CN108536769A - Image analysis method, searching method and device, computer installation and storage medium - Google Patents
Image analysis method, searching method and device, computer installation and storage medium Download PDFInfo
- Publication number
- CN108536769A CN108536769A CN201810241452.7A CN201810241452A CN108536769A CN 108536769 A CN108536769 A CN 108536769A CN 201810241452 A CN201810241452 A CN 201810241452A CN 108536769 A CN108536769 A CN 108536769A
- Authority
- CN
- China
- Prior art keywords
- cluster
- image
- vector
- high dimension
- vectorial
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of image analysis method, searching method and device, computer installation and storage mediums, are related to picture search technical field, wherein described image analysis method includes：It treats analysis picture and carries out feature extraction；Neural network is established according to the feature extracted；Convolutional calculation is carried out to the neural unit of the neural network and obtains handling result；The handling result is exported to obtain corresponding high dimension vector by corresponding full articulamentum；Merge the high dimension vector in pre-determined distance as vectorial cluster.According to scheme provided by the present invention, it can be achieved that more preferably picture analyzing and picture searching effect.
Description
Technical field
The present invention relates to picture search technical field more particularly to image analysis method, searching method and device, computers
Device and storage medium.
Background technology
Picture search is one of big event of big data analysis, and the accuracy and efficiency of search are the research in market instantly
Hot spot.
When carrying out picture search, comparing search and the conventional search for carrying out precise image content according to picture material has
Institute is different, emphasis not in the specified target in searching for picture material, but in global level using similarity as foundation
It goes to obtain search result.
One of prior art has by directory index（It is mostly tree structure）Mode is filed, and is closed to image content
Connection classification.Each subclass is all the upper level based on it（Lower level-one）Subclass according to customized feature carry out further it is thin
Section is sorted out, until the image file of tree structure bottom connection.However, 1. need manual sort when filing, manpower and when
Between cost it is higher；2. manual sort is error-prone；3. building, library structure is complicated and flexibility is low；4. some minutias are held when filing
Easily it is ignored；5. the problems such as search result can not be ranked up by comparison degree.
The two of the prior art have the high dimensional feature vector for all images to be compared of extracting with neural network, using pre- place
Reason（Constituent analysis is extracted in a patent-pending as above）Afterwards, vector to be compared is subjected to distance operation one by one with object vector,
Similarity is obtained after final ranking operation.Demand much cannot be satisfied for the data search speed of mass data.
Invention content
The present invention for existing image analysis and corresponding picture search technology there are the problem of, provide image analysis
Method, searching method and device, computer installation and storage medium, to realize more preferably picture analyzing and picture searching effect.
The technical solution that the present invention is proposed with regard to above-mentioned technical problem is as follows：
In a first aspect, the present invention provides a kind of image analysis method, the method includes：
It treats analysis picture and carries out feature extraction；
Neural network is established according to the feature extracted；
Convolutional calculation is carried out to the neural unit of the neural network and obtains handling result；
The handling result is exported to obtain corresponding high dimension vector by corresponding full articulamentum；
Merge the high dimension vector in pre-determined distance as vectorial cluster.
According to above-mentioned image analysis method, feature that the basis is extracted establishes neural network and is：
According to the feature extracted according to AlexNet model foundation neural networks.
According to above-mentioned image analysis method, the method further includes：
The vectorial cluster is limited into line range with predetermined manner so that each high dimension vector is close to the center of the vector cluster
High dimension vector.
Second aspect, the present invention provide a kind of image search method, the method includes：
Target Photo is analyzed using image analysis method as described above to obtain target high dimension vector；
Vectorial cluster corresponding with the target high dimension vector is determined from the vectorial cluster of default number of clusters using index；
Obtain the image corresponding to the vectorial cluster within the scope of the pre-determined distance of the corresponding vectorial cluster of the target high dimension vector.
According to the image search method of preceding claim, the method further includes：
Each picture to be analyzed is analyzed using image analysis method as described above to obtain corresponding high dimensional feature vector；
Clustering processing is carried out to each high dimensional feature vector and obtains the vectorial cluster of the default number of clusters；
The corresponding index is created according to the vectorial cluster of the default number of clusters.
The third aspect, the present invention also provides a kind of image analysis apparatus, including：
Characteristic extracting module carries out feature extraction for treating analysis picture；
Neural network module, for establishing neural network according to the feature extracted；
Computing module carries out convolutional calculation for the neural unit to the neural network and obtains handling result；
Vectorial output module, for exporting to obtain corresponding high dimension vector the handling result by corresponding full articulamentum；
Vectorial cluster generation module, for merging the high dimension vector in pre-determined distance as vectorial cluster.
Fourth aspect, the present invention also provides a kind of image search apparatus, including：
Analysis module obtains target high dimensional feature for being analyzed target image using image analysis apparatus as described above
Vector；
Determining module, for determining vector corresponding with the target high dimension vector from the vectorial cluster of default number of clusters using index
Cluster；
Acquisition module, for obtaining the vectorial cluster pair within the scope of the pre-determined distance of the corresponding vectorial cluster of the target high dimension vector
The image answered.
According to above-mentioned image search apparatus, the analysis module is also with for utilizing image analysis apparatus as described above
Each image to be analyzed is analyzed to obtain corresponding high dimensional feature vector；
Described image searcher further includes：
Cluster module obtains the vectorial cluster of the default number of clusters for carrying out clustering processing to each high dimensional feature vector；
Index creation module, for creating the corresponding index according to the vectorial cluster of the default number of clusters.
5th aspect, the present invention also provides a kind of computer installation, the computer installation includes processor, the processing
Device is for realizing image analysis method as described above and figure as described above when executing the computer program stored in memory
As the step in searching method.
6th aspect, the present invention also provides a kind of storage mediums, are stored thereon with computer program, the computer program
The step in image analysis method as described above and image search method as described above is realized when being executed by processor.
The advantageous effect that technical solution provided in an embodiment of the present invention is brought is：
Image analysis method provided by the present invention is being treated the progress feature extraction of analysis picture and is being built according to the feature extracted
Vertical neural network, and convolutional calculation is carried out to the neural unit of neural network and obtains processing structure, thereafter, by the handling result
It exports to obtain corresponding high dimension vector by corresponding full articulamentum, finally, merges the high dimension vector in pre-determined distance as vector
Cluster, the vector cluster be include multiple high dimension vectors with corresponding to similar and/or approximation characteristic, thus using it is described to
Amount cluster obtains the degree of association of the two at a distance from the high dimension vector corresponding to each feature in Target Photo, is thus conducive to image
Between the degree of association analysis, contribute to fast implementing for picture search.
Image search method provided by the present invention, using image analysis method above-mentioned to Target Photo handled with
Target high dimension vector corresponding with the feature in the Target Photo is obtained, thereafter, utilizes the vector indexed from default number of clusters
Determine that vectorial cluster corresponding with the target high dimension vector is obtained in the corresponding vector of the target high dimension vector later in cluster
The image corresponding to vectorial cluster within the scope of the pre-determined distance of cluster, to realize search picture associated with Target Photo content
Function.Relatively existing picture searching technology can reduce the complexity for building the associated image library of image content, and can
The step of removing manual sort, can largely meet the capture to the minutia of image content, search speed is fast, can
Meet technical grade picture searching requirement.
Description of the drawings
To describe the technical solutions in the embodiments of the present invention more clearly, make required in being described below to embodiment
Attached drawing is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for
For those of ordinary skill in the art, without creative efforts, other are can also be obtained according to these attached drawings
Attached drawing.
Fig. 1 is flow chart schematic diagram of the image analysis method provided by the invention under an embodiment；
Fig. 2 is flow chart schematic diagram of the image search method provided by the invention under an embodiment；
Fig. 3 is the structural schematic diagram of image analysis apparatus provided by the invention；
Fig. 4 is the structural schematic diagram of image search apparatus provided by the invention.
Specific implementation mode
To make the object, technical solutions and advantages of the present invention clearer, below in conjunction with attached drawing to embodiment party of the present invention
Formula is described in further detail.
Fig. 1 is flow chart schematic diagram of the image analysis method provided by the invention under an embodiment, see Fig. 1 institutes
Show, described image analysis method includes the following steps：
Step s101：It treats analysis picture and carries out feature extraction.Wherein, the feature extraction is to be different from having in image
One class object of the corresponding feature of other class objects or the set of object extract.
Certainly, specific characteristic is extracted using conventional images identification technology to obtain accordingly to carry out classification precipitation to image
The classification of multi-layer and its subclassification, the matrix conversion that may be based on pixel value carry out multi-C vector calculating and conversion with to image
It carries out classification precipitation and obtains classification and its subclassification of corresponding multi-layer.
The picture to be analyzed can be the picture of one and the above quantity.
Step s102：Neural network is established according to the feature extracted.Wherein, the neural unit of the neural network can be
Convolutional Neural unit, and each neural unit can with it is above-mentioned it is corresponding classify it is corresponding.
The image analysis method that present embodiment is provided can be according to the feature extracted according to AlexNet model foundation nerves
Network.
The feature that the basis is extracted establishes neural network：
According to the feature extracted according to AlexNet model foundation neural networks, the neural network of the AlexNet model foundations can
Including five convolutional layers set gradually and three full articulamentums, five convolutional layers can be denoted as the first convolutional layer, volume Two respectively
Lamination, third convolutional layer, Volume Four lamination and the 5th convolutional layer, and three full articulamentums can be denoted as respectively the first full articulamentum,
Second full articulamentum and the full articulamentum of third.
Step s103：Convolutional calculation is carried out to the neural unit of the neural network and obtains handling result.
Wherein, the part of the image in each neural unit and the part of the image with standard feature can be carried out one by one
The convolutional layer that contrast conting obtains corresponding level is handling result.
Step s104：The handling result is exported to obtain corresponding high dimension vector by corresponding full articulamentum.
Wherein, the high dimension vector can be 4096 dimensional vectors, and each high dimension vector and the feature phase in picture to be analyzed
It is corresponding.
Step s105：Merge the high dimension vector in pre-determined distance as vectorial cluster.
Wherein, the vectorial cluster is to merge multiple obtained aggregates of data of high dimension vector, and combined foundation is each high
The distance between dimensional vector can be by corresponding high dimension vector only when the distance between each high dimension vector meets pre-determined distance
It is incorporated into an aggregate of data.
The distance in pre-determined distance in present embodiment is the COS distance between two high dimension vectors.
Certainly, the number that combined high dimension vector can be also determined according to user demand, as determined and merged when user demand
When the number of high dimension vector is 10 or less, 10 high dimension vectors below for meeting pre-determined distance are only merged into a vector
Cluster.And during screening meets the high dimension vector of demand number, it can be according to the preferential of the feature corresponding to each high dimension vector
Grade setting is screened, and includes on earth as the priority of feature is arranged by height：When shape, aberration, color, then during screening,
Preferential selection meets high dimension vector corresponding with shape feature, and selection of taking second place meets the corresponding high dimension vector of aberration feature,
Finally selection meets high dimension vector corresponding with color characteristic, and by the control of the quantity of the high dimension vector of each cluster 10 with
Under.
The center high dimension vector of vectorial cluster in present embodiment can voluntarily be selected by user, also but default choice waits for point
Analyse the high dimension vector pointed by most characteristic feature in picture.
The image analysis method that present embodiment is provided carries out feature extraction and according to extracting treating analysis picture
Feature establish neural network, and convolutional calculation is carried out to the neural unit of neural network and obtains processing structure, thereafter, will be described
Handling result is exported to obtain corresponding high dimension vector by corresponding full articulamentum, finally, merges the high dimension vector in pre-determined distance
As vectorial cluster, the vector cluster be include multiple high dimension vectors with corresponding to similar and/or approximation characteristic, thus can profit
The degree of association of the two is obtained at a distance from the high dimension vector corresponding to each feature in Target Photo with the vectorial cluster, thus
Conducive to the analysis of the degree of association between image, contribute to fast implementing for picture search.
In addition, after merging the high dimension vector in pre-determined distance as vectorial cluster, can with predetermined manner to it is described to
Amount cluster limited into line range so that each high dimension vector close to the vectorial cluster center high dimension vector, specifically can will described in
Vectorial cluster is multiplied with a contraction factor, and the contraction factor can be with constant, and value can be 0.618.
It should be understood that when high dimension vector is closer to the center high dimension vector of the vectorial cluster, you can think the higher-dimension
Vector is is distributed preferable point, and distance value too far can be for the farthest high dimension vector of the center high dimension vector of distance vector cluster extremely
The distance of the center high dimension vector of vectorial cluster and the nearest high dimension vector of the center high dimension vector of distance vector cluster to vectorial cluster
The half of the sum of the distance of center high dimension vector.
It should be understood that the use of contraction factor can reduce the influence that the noise in image forms vectorial cluster.
The present invention also provides a kind of image search method, Fig. 2 is image search method provided by the invention in an embodiment party
Flow chart schematic diagram under formula, please as shown in Figure 2, image search method provided by the invention includes the following steps：
Step s201：Feature extraction is carried out to Target Photo.Wherein, special using analysis picture progress is treated in aforementioned step
The method for levying extraction carries out feature extraction to Target Photo.
Step s202：Neural network is established according to the feature extracted.
Step s203：Convolutional calculation is carried out to the neural unit of the neural network and obtains handling result.
Step s204：The handling result is exported by corresponding full articulamentum to obtain corresponding target high dimension vector.
Step s205：Vector corresponding with the target high dimension vector is determined from the vectorial cluster of default number of clusters using index
Cluster.
Wherein, the index can create by the following method, i.e.,：
Analysis picture is treated using above-mentioned image analysis method to be analyzed to obtain corresponding high dimensional feature vector；
Merge the high dimension vector in pre-determined distance as vectorial cluster.
It should be understood that the corresponding picture to be analyzed of multiple, after carrying out respective handling, obtained vector cluster
It mutually should be the vectorial cluster of multiple number of clusters.
The corresponding index is created according to the vectorial cluster of the default number of clusters, i.e., is selected from obtained multiple vectorial clusters
The vectorial cluster of default number of clusters is using as index.
Step s206：Obtain the vectorial cluster pair within the scope of the pre-determined distance of the corresponding vectorial cluster of the target high dimension vector
The image answered.
It should be understood that the size of the pre-determined distance range of the corresponding vectorial cluster of the target high dimension vector determines target
The size of the image content degree of association of the picture picture pointed with search, apart from the corresponding vectorial cluster of the target high dimension vector
Vectorial cluster it is closer, the content degree of association of Target Photo and the pointed picture of search is bigger, otherwise smaller.And herein default
Distance range then indicates that the degree of association of the image content and Target Photo of the pointed picture of user controllable system search, realization have needle
To the image of the search specified conditions of property.
The image search method that present embodiment is provided, using image analysis method above-mentioned to Target Photo at
Reason is to obtain target high dimension vector corresponding with the feature in the Target Photo, thereafter, using index from default number of clusters
Determine that vectorial cluster corresponding with the target high dimension vector obtains corresponding in the target high dimension vector later in vectorial cluster
The image corresponding to vectorial cluster within the scope of the pre-determined distance of vectorial cluster, to realize that search is associated with Target Photo content
The function of picture.
It should be understood that carrying out feature extraction and using related algorithm to picture to obtain high dimension vector, and merge pre-
If the high dimension vector in distance is as vectorial cluster.Similarly, multiple vectorial clusters can be similarly handled plurality of pictures and be obtained,
It can be indexed accordingly using obtained multiple vectorial clusters, to utilize the associated picture of Target Photo search pictures content
When, to Target Photo carry out respective handling obtain target high dimension vector when, according to the index can fast search scheme therewith
The associated picture of piece content.
Relatively existing picture searching technology can reduce the complexity for building the associated image library of image content, and energy
The step of enough removing manual sort, can largely meet the capture to the minutia of image content, and search speed is fast,
Technical grade picture searching requirement can be met.
The present invention also provides a kind of image analysis apparatus, Fig. 3 is the structural representation of image analysis apparatus provided by the invention
Figure, please as shown in Figure 3, image analysis apparatus 11 provided by the invention includes：
Characteristic extracting module 111 carries out feature extraction for treating analysis picture.
Neural network module 112, for establishing neural network according to the feature extracted.
Computing module 113 carries out convolutional calculation for the neural unit to the neural network and obtains handling result.
Vectorial output module 114, for exporting to obtain corresponding higher-dimension the handling result by corresponding full articulamentum
Vector.
Vectorial cluster generation module 115, for merging the high dimension vector in pre-determined distance as vectorial cluster.
It should be understood that after corresponding module executes corresponding function, achieved effect can be with image above-mentioned
Analysis method is identical, therefore details are not described herein again.
The present invention also provides a kind of image search apparatus, Fig. 4 is the structural representation of image search apparatus provided by the invention
Figure, please as shown in Figure 4, image search apparatus 21 provided by the invention includes：
Analysis module 211 obtains target higher-dimension for being analyzed target image using image analysis apparatus as described above
Feature vector.
Determining module 212, for being determined and the target high dimension vector pair from the vectorial cluster of default number of clusters using index
The vectorial cluster answered.
Acquisition module 213, for obtaining within the scope of the pre-determined distance of the corresponding vectorial cluster of the target high dimension vector
The corresponding image of vectorial cluster.
Wherein, the analysis module 211 is also with for utilizing image analysis apparatus as described above to each figure to be analyzed
As being analyzed to obtain corresponding high dimensional feature vector；
Cluster module 214 obtains the vectorial cluster of the default number of clusters for carrying out clustering processing to each high dimensional feature vector.
Index creation module 215, for creating the corresponding index according to the vectorial cluster of the default number of clusters.
It should be understood that after corresponding module executes corresponding function, achieved effect can be with image above-mentioned
Searching method is identical, therefore details are not described herein again.
In addition, the present invention provides a kind of computer installation, the computer installation includes processor, and the processor is used for
The step in above-mentioned image analysis method and image search method is realized when executing the computer program stored in memory.
The processor can be central processing unit (Central Processing Unit, CPU), can also be it
His general processor, digital signal processor (Digital Signal Processor, DSP), application-specific integrated circuit
(Application Specific Integrated Circuit, ASIC), ready-made programmable gate array (Field-
Programmable Gate Array, FPGA) either other programmable logic device, discrete gate or transistor logic device
Part, discrete hardware components etc..General processor can be microprocessor or the processor can also be any conventional processing
Device etc., the processor are the control centres of the computer installation, are filled using various interfaces and the entire computer of connection
The various pieces set.
The memory can be used for storing the computer program and/or module, and the processor is by running or executing
Computer program in the memory and/or module are stored, and calls the data being stored in memory, described in realization
The various functions of computer installation.The memory can include mainly storing program area and storage data field, wherein storage program
It area can storage program area, the application program etc. needed at least one function；Storage data field can store the use according to mobile phone
The data etc. created.In addition, memory may include high-speed random access memory, can also include non-volatile memories
Device, such as hard disk, memory, plug-in type hard disk, intelligent memory card（Smart Media Card, SMC）, secure digital（Secure
Digital, SD）Card, flash card（Flash Card）, at least one disk memory, flush memory device or other volatibility
Solid-state memory.
In addition, the present invention also provides a kind of storage medium, it is stored thereon with computer program, the computer program is located
Reason device realizes the step in image analysis method above-mentioned and image search method above-mentioned when executing.
The foregoing is merely presently preferred embodiments of the present invention, is not intended to limit the invention, it is all the present invention spirit and
Within principle, any modification, equivalent replacement, improvement and so on should all be included in the protection scope of the present invention.
Claims (10)
1. a kind of image analysis method, which is characterized in that the method includes：
It treats analysis picture and carries out feature extraction；
Neural network is established according to the feature extracted；
Convolutional calculation is carried out to the neural unit of the neural network and obtains handling result；
The handling result is exported to obtain corresponding high dimension vector by corresponding full articulamentum；
Merge the high dimension vector in pre-determined distance as vectorial cluster.
2. image analysis method according to claim 1, which is characterized in that the feature that the basis is extracted establishes nerve
Network is：
According to the feature extracted according to AlexNet model foundation neural networks.
3. image analysis method according to claim 2, which is characterized in that the method further includes：
The vectorial cluster is limited into line range with predetermined manner so that each high dimension vector is close to the center of the vector cluster
High dimension vector.
4. a kind of image search method, which is characterized in that the method includes：
Target Photo is analyzed using the image analysis method as described in claims 1 to 3 any one to obtain target height
Dimensional vector；
Vectorial cluster corresponding with the target high dimension vector is determined from the vectorial cluster of default number of clusters using index；
Obtain the image corresponding to the vectorial cluster within the scope of the pre-determined distance of the corresponding vectorial cluster of the target high dimension vector.
5. image search method according to claim 4, which is characterized in that the method further includes：
Each picture to be analyzed is analyzed to obtain using the image analysis method as described in claims 1 to 3 any one
Corresponding high dimensional feature vector；
Clustering processing is carried out to each high dimensional feature vector and obtains the vectorial cluster of the default number of clusters；
The corresponding index is created according to the vectorial cluster of the default number of clusters.
6. a kind of image analysis apparatus, which is characterized in that including：
Characteristic extracting module carries out feature extraction for treating analysis picture；
Neural network module, for establishing neural network according to the feature extracted；
Computing module carries out convolutional calculation for the neural unit to the neural network and obtains handling result；
Vectorial output module, for exporting to obtain corresponding high dimension vector the handling result by corresponding full articulamentum；
Vectorial cluster generation module, for merging the high dimension vector in pre-determined distance as vectorial cluster.
7. a kind of image search apparatus, which is characterized in that including：
Analysis module obtains target for being analyzed target image using image analysis apparatus as claimed in claim 6
High dimensional feature vector；
Determining module, for determining vector corresponding with the target high dimension vector from the vectorial cluster of default number of clusters using index
Cluster；
Acquisition module, for obtaining the vectorial cluster pair within the scope of the pre-determined distance of the corresponding vectorial cluster of the target high dimension vector
The image answered.
8. image search apparatus according to claim 7, which is characterized in that the analysis module is also with for using as weighed
Profit requires the image analysis apparatus described in 6 to be analyzed each image to be analyzed to obtain corresponding high dimensional feature vector；
Described image searcher further includes：
Cluster module obtains the vectorial cluster of the default number of clusters for carrying out clustering processing to each high dimensional feature vector；
Index creation module, for creating the corresponding index according to the vectorial cluster of the default number of clusters.
9. a kind of computer installation, which is characterized in that the computer installation includes processor, and the processor is deposited for executing
The image analysis method and such as right as described in any one of claim 1-3 are realized when the computer program stored in reservoir
It is required that the step in image search method described in 4-5 any one.
10. a kind of storage medium, is stored thereon with computer program, which is characterized in that the computer program is held by processor
Image analysis method as described in any one of claim 1-3 is realized when row and as described in claim 4-5 any one
Image search method in step.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810241452.7A CN108536769B (en) | 2018-03-22 | 2018-03-22 | Image analysis method, search method and device, computer device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810241452.7A CN108536769B (en) | 2018-03-22 | 2018-03-22 | Image analysis method, search method and device, computer device and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108536769A true CN108536769A (en) | 2018-09-14 |
| CN108536769B CN108536769B (en) | 2023-01-03 |
Family
ID=63483730
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810241452.7A Active CN108536769B (en) | 2018-03-22 | 2018-03-22 | Image analysis method, search method and device, computer device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108536769B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112052866A (en) * | 2019-06-05 | 2020-12-08 | 株式会社理光 | Image verification method and device and computer readable storage medium |
| CN112115281A (en) * | 2020-09-17 | 2020-12-22 | 杭州海康威视数字技术股份有限公司 | Data retrieval method, device and storage medium |
Citations (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030074368A1 (en) * | 1999-01-26 | 2003-04-17 | Hinrich Schuetze | System and method for quantitatively representing data objects in vector space |
| US20030110181A1 (en) * | 1999-01-26 | 2003-06-12 | Hinrich Schuetze | System and method for clustering data objects in a collection |
| CN102291577A (en) * | 2010-06-21 | 2011-12-21 | 北京中星微电子有限公司 | Method and device for calculating macroblock motion vector |
| CN102341824A (en) * | 2009-03-04 | 2012-02-01 | 公立大学法人大阪府立大学 | Method of compiling image database, image database compilation program, image retrieval method |
| CN103714354A (en) * | 2014-01-16 | 2014-04-09 | 西安电子科技大学 | Hyperspectral image wave band selection method based on quantum-behaved particle swarm optimization algorithm |
| CN105045841A (en) * | 2015-07-01 | 2015-11-11 | 北京理工大学 | Image feature query method in combination with gravity sensor and image feature point angles |
| CN105373583A (en) * | 2015-10-12 | 2016-03-02 | 国家计算机网络与信息安全管理中心 | Modeling method for support vector machine based on data compression |
| US20160103900A1 (en) * | 2014-10-08 | 2016-04-14 | University Of Lancaster | Data structuring and searching methods and apparatus |
| CN105631037A (en) * | 2015-12-31 | 2016-06-01 | 北京恒冠网络数据处理有限公司 | Image retrieval method |
| US20160267351A1 (en) * | 2013-07-08 | 2016-09-15 | University Of Surrey | Compact and robust signature for large scale visual search, retrieval and classification |
| CN106503656A (en) * | 2016-10-24 | 2017-03-15 | 厦门美图之家科技有限公司 | A kind of image classification method, device and computing device |
| CN106503106A (en) * | 2016-10-17 | 2017-03-15 | 北京工业大学 | A kind of image hash index construction method based on deep learning |
| CN106528662A (en) * | 2016-10-20 | 2017-03-22 | 中山大学 | Quick retrieval method and system of vehicle image on the basis of feature geometric constraint |
| CN107122809A (en) * | 2017-04-24 | 2017-09-01 | 北京工业大学 | Neural network characteristics learning method based on image own coding |
| CN107256246A (en) * | 2017-06-06 | 2017-10-17 | 西安工程大学 | PRINTED FABRIC image search method based on convolutional neural networks |
| CN107451200A (en) * | 2017-07-06 | 2017-12-08 | 西安交通大学 | Search method using Randomized Quantizing words tree and the image search method based on it |
| CN107515905A (en) * | 2017-08-02 | 2017-12-26 | 北京邮电大学 | A kind of interactive type image search and fusion method based on sketch |
| CN107563439A (en) * | 2017-08-31 | 2018-01-09 | 湖南麓川信息科技有限公司 | A kind of model for identifying cleaning food materials picture and identification food materials class method for distinguishing |
| CN107679078A (en) * | 2017-08-29 | 2018-02-09 | 银江股份有限公司 | A kind of bayonet socket image vehicle method for quickly retrieving and system based on deep learning |
| CN107729935A (en) * | 2017-10-12 | 2018-02-23 | 杭州贝购科技有限公司 | The recognition methods of similar pictures and device, server, storage medium |
-
2018
- 2018-03-22 CN CN201810241452.7A patent/CN108536769B/en active Active
Patent Citations (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030110181A1 (en) * | 1999-01-26 | 2003-06-12 | Hinrich Schuetze | System and method for clustering data objects in a collection |
| US20030074368A1 (en) * | 1999-01-26 | 2003-04-17 | Hinrich Schuetze | System and method for quantitatively representing data objects in vector space |
| CN102341824A (en) * | 2009-03-04 | 2012-02-01 | 公立大学法人大阪府立大学 | Method of compiling image database, image database compilation program, image retrieval method |
| CN102291577A (en) * | 2010-06-21 | 2011-12-21 | 北京中星微电子有限公司 | Method and device for calculating macroblock motion vector |
| US20160267351A1 (en) * | 2013-07-08 | 2016-09-15 | University Of Surrey | Compact and robust signature for large scale visual search, retrieval and classification |
| CN103714354A (en) * | 2014-01-16 | 2014-04-09 | 西安电子科技大学 | Hyperspectral image wave band selection method based on quantum-behaved particle swarm optimization algorithm |
| US20160103900A1 (en) * | 2014-10-08 | 2016-04-14 | University Of Lancaster | Data structuring and searching methods and apparatus |
| CN105045841A (en) * | 2015-07-01 | 2015-11-11 | 北京理工大学 | Image feature query method in combination with gravity sensor and image feature point angles |
| CN105373583A (en) * | 2015-10-12 | 2016-03-02 | 国家计算机网络与信息安全管理中心 | Modeling method for support vector machine based on data compression |
| CN105631037A (en) * | 2015-12-31 | 2016-06-01 | 北京恒冠网络数据处理有限公司 | Image retrieval method |
| CN106503106A (en) * | 2016-10-17 | 2017-03-15 | 北京工业大学 | A kind of image hash index construction method based on deep learning |
| CN106528662A (en) * | 2016-10-20 | 2017-03-22 | 中山大学 | Quick retrieval method and system of vehicle image on the basis of feature geometric constraint |
| CN106503656A (en) * | 2016-10-24 | 2017-03-15 | 厦门美图之家科技有限公司 | A kind of image classification method, device and computing device |
| CN107122809A (en) * | 2017-04-24 | 2017-09-01 | 北京工业大学 | Neural network characteristics learning method based on image own coding |
| CN107256246A (en) * | 2017-06-06 | 2017-10-17 | 西安工程大学 | PRINTED FABRIC image search method based on convolutional neural networks |
| CN107451200A (en) * | 2017-07-06 | 2017-12-08 | 西安交通大学 | Search method using Randomized Quantizing words tree and the image search method based on it |
| CN107515905A (en) * | 2017-08-02 | 2017-12-26 | 北京邮电大学 | A kind of interactive type image search and fusion method based on sketch |
| CN107679078A (en) * | 2017-08-29 | 2018-02-09 | 银江股份有限公司 | A kind of bayonet socket image vehicle method for quickly retrieving and system based on deep learning |
| CN107563439A (en) * | 2017-08-31 | 2018-01-09 | 湖南麓川信息科技有限公司 | A kind of model for identifying cleaning food materials picture and identification food materials class method for distinguishing |
| CN107729935A (en) * | 2017-10-12 | 2018-02-23 | 杭州贝购科技有限公司 | The recognition methods of similar pictures and device, server, storage medium |
Non-Patent Citations (2)
| Title |
|---|
| ANIRBAN MUKHOPADHYAY 等: ""Unsupervised Satellite Image Segmentation by Combining SA Based Fuzzy Clustering with Support Vector Machine"", 《2009 SEVENTH INTERNATIONAL CONFERENCE ON ADVANCES IN PATTERN RECOGNITION》 * |
| 章登义 等: ""基于分块特征收缩的行人检测方法"", 《计算机科学》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112052866A (en) * | 2019-06-05 | 2020-12-08 | 株式会社理光 | Image verification method and device and computer readable storage medium |
| CN112052866B (en) * | 2019-06-05 | 2024-08-06 | 株式会社理光 | Image verification method and device and computer readable storage medium |
| CN112115281A (en) * | 2020-09-17 | 2020-12-22 | 杭州海康威视数字技术股份有限公司 | Data retrieval method, device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108536769B (en) | 2023-01-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3044758B1 (en) | Automated and scalable object and feature extraction from imagery | |
| Paulin et al. | Local convolutional features with unsupervised training for image retrieval | |
| CN109376596A (en) | Face matching process, device, equipment and storage medium | |
| CN111382300B (en) | Multi-view three-dimensional model retrieval method and system based on pairing depth feature learning | |
| CN109948641A (en) | Anomaly groups recognition methods and device | |
| CN108280190A (en) | Image classification method, server and storage medium | |
| CN108536769A (en) | Image analysis method, searching method and device, computer installation and storage medium | |
| CN103353881A (en) | Method and device for searching application | |
| Yu et al. | A rule based technique for extraction of visual attention regions based on real-time clustering | |
| CN115239886A (en) | Remote sensing UAV-MVS image point cloud data processing method, device, equipment and medium | |
| Abdullah et al. | Spatial pyramids and two-layer stacking SVM classifiers for image categorization: A comparative study | |
| Shah et al. | Object detection using convolutional neural networks and transformer-based models: a review | |
| Pultar | Improving the hardnet descriptor | |
| CN106056090A (en) | Fingerprint image processing method and device | |
| CN109190637A (en) | A kind of image characteristic extracting method | |
| WO2021041342A1 (en) | Semantic image retrieval for whole slide images | |
| US20230196831A1 (en) | Image Group Classifier in a User Device | |
| CN116912537A (en) | Communication tower type identification method, device, equipment and storage medium | |
| Mansoor Hussain et al. | RETRACTED ARTICLE: Content based image retrieval using bees algorithm and simulated annealing approach in medical big data applications | |
| Piao et al. | Vdetor: an effective and efficient neural network for vehicle detection in aerial image | |
| Kang et al. | A novel unsupervised approach for multilevel image clustering from unordered image collection | |
| CN113656625A (en) | Method and device for determining human body space domain and electronic equipment | |
| CN111914697A (en) | Multi-view target identification method based on view semantic information and sequence context information | |
| Zhang et al. | Delving into fully convolutional networks activations for visual recognition | |
| Xu et al. | Spatial--spectral FFPNet: Attention-Based Pyramid Network for Segmentation and Classification of Remote Sensing Images |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |