CN110175249A - A kind of search method and system of similar pictures - Google Patents
A kind of search method and system of similar pictures Download PDFInfo
- Publication number
- CN110175249A CN110175249A CN201910470653.9A CN201910470653A CN110175249A CN 110175249 A CN110175249 A CN 110175249A CN 201910470653 A CN201910470653 A CN 201910470653A CN 110175249 A CN110175249 A CN 110175249A
- Authority
- CN
- China
- Prior art keywords
- picture
- feature
- library
- module
- pretreatment
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/51—Indexing; Data structures therefor; Storage structures
-
- 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
- G06F16/53—Querying
- G06F16/535—Filtering based on additional data, e.g. user or group profiles
-
- 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
- G06F16/55—Clustering; Classification
-
- 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
Abstract
The invention discloses a kind of search methods of similar pictures, this method comprises: acquisition original image, original image is put into searching database；Pretreatment is carried out to original image and generates pretreatment picture；Pretreatment picture is trained to obtain picture classification model, the feature of picture is pre-processed according to picture classification model extraction, goes forward side by side row vector to obtain feature vector；Described eigenvector is imported in the searching database, picture indices library is established；It is retrieved using the feature of the picture classification model extraction picture to be retrieved, and in the picture indices library to generate search result.The invention also discloses a kind of searching systems of similar pictures.The present invention carries out feature when treating retrieving image and retrieving, first with picture classification model extraction picture to be retrieved, improves the accuracy rate of this feature；In addition, further indexed using picture indices library to picture to be retrieved is retrieved, it is more efficient than directly doing full planting modes on sink characteristic match party.
Description
Technical field
The invention belongs to the technical field of image retrieval more particularly to the search methods and system of a kind of similar pictures.
Background technique
Since the 1970s, the research in relation to image retrieval has just started, and is mainly based upon the figure of text at that time
As retrieval technique (Text-based Image Retrieval, abbreviation TBIR), image is described in the way of text description
Feature, such as the author of paint, age, school, size.To after the nineties, there is the contents semantic to image, such as
The image retrieval technologies that color, texture, layout of image etc. are analyzed and retrieved, i.e. content-based image retrieval
(Content-based Image Retrieval, abbreviation CBIR) technology.
Existing similar pictures search method majority is by traditional image texture, shape, and the features such as edge carry out figure
Piece description.Such as:
Phash (perceptual hash), it describes the class for having comparable hash function, and characteristics of image be used to generate
Unique (but not being unique) fingerprint, and these fingerprints are comparable；
BOW (Bag of words) carries out bag of words by the sift corner feature to picture and summarizes, and obtained statistical is special
Sign.
It is huge by feature and the upper layer semantic information for itself being included but in coping with complicated picture scene
Gap is difficult to retrieve the image based on content using above-mentioned existing method, therefore search result is also not accurate enough, efficient.It takes
Obtain satisfactory effect.
Summary of the invention
In view of this, the embodiment of the present invention provides the search method and system of a kind of similar pictures, so as to accurate, high
Retrieve to effect the similar pictures in database.
In a first aspect, the embodiment of the invention provides a kind of search methods of similar pictures, this method comprises:
Original image is acquired, original image is put into searching database；
Pretreatment is carried out to original image and generates pretreatment picture；
Pretreatment picture is trained to obtain picture classification model, picture is pre-processed according to picture classification model extraction
Feature goes forward side by side row vector to obtain feature vector；
Described eigenvector is imported in the searching database, picture indices library is established；
It is retrieved using the feature of the picture classification model extraction picture to be retrieved, and in the picture indices library
To generate search result.
Second aspect, the embodiment of the invention provides a kind of searching system of similar pictures, the system include acquisition module,
Preprocessing module, training module, feature vector obtain module, index database establishes module and retrieval module；
Original image is put into searching database by the acquisition module for acquiring original image；
The preprocessing module is used to carry out original image pretreatment and generates pretreatment picture；
The training module is trained to obtain picture classification model to pretreatment picture, extracts the spy of pretreatment picture
Sign, goes forward side by side row vector to obtain feature vector；The index database establishes module and described eigenvector is imported the retrieval number
According in library and establishing picture indices library；
The retrieval module is using the feature of the picture classification model extraction picture to be retrieved and in the picture indices
It is retrieved in library to generate search result.
The present invention is carried out when treating retrieving image and retrieving first with picture classification model extraction picture to be retrieved
Feature improves the accuracy rate of this feature；In addition, when retrieving the feature of picture to be retrieved, due to treating the spy of retrieving image
Sign has carried out further index, and than directly doing full planting modes on sink characteristic match party efficiency, there has also been significantly improve.
Detailed description of the invention
Fig. 1 is a kind of flow diagram of the search method of similar pictures provided in an embodiment of the present invention；
Fig. 2 is a kind of structural schematic diagram of the searching system of similar pictures provided in an embodiment of the present invention.
Specific embodiment
The present invention is described in detail for each embodiment shown in reference to the accompanying drawing, but it should be stated that, these
Embodiment is not limitation of the present invention, those of ordinary skill in the art according to these embodiments made by function, method,
Or equivalent transformation or substitution in structure, all belong to the scope of protection of the present invention within.
Embodiment one
A kind of specific embodiment of the search method of similar pictures provided in an embodiment of the present invention is described below, referring to figure
1, which includes:
S100: original image is put into searching database by acquisition original image；
Specifically, can use video camera or acquire original image from network, and original image is put into searching database
In.
S200: pretreatment is carried out to original image and generates pretreatment picture；
Specifically, carrying out pretreatment to the original image in searching database generates pretreatment picture；
Wherein, described that original image pre-process including following sub-step S210-S220；
S210: fuzzy denoising is carried out to original image, the picture after being denoised, to exclude original image details to picture
The interference of retrieval；In the present embodiment, it can use Gaussian convolution and fuzzy denoising carried out to original image.
S220: being normalized the pixel of the picture after denoising, generates pretreatment picture, the specific following institute of formula
Show:
Wherein, Ix,yThat picture after denoising is set to x in place, the place y, channel for c pixel value；I′x,yIt is picture to be trained
Be set to x in place, the place y, channel for c pixel value.
S300: pretreatment picture is trained to obtain picture classification model, is pre-processed according to picture classification model extraction
The feature of picture goes forward side by side row vector to obtain feature vector；
The training in the present embodiment is described to obtain to pretreatment picture progress classification based training using residual error network
Picture classification model；It is specific as follows；
S310: in the present embodiment, can use inception residual error network and carry out classification based training to pretreatment picture, from
And obtain picture classification model.
Further, in a preferred embodiment, pretreatment picture can also be classified in advance, such as will schemed to training
Piece is divided into 20000 class pictures, and preferably to train picture, this 20000 class picture is the classification carried out according to the subjective thinking of people,
The subjective consciousness of people can be reflected according to the model of these sample trainings.
S320: the feature of picture classification model extraction pretreatment picture is utilized；
Specifically, for example can use the convolutional layer knot that picture classification model calculates mixed_conv (mixing convolutional layer) layer
Fruit, the convolutional layer result are the feature of the pretreatment picture extracted；
In the present embodiment, the characteristic of this layer is shown using mixed_conv layers of convolutional layer result during the test
More preferable according to effect, the result accuracy rate retrieved is higher；It should be noted that in other embodiments, picture can also be calculated
The convolutional layer result of other in disaggregated model layer.
S330: the feature for pre-processing picture is input in Fisher Vector (Fei Sheer vector coding) and obtains feature
Vector；
In the present embodiment, above-mentioned mixed_conv layers of convolution characteristic results can be input to Fisher Vector, most
Feature vector after being normalized eventually, such as the dimension of this feature vector is 2048；It is understood that in other embodiments
In, described eigenvector may be the feature vector of any dimension.
The step of obtaining feature vector using Fisher Vector includes: the mixed_conv layer for reading whole pictures
Convolution feature seeks the parameter of mixed Gauss model by EM algorithm using these convolution features；Read single figure
This feature is substituted into aforementioned mixed Gauss model, it is each to seek mixed Gauss model by the convolution feature of the mixed_conv layer of piece
The derivative of parameter is Fisher Vector characteristic vector, utilizes the standard of the Fisher Vector feature vector that can make to obtain
True rate is higher.
S400: described eigenvector is imported in the searching database, establishes picture indices library；Establish the picture indices
Library can accurately and efficiently retrieve similar pictures.
Specifically, the step S400 includes being clustered the feature vector in step S330 to form Clustering Model；
Preferably, in one embodiment, it can use kmean (k means clustering algorithm) clustering method to be clustered, i.e., will retrieve number
It is assigned at different cluster centres according to the feature of the pretreatment picture in library, and using the Clustering Model to pretreatment picture
Classify, i.e., is classified according to the feature vector of pretreatment picture to pretreatment picture.
Such as: pretreatment picture has 1,000,000,000 feature vectors, can first randomly select wherein 100w feature vector and carry out
Training obtains Clustering Model, and choosing the reason of wherein 100w feature vector is trained is, if to all 1,000,000,000 features
It is trained, by the calculation amount of the increase cluster training of geometry times, so that substantially reducing storage；
In storage, this 1,000,000,000 feature vector is referred to respectively away from nearest cluster centre, i.e., it will be sorted
It pre-processes picture to import in the searching database, establishes picture indices library；It, can be with it should be noted that when being retrieved
Only calculate the characteristic point that single cluster centre is included.
The step of Kmean clustering algorithm is to randomly select K object as initial cluster centre, and it is each right then to calculate
As each object being distributed to the cluster centre nearest apart from it the distance between with each seed cluster centre.Cluster centre
And it distributes to their object and just represents a cluster.One sample of every distribution, the cluster centre of cluster can be according in cluster
Existing object is recalculated.This process is repeated continuous until meeting some termination condition.Termination condition can be not
There is (or minimal amount) object to be reassigned to different clusters, there is no (or minimal amount) cluster centre to change again,
Error sum of squares Local Minimum.
Further, the step 400 further includes compression step, i.e., to the feature of the picture in the picture indices library to
Amount is compressed with the picture indices library after being optimized；Preferably, (PQ, PRODUCT can be quantified using product
QUANTIZATION) method compresses sorted picture, that is, utilizes using product quantization method to sorted picture
Feature vector compressed because the feature vector of sorted picture be higher-dimension floating point features vector, to the higher-dimension
Floating point features vector, which carries out compression, can reduce the occupied space of picture, to ensure that the demand of the present embodiment quick-searching.
It should be noted that can also first be compressed to pretreatment picture, then compressed picture is clustered,
To establish picture indices library.
S500: using the feature vector of the picture classification model extraction picture to be retrieved, and in the picture indices library
In retrieved to generate search result.
Wherein, step S500 includes following sub-step S510-S530, specific as follows；
S510: the feature of picture to be retrieved is compared with the feature of the picture in picture indices library；
Specifically, utilizing the feature of the picture classification model extraction picture to be retrieved；According to the feature of picture to be retrieved
Second feature vector is obtained, the second feature vector is clustered and compressed using Clustering Model, it will be compressed to be checked
Rope picture is compared with the picture in picture indices library；
S520: the COS distance between the picture feature in the feature and picture indices library of picture to be retrieved is calculated；
S530: the COS distance is ranked up according to sequence from small to large, and will be apart from the smallest predetermined quantity
A picture output is used as search result.
Embodiment two
A kind of specific embodiment of the searching system of similar pictures provided in an embodiment of the present invention is described below, referring to figure
2, which includes that module, index database foundation are obtained including acquisition module, preprocessing module, training module, feature vector
Module and retrieval module；
Original image is put into searching database by the acquisition module for acquiring original image；
The preprocessing module is used to carry out original image pretreatment and generates pretreatment picture；
The training module is trained to obtain picture classification model to pretreatment picture, extracts the spy of pretreatment picture
Sign, goes forward side by side row vector to obtain feature vector；The index database establishes module and described eigenvector is imported the retrieval number
According in library and establishing picture indices library；
The retrieval module is using the feature of the picture classification model extraction picture to be retrieved and in the picture indices
It is retrieved in library to generate search result.
Further, it includes cluster module and categorization module that the index database, which establishes module,；
The cluster module clusters to form Clustering Model described eigenvector；Preferably, it can use kmean
Clustering method clusters described eigenvector；
The categorization module using the Clustering Model carries out classification to pretreatment picture and by sorted pretreatment picture
It imports in the searching database to establish picture indices library.
Further, the retrieval module includes comparison module, distance acquisition module and sorting module；
The comparison module is for the feature of picture to be retrieved to be compared with the feature of the picture in picture indices library；
The distance acquisition module calculates remaining between the picture feature in the feature and picture indices library of picture to be retrieved
Chordal distance；
The sorting module is used to for the COS distance being ranked up according to sequence from small to large and will be apart from minimum
Predetermined quantity output be used as search result.
The retrieving of the present embodiment is consistent with the search method of embodiment one, and details are not described herein.
Beneficial effects of the present invention:
The present invention is carried out when treating retrieving image and retrieving first with picture classification model extraction picture to be retrieved
Feature improves the accuracy rate of this feature；In addition, when retrieving the feature of picture to be retrieved, due to treating the spy of retrieving image
Sign has carried out further index, and than directly doing full planting modes on sink characteristic match party efficiency, there has also been significantly improve.
Those of ordinary skill in the art may be aware that the embodiment in conjunction with disclosed in the embodiment of the present invention describe it is each
Exemplary unit and algorithm steps can be realized with the combination of electronic hardware or computer software and electronic hardware.These
Function is implemented in hardware or software actually, the specific application and design constraint depending on technical solution.Profession
Technical staff can use different methods to achieve the described function each specific application, but this realization is not answered
Think beyond the scope of this invention.
In embodiment provided herein, it should be understood that disclosed device and method can pass through others
Mode is realized.For example, the apparatus embodiments described above are merely exemplary, for example, the division of the unit, only
A kind of logical function partition, there may be another division manner in actual implementation, for example, multiple units or components can combine or
Person is desirably integrated into another system, or some features can be ignored or not executed.Another point, shown or discussed is mutual
Between coupling, direct-coupling or communication connection can be through some interfaces, the INDIRECT COUPLING or communication link of device or unit
It connects, can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.
It, can be with if the function is realized in the form of SFU software functional unit and when sold or used as an independent product
It is stored in a computer readable storage medium.Based on this understanding, technical solution of the present invention is substantially in other words
The part of the part that contributes to existing technology or the technical solution can be embodied in the form of software products, the meter
Calculation machine software product is stored in a storage medium, including some instructions are used so that a computer equipment (can be a
People's computer, server or network equipment etc.) it performs all or part of the steps of the method described in the various embodiments of the present invention.
And storage medium above-mentioned includes: that USB flash disk, mobile hard disk, ROM, RAM, magnetic or disk etc. are various can store program code
Medium.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can easily think of the change or the replacement, and should all contain
Lid is within protection scope of the present invention.Therefore, protection scope of the present invention should be subject to the protection scope in claims.
Claims (10)
1. a kind of search method of similar pictures, which is characterized in that this method comprises:
Original image is acquired, original image is put into searching database；
Pretreatment is carried out to original image and generates pretreatment picture；
Pretreatment picture is trained to obtain picture classification model, the spy of picture is pre-processed according to picture classification model extraction
Sign, goes forward side by side row vector to obtain feature vector；
Described eigenvector is imported in the searching database, picture indices library is established；
It is retrieved using the feature of the picture classification model extraction picture to be retrieved, and in the picture indices library with life
At search result.
2. search method according to claim 1, which is characterized in that the pretreatment includes:
Fuzzy denoising is carried out to original image, the picture after being denoised；
The pixel of picture after denoising is normalized, pretreatment picture is generated.
3. search method according to claim 1, which is characterized in that the training is using residual error network to pretreatment figure
Piece carries out classification based training to obtain the picture classification model.
4. search method according to claim 1, which is characterized in that the picture indices library of establishing includes:
Described eigenvector is clustered to form Clustering Model；
Classified using the Clustering Model to pretreatment picture；
Sorted pretreatment picture is imported in the searching database, picture indices library is established.
5. search method according to claim 1, which is characterized in that the picture indices library of establishing further includes compression step
Suddenly, i.e., the picture in the picture indices library is compressed with the picture indices library after being optimized.
6. search method according to claim 5, which is characterized in that using product quantization method to the picture indices library
In picture compressed.
7. search method according to claim 1-6, which is characterized in that it is described in the picture indices library into
Row retrieval includes following sub-step to generate search result:
The feature of picture to be retrieved is compared with the feature of the picture in picture indices library；
Calculate the COS distance between the picture feature in the feature and picture indices library of picture to be retrieved；
The COS distance is ranked up according to sequence from small to large, and will be exported apart from the smallest predetermined quantity picture
As search result.
8. a kind of searching system of similar pictures, which is characterized in that the system includes acquisition module, preprocessing module, training mould
Block, feature vector obtain module, index database establishes module and retrieval module；
Original image is put into searching database by the acquisition module for acquiring original image；
The preprocessing module is used to carry out original image pretreatment and generates pretreatment picture；
The training module is trained to obtain picture classification model to pretreatment picture, extracts the feature of pretreatment picture, and
Vectorization is carried out to obtain feature vector；The index database establishes module and imports described eigenvector in the searching database
And establish picture indices library；
The retrieval module is using the feature of the picture classification model extraction picture to be retrieved and in the picture indices library
It is retrieved to generate search result.
9. searching system according to claim 8, which is characterized in that the index database establish module include cluster module and
Categorization module；
The cluster module clusters to form Clustering Model described eigenvector；
The categorization module carries out classification to pretreatment picture using the Clustering Model and imports sorted pretreatment picture
To establish picture indices library in the searching database.
10. searching system according to claim 8, which is characterized in that the retrieval module includes that comparison module, distance obtain
Modulus block and sorting module；
The comparison module is for the feature of picture to be retrieved to be compared with the feature of the picture in picture indices library；
It is described distance obtain module calculate the cosine between the picture feature in the feature and picture indices library of picture to be retrieved away from
From；
The sorting module is used to for the COS distance being ranked up according to sequence from small to large and will be apart from the smallest pre-
The output of fixed number amount is used as search result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910470653.9A CN110175249A (en) | 2019-05-31 | 2019-05-31 | A kind of search method and system of similar pictures |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910470653.9A CN110175249A (en) | 2019-05-31 | 2019-05-31 | A kind of search method and system of similar pictures |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110175249A true CN110175249A (en) | 2019-08-27 |
Family
ID=67696222
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910470653.9A Pending CN110175249A (en) | 2019-05-31 | 2019-05-31 | A kind of search method and system of similar pictures |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110175249A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110609916A (en) * | 2019-09-25 | 2019-12-24 | 四川东方网力科技有限公司 | Video image data retrieval method, device, equipment and storage medium |
| CN111125418A (en) * | 2020-01-15 | 2020-05-08 | 广东工业大学 | Trademark retrieval system |
| CN111324760A (en) * | 2020-02-19 | 2020-06-23 | 名创优品（横琴）企业管理有限公司 | Image retrieval method and device |
| CN111506761A (en) * | 2020-04-22 | 2020-08-07 | 上海极链网络科技有限公司 | Similar picture query method, device, system and storage medium |
| CN112015937A (en) * | 2020-08-31 | 2020-12-01 | 核工业北京地质研究院 | Picture geographical positioning method and system |
| CN112287142A (en) * | 2020-11-18 | 2021-01-29 | 上海依图网络科技有限公司 | Picture retrieval method and device, electronic equipment and computer-readable storage medium |
| CN113282781A (en) * | 2021-05-18 | 2021-08-20 | 稿定（厦门）科技有限公司 | Image retrieval method and device |
| CN113362543A (en) * | 2021-05-21 | 2021-09-07 | 北京百度网讯科技有限公司 | Settlement method, settlement device, electronic device and storage medium |
| US11663260B2 (en) | 2020-04-08 | 2023-05-30 | Baidu Online Network Technology (Beijing) Co., Ltd. | Method and apparatus for searching multimedia content device, and storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040249774A1 (en) * | 1993-09-20 | 2004-12-09 | Caid William R. | Representation and retrieval of images using context vectors derived from image information elements |
| CN102521366A (en) * | 2011-12-16 | 2012-06-27 | 华中科技大学 | Image retrieval method integrating classification with hash partitioning and image retrieval system utilizing same |
| CN103399870A (en) * | 2013-07-08 | 2013-11-20 | 华中科技大学 | Visual word bag feature weighting method and system based on classification drive |
| CN104008395A (en) * | 2014-05-20 | 2014-08-27 | 中国科学技术大学 | Intelligent bad video detection method based on face retrieval |
| CN104376003A (en) * | 2013-08-13 | 2015-02-25 | 深圳市腾讯计算机系统有限公司 | Video retrieval method and device |
| CN106649490A (en) * | 2016-10-08 | 2017-05-10 | 中国人民解放军理工大学 | Depth feature-based image retrieval method and apparatus |
| CN107193979A (en) * | 2017-05-26 | 2017-09-22 | 成都掌中全景信息技术有限公司 | A kind of method of homologous picture retrieval |
-
2019
- 2019-05-31 CN CN201910470653.9A patent/CN110175249A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040249774A1 (en) * | 1993-09-20 | 2004-12-09 | Caid William R. | Representation and retrieval of images using context vectors derived from image information elements |
| CN102521366A (en) * | 2011-12-16 | 2012-06-27 | 华中科技大学 | Image retrieval method integrating classification with hash partitioning and image retrieval system utilizing same |
| CN103399870A (en) * | 2013-07-08 | 2013-11-20 | 华中科技大学 | Visual word bag feature weighting method and system based on classification drive |
| CN104376003A (en) * | 2013-08-13 | 2015-02-25 | 深圳市腾讯计算机系统有限公司 | Video retrieval method and device |
| CN104008395A (en) * | 2014-05-20 | 2014-08-27 | 中国科学技术大学 | Intelligent bad video detection method based on face retrieval |
| CN106649490A (en) * | 2016-10-08 | 2017-05-10 | 中国人民解放军理工大学 | Depth feature-based image retrieval method and apparatus |
| CN107193979A (en) * | 2017-05-26 | 2017-09-22 | 成都掌中全景信息技术有限公司 | A kind of method of homologous picture retrieval |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110609916A (en) * | 2019-09-25 | 2019-12-24 | 四川东方网力科技有限公司 | Video image data retrieval method, device, equipment and storage medium |
| CN111125418A (en) * | 2020-01-15 | 2020-05-08 | 广东工业大学 | Trademark retrieval system |
| CN111324760A (en) * | 2020-02-19 | 2020-06-23 | 名创优品（横琴）企业管理有限公司 | Image retrieval method and device |
| CN111324760B (en) * | 2020-02-19 | 2023-09-26 | 创优数字科技(广东)有限公司 | Image retrieval method and device |
| US11663260B2 (en) | 2020-04-08 | 2023-05-30 | Baidu Online Network Technology (Beijing) Co., Ltd. | Method and apparatus for searching multimedia content device, and storage medium |
| CN111506761A (en) * | 2020-04-22 | 2020-08-07 | 上海极链网络科技有限公司 | Similar picture query method, device, system and storage medium |
| CN112015937A (en) * | 2020-08-31 | 2020-12-01 | 核工业北京地质研究院 | Picture geographical positioning method and system |
| CN112015937B (en) * | 2020-08-31 | 2024-01-19 | 核工业北京地质研究院 | Picture geographic positioning method and system |
| CN112287142A (en) * | 2020-11-18 | 2021-01-29 | 上海依图网络科技有限公司 | Picture retrieval method and device, electronic equipment and computer-readable storage medium |
| CN113282781A (en) * | 2021-05-18 | 2021-08-20 | 稿定（厦门）科技有限公司 | Image retrieval method and device |
| CN113362543A (en) * | 2021-05-21 | 2021-09-07 | 北京百度网讯科技有限公司 | Settlement method, settlement device, electronic device and storage medium |
| CN113362543B (en) * | 2021-05-21 | 2023-10-31 | 北京百度网讯科技有限公司 | Settlement method, settlement device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110175249A (en) | A kind of search method and system of similar pictures | |
| CN109117777A (en) | The method and apparatus for generating information | |
| CN109325148A (en) | The method and apparatus for generating information | |
| US11429809B2 (en) | Image processing method, image processing device, and storage medium | |
| CN111914937A (en) | Lightweight improved target detection method and detection system | |
| Kadam et al. | Detection and localization of multiple image splicing using MobileNet V1 | |
| CN110209859A (en) | The method and apparatus and electronic equipment of place identification and its model training | |
| Lewis et al. | An integrated content and metadata based retrieval system for art | |
| Giveki et al. | A new content based image retrieval model based on wavelet transform | |
| CN102156686B (en) | Method for detecting specific contained semantics of video based on grouped multi-instance learning model | |
| CN111782852B (en) | Deep learning-based high-level semantic image retrieval method | |
| CN110825902B (en) | Method and device for realizing feature similarity search, electronic equipment and storage medium | |
| Lu et al. | Combining context, consistency, and diversity cues for interactive image categorization | |
| CN105389588A (en) | Multi-semantic-codebook-based image feature representation method | |
| CN108764258B (en) | Optimal image set selection method for group image insertion | |
| CN114077682B (en) | Intelligent recognition matching processing method and system for image retrieval and storage medium | |
| CN110162654A (en) | It is a kind of that image retrieval algorithm is surveyed based on fusion feature and showing for search result optimization | |
| Chen et al. | Photo retrieval based on spatial layout with hardware acceleration for mobile devices | |
| CN114692715A (en) | Sample labeling method and device | |
| Sasithradevi et al. | Content based video retrieval via object based approach | |
| Varish et al. | Color Image Retrieval Method Using Low Dimensional Salient Visual Feature Descriptors for IoT Applications | |
| CN111797765A (en) | Image processing method, image processing apparatus, server, and storage medium | |
| Al-Jubouri | Multi Evidence Fusion Scheme for Content-Based Image Retrieval by Clustering Localised Colour and Texture | |
| CN111768214A (en) | Product attribute prediction method, system, device and storage medium | |
| CN110163222A (en) | A kind of method of image recognition, the method for model training and server |
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: 20190827 |