user wishes
The picture of lookup in the buffer, does not need to read candidate from hard disk at this time, to complete the retrieval of picture.
Specifically, in embodiments of the present invention, according between picture query sample and the picture candidate of reading away from
When whether the picture candidate from determining reading is search result, calculating picture query sample and the picture of the reading first is waited
The distance between the person of choosing, when the distance between picture query sample and the picture candidate of reading are less than third predetermined threshold value,
The picture candidate of reading is determined as search result.Wherein, third predetermined threshold value and the first preset threshold can be identical, can also
With difference, specifically can require to be configured according to retrieval accuracy.
Those of ordinary skill in the art will appreciate that implement the method for the above embodiments be can be with
Relevant hardware is instructed to complete by program, the program can be stored in a computer readable storage medium,
The storage medium, such as ROM/RAM, disk, CD.
Embodiment three:
The structure that Fig. 3 shows the image searching system of the offer of the embodiment of the present invention three illustrates only for ease of description
Part related to the embodiment of the present invention, including:
Vector extraction unit 31 extracts the feature of picture query sample for receiving the picture query sample of user's submission
Vector;
Mark acquiring unit 32, for the searching characteristic vector in index structure, to obtain associated candidate mark;
Vector search unit 33, for searching picture corresponding with candidate mark in the caching histogram constructed in advance
The approximate characteristic vector of candidate;
Metrics calculation unit 34, between the feature vector and approximate characteristic vector for calculating picture query sample away from
From;And
Search result processing unit 35, for when distance is less than the first preset threshold, picture candidate being determined as examining
Hitch fruit simultaneously exports candidate mark, when distance is greater than the second preset threshold, abandons associated candidate mark.
In embodiments of the present invention, each unit of image searching system can be realized by corresponding hardware or software unit, respectively
Unit can be independent soft and hardware unit, also can integrate as a soft and hardware unit, herein not to limit the present invention.
The specific embodiment of each unit can refer to the description of embodiment one, and details are not described herein.
Example IV:
The structure that Fig. 4 shows the image searching system of the offer of the embodiment of the present invention four illustrates only for ease of description
Part related to the embodiment of the present invention, including:
Picture set acquiring unit 41, when being preset for obtaining according to user's history search record in preset period of time
Between frequency of occurrence is more than the picture set of preset times in search result in the period;
Statistical analysis unit 42, for extracting the feature vector of all pictures in picture set, to all feature vectors
It is for statistical analysis, to obtain the frequency of occurrence of each characteristic value in all feature vectors;
Data cached determination unit 43 will be segmented for being segmented all characteristic values according to obtained frequency of occurrence
Characteristic value region be deposited into caching in the histogram that constructs in advance, which is determined as to cache histogram;
Vector extraction unit 44 extracts the feature of picture query sample for receiving the picture query sample of user's submission
Vector;
Mark acquiring unit 45, for the searching characteristic vector in index structure, to obtain associated candidate mark;
Vector search unit 46, for searching picture corresponding with candidate mark in the caching histogram constructed in advance
The approximate characteristic vector of candidate;
Metrics calculation unit 47, between the feature vector and approximate characteristic vector for calculating picture query sample away from
From;
Search result processing unit 48, for when distance is less than the first preset threshold, picture candidate being determined as examining
Hitch fruit simultaneously exports candidate mark, when distance is greater than the second preset threshold, abandons associated candidate mark;And
As a result determination unit 49, for using multistep k when distance is between the first preset threshold and the second pre- threshold value
Nearest neighbor algorithm reads picture candidate from hard disk, according to the distance between picture query sample and the picture candidate of reading
Determine whether the picture candidate read is search result.In addition, result determination unit 49 can be also used for not tying in caching
When finding approximate characteristic vector in structure, candidate is read from hard disk using multistep k nearest neighbor algorithm and identifies corresponding picture
Candidate, according to the distance between picture query sample and the picture candidate of reading determine read picture candidate whether be
Search result.
Specifically, as a result determination unit 49 can include:
Apart from computation subunit 491, for calculating the distance between the picture candidate of picture query sample and reading;With
And
As a result subelement 492 is determined, for being less than when the distance between picture query sample and the picture candidate of reading
When third predetermined threshold value, the picture candidate of reading is determined as search result.
In embodiments of the present invention, each unit of image searching system can be realized by corresponding hardware or software unit, respectively
Unit can be independent soft and hardware unit, also can integrate as a soft and hardware unit, herein not to limit the present invention.
The specific embodiment of each unit can refer to the description of embodiment one, and details are not described herein.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Made any modifications, equivalent replacements, and improvements etc., should all be included in the protection scope of the present invention within mind and principle.
Claims (8)
1. a kind of image searching method, which is characterized in that the method includes the following steps:
The picture query sample that user submits is received, the feature vector of the picture query sample is extracted;
Described eigenvector is retrieved, in index structure to obtain associated candidate mark;
Searched in the caching histogram constructed in advance with the candidate identify the approximation characteristic of corresponding picture candidate to
Amount;
Calculate the feature vector and the distance between the approximate characteristic vector of the picture query sample;
When the distance is less than the first preset threshold, the picture candidate is determined as search result and exports the candidate
Person's mark abandons the associated candidate mark when the distance is greater than the second preset threshold;
Searched in the caching histogram constructed in advance with the candidate identify the approximation characteristic of corresponding picture candidate to
Before the step of amount, the method also includes:
According to user's history search record in preset period of time, obtains and go out occurrence in the preset period of time in search result
Number is more than the picture set of preset times;
The feature vector for extracting all pictures in the picture set, it is for statistical analysis to all feature vectors, with
Obtain the frequency of occurrence of each characteristic value in all feature vectors;
All characteristic values are segmented according to the obtained frequency of occurrence, the characteristic value region of segmentation is deposited into caching
In the histogram constructed in advance, which is determined as to cache histogram.
2. the method as described in claim 1, which is characterized in that the method also includes:
When the distance is between first preset threshold and the second pre- threshold value, using multistep k nearest neighbor algorithm from hard
The picture candidate is read on disk, it is true according to the distance between the picture query sample and the picture candidate of the reading
Whether the picture candidate of the fixed reading is search result.
3. the method as described in claim 1, which is characterized in that the method also includes:
When not finding the approximate characteristic vector in the buffer structure, using multistep k nearest neighbor algorithm from hard disk
It reads the candidate and identifies corresponding picture candidate, according to the picture candidate of the picture query sample and the reading
The distance between determine whether the picture candidate of the reading is search result.
4. method as claimed in claim 2 or claim 3, which is characterized in that according to the figure of the picture query sample and the reading
The distance between piece candidate determines the step of whether the picture candidate of the reading is search result, comprising:
Calculate the distance between the picture candidate of the picture query sample and the reading;
When the distance between the picture query sample and the picture candidate of the reading are less than third predetermined threshold value, by institute
The picture candidate for stating reading is determined as search result.
5. a kind of image searching system, which is characterized in that the system comprises:
Picture set acquiring unit, for obtaining the preset time according to user's history search record in preset period of time
Frequency of occurrence is more than the picture set of preset times in search result in period;
Statistical analysis unit, for extracting the feature vector of all pictures in the picture set, to all features to
Measure it is for statistical analysis, to obtain the frequency of occurrence of each characteristic value in all feature vectors;
Data cached determination unit, all characteristic values are segmented by the frequency of occurrence for obtaining according to, by segmentation
Characteristic value region is deposited into the histogram constructed in advance in caching, which is determined as to cache histogram;
Vector extraction unit, for receiving the picture query sample of user's submission, extract the feature of the picture query sample to
Amount;
Mark acquiring unit, for retrieving described eigenvector in index structure, to obtain associated candidate mark;
Vector search unit is waited for searching picture corresponding with candidate mark in the caching histogram constructed in advance
The approximate characteristic vector of the person of choosing;
Metrics calculation unit, between the feature vector and the approximate characteristic vector for calculating the picture query sample away from
From;And
Search result processing unit, for when the distance is less than the first preset threshold, the picture candidate to be determined as
Search result simultaneously exports the candidate mark, when the distance is greater than the second preset threshold, the discarding associated candidate
Person's mark.
6. system as claimed in claim 5, which is characterized in that the system also includes:
As a result determination unit is used for when the distance is between first preset threshold and the second pre- threshold value, using more
Step k nearest neighbor algorithm reads the picture candidate from hard disk, according to the picture of the picture query sample and the reading
The distance between candidate determines whether the picture candidate of the reading is search result.
7. system as claimed in claim 5, which is characterized in that the result determination unit is also used to not tie in the caching
When finding the approximate characteristic vector in structure, the candidate mark pair is read from hard disk using multistep k nearest neighbor algorithm
The picture candidate answered determines the reading according to the distance between the picture query sample and the picture candidate of the reading
Whether the picture candidate taken is search result.
8. system as claimed in claims 6 or 7, which is characterized in that the result determination unit includes:
Apart from computation subunit, for calculating the distance between the picture candidate of the picture query sample and the reading;
And
As a result subelement is determined, for being less than when the distance between the picture query sample and the picture candidate of the reading
When third predetermined threshold value, the picture candidate of the reading is determined as search result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610300183.8A CN105930499B (en) | 2016-05-09 | 2016-05-09 | A kind of image searching method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610300183.8A CN105930499B (en) | 2016-05-09 | 2016-05-09 | A kind of image searching method and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105930499A CN105930499A (en) | 2016-09-07 |
CN105930499B true CN105930499B (en) | 2019-11-22 |
Family
ID=56835280
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610300183.8A Active CN105930499B (en) | 2016-05-09 | 2016-05-09 | A kind of image searching method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105930499B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108287833A (en) * | 2017-01-09 | 2018-07-17 | 北京艺鉴通科技有限公司 | It is a kind of for the art work identification to scheme to search drawing method |
CN108920550A (en) * | 2018-06-15 | 2018-11-30 | 广州视源电子科技股份有限公司 | file searching method and device |
CN111177450B (en) * | 2019-12-31 | 2021-03-23 | 视辰信息科技(上海)有限公司 | Image retrieval cloud identification method and system and computer readable storage medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102110122A (en) * | 2009-12-24 | 2011-06-29 | 阿里巴巴集团控股有限公司 | Method and device for establishing sample picture index table, method and device for filtering pictures and method and device for searching pictures |
CN103336776A (en) * | 2013-05-13 | 2013-10-02 | 云南瑞攀科技有限公司 | Image searching method based on image content |
CN103984776A (en) * | 2014-06-05 | 2014-08-13 | 北京奇虎科技有限公司 | Repeated image identification method and image search duplicate removal method and device |
CN105095215A (en) * | 2014-04-22 | 2015-11-25 | 富士通株式会社 | Information acquisition device, information acquisition method and server |
CN105404657A (en) * | 2015-11-04 | 2016-03-16 | 北京工业大学 | CEDD feature and PHOG feature based image retrieval method |
-
2016
- 2016-05-09 CN CN201610300183.8A patent/CN105930499B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102110122A (en) * | 2009-12-24 | 2011-06-29 | 阿里巴巴集团控股有限公司 | Method and device for establishing sample picture index table, method and device for filtering pictures and method and device for searching pictures |
CN103336776A (en) * | 2013-05-13 | 2013-10-02 | 云南瑞攀科技有限公司 | Image searching method based on image content |
CN105095215A (en) * | 2014-04-22 | 2015-11-25 | 富士通株式会社 | Information acquisition device, information acquisition method and server |
CN103984776A (en) * | 2014-06-05 | 2014-08-13 | 北京奇虎科技有限公司 | Repeated image identification method and image search duplicate removal method and device |
CN105404657A (en) * | 2015-11-04 | 2016-03-16 | 北京工业大学 | CEDD feature and PHOG feature based image retrieval method |
Also Published As
Publication number | Publication date |
---|---|
CN105930499A (en) | 2016-09-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11126654B1 (en) | Method and apparatus for multi-dimensional content search and video identification | |
WO2017012491A1 (en) | Similarity comparison method and apparatus for high-dimensional image features | |
CN107256262B (en) | Image retrieval method based on object detection | |
US20130254191A1 (en) | Systems and methods for mobile search using bag of hash bits and boundary reranking | |
US20220254095A1 (en) | Apparatus and method for searching for global minimum of point cloud registration error | |
CN105930499B (en) | A kind of image searching method and system | |
CN112926592B (en) | Trademark retrieval method and device based on improved Fast algorithm | |
CN109492537B (en) | Object identification method and device | |
JP5367488B2 (en) | Data classification apparatus and program | |
CN116108150A (en) | Intelligent question-answering method, device, system and electronic equipment | |
KR101350335B1 (en) | Content based image retrieval apparatus and method | |
CN110209895B (en) | Vector retrieval method, device and equipment | |
Yu et al. | Key frame extraction scheme based on sliding window and features | |
CN112580438A (en) | Point cloud identification method in three-dimensional scene | |
JP2017162230A (en) | Information processor, similar data search method and similar data search program | |
Peng et al. | KISS: knowing camera prototype system for recognizing and annotating places-of-interest | |
KR102054211B1 (en) | Method and system for video retrieval based on image queries | |
Cao et al. | A key frame selection algorithm based on sliding window and image features | |
JP4343038B2 (en) | Partial image search system, method and program | |
JPH08249467A (en) | Device and method for retrieving similar picture | |
CN114219978B (en) | Target multi-part association method and device, terminal and computer readable storage medium | |
CN112559170B (en) | Approximate matching method for cache data in edge computing environment | |
Chen et al. | Deck: Detecting events from web click-through data | |
Toyama et al. | A fast template matching using adaptive window skipping method considering position of reference template | |
Liu et al. | cutting Image Retrieval Technology based on LSH Improvement |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right |
Effective date of registration: 20200624 Address after: No. 1088 Nanshan District Xueyuan Road in Shenzhen city in Guangdong province 518000 Patentee after: Southern University of Science and Technology Address before: 518057 Guangdong city of Shenzhen province Nanshan District high tech Industrial Park South Hing a No. 18 309B Patentee before: SHENZHEN INFIDATA TECHNOLOGY Co.,Ltd. |
|
TR01 | Transfer of patent right |