CN110390030A - The storage method and device of pictorial information - Google Patents
The storage method and device of pictorial information Download PDFInfo
- Publication number
- CN110390030A CN110390030A CN201910572522.1A CN201910572522A CN110390030A CN 110390030 A CN110390030 A CN 110390030A CN 201910572522 A CN201910572522 A CN 201910572522A CN 110390030 A CN110390030 A CN 110390030A
- Authority
- CN
- China
- Prior art keywords
- information
- suffix
- domain
- image data
- various features
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 43
- 238000004590 computer program Methods 0.000 claims description 15
- 238000010276 construction Methods 0.000 claims description 10
- 229910002056 binary alloy Inorganic materials 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 abstract description 10
- 238000000605 extraction Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 239000012141 concentrate Substances 0.000 description 2
- 230000008602 contraction Effects 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000012217 deletion Methods 0.000 description 2
- 230000037430 deletion Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 235000004789 Rosa xanthina Nutrition 0.000 description 1
- 241000109329 Rosa xanthina Species 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000002360 explosive Substances 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000013332 literature search Methods 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000011160 research 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
- G06F16/51—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
- G06F16/58—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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Library & Information Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the present invention is suitable for information technology field, provides the storage method and device of a kind of pictorial information, which comprises obtain image data to be processed;Semantic analysis is made to the image data, obtains the semantic text for describing the image data;The various features information is converted to feature vector by the various features information for extracting the image data;The semantic text and the corresponding feature vector of the various features information are stored into multiple Attribute domains of suffix index to be built respectively;Suffix index is constructed based on the multiple Attribute domain, realizes the storage to the information of the image data.The present embodiment pass through by describe image content semantic text and and be used to indicate that the binary vector of picture feature to be stored into Suffix array clustering jointly, so as to be indexed building with unified technology, so that the index constructed can meet the searching method based on text or based on image content feature simultaneously.
Description
Technical field
The invention belongs to information technology fields, storage method, a kind of pictorial information more particularly to a kind of pictorial information
Storage device, a kind of server and a kind of computer readable storage medium.
Background technique
With the development of Internet application technology, picture becomes one of the main source for obtaining information.In internet mass
In information, explosive growth is also presented in the quantity of picture.How fast and accurate in numerous pictures required figure is searched
Piece is the hot spot of internet application field research.
Traditional image searching method is text based search, with reference to the key being often used in literature search
Word and search technology, this method need to carry out picture relevant text information the extraction and analysis of keyword, and establish keyword
Index.But this method is often because the relevant text information of picture differs greatly to image content and causes search result related
It spends lower.In order to solve the deficiency based on text search method, occur the method based on image content feature, the party again later
For method by the content characteristic of extraction picture, color, texture, shape, spatial relationship including picture itself etc. establish aspect indexing.
One master drawing need to be provided when retrieval, by the content characteristic of extraction and analysis master drawing, compare, return with the aspect indexing constructed
Return the similar picture on content characteristic.Although this method can carry out analysis comparison from the feature of picture itself, lack figure
Piece expression semantic information, can only retrieve " likeness in form " as a result, and the content characteristic of picture is indexed be built with compared with
High calculating requirement.
The traditional picture retrieval method of both the above has its respective limitation, for above two method, existing skill
It is matched that different index construct technologies must be respectively adopted in art, be not available unified technology and be indexed building, rope
It is lower to draw creation efficiency.
Summary of the invention
In view of this, the embodiment of the invention provides a kind of storage method of pictorial information and device, to solve existing skill
Need to construct index respectively for different picture retrieval methods in art, building process not only inefficiency, to calculating requirement
Higher problem.
The first aspect of the embodiment of the present invention provides a kind of storage method of pictorial information, comprising:
Obtain image data to be processed;
Semantic analysis is made to the image data, obtains the semantic text for describing the image data;
The various features information is converted to feature vector by the various features information for extracting the image data;
The semantic text and the corresponding feature vector of the various features information are stored to suffix to be built respectively
In multiple Attribute domains of index;
Suffix index is constructed based on the multiple Attribute domain, realizes the storage to the information of the image data.
The second aspect of the embodiment of the present invention provides a kind of storage device of pictorial information, comprising:
Image data obtains module, for obtaining image data to be processed;
Semantic text analysis module is obtained for making semantic analysis to the image data for describing the picture number
According to semantic text;
Characteristic information extracting module believes the various features for extracting the various features information of the image data
Breath is converted to feature vector;
Metadata storage module, for respectively by the semantic text and the corresponding feature vector of the various features information
It stores into multiple Attribute domains of suffix index to be built;
Suffix index constructs module, for constructing suffix index based on the multiple Attribute domain, realizes to the picture number
According to information storage.
The third aspect of the embodiment of the present invention provides a kind of server, including memory, processor and is stored in institute
The computer program that can be run in memory and on the processor is stated, the processor executes real when the computer program
Now as described in relation to the first aspect the storage method of pictorial information the step of.
The fourth aspect of the embodiment of the present invention provides a kind of computer readable storage medium, the computer-readable storage
Media storage has computer program, and pictorial information as described in relation to the first aspect is realized when the computer program is executed by processor
The step of storage method.
Compared with prior art, the embodiment of the present invention includes following advantages:
The embodiment of the present invention, by making semantic analysis to image data, is used after obtaining image data to be processed
In the semantic text of the description image data, and the various features information of image data is extracted, is converted into feature vector, from
And semantic text and the corresponding feature vector of various features information can be stored respectively multiple to suffix index to be built
In Attribute domain, and then suffix index is constructed based on multiple Attribute domains, realizes the storage to the information of image data.The present embodiment is logical
The semantic text of image content will be described and and is used to indicate that the binary vector of picture feature to be stored jointly to Suffix array clustering by crossing
In, so as to be indexed building with unified technology, the index constructed is met simultaneously based on text or base
In the searching method of image content feature.Also, the building speed by indexing in this present embodiment be it is linear, with traditional rope
Draw method comparison, constructs more efficient, advantage and become apparent from.
Detailed description of the invention
It to describe the technical solutions in the embodiments of the present invention more clearly, below will be to embodiment or description of the prior art
Needed in attached drawing be briefly described.It should be evident that the accompanying drawings in the following description is only of the invention some
Embodiment for those of ordinary skill in the art without any creative labor, can also be according to these
Attached drawing obtains other attached drawings.
Fig. 1 is a kind of step flow diagram of the storage method of pictorial information of one embodiment of the invention;
Fig. 2 is a kind of schematic diagram of the storage device of pictorial information of one embodiment of the invention;
Fig. 3 is a kind of schematic diagram of server of one embodiment of the invention.
Specific embodiment
In being described below, for illustration and not for limitation, the tool of such as particular system structure, technology etc is proposed
Body details, to understand thoroughly the embodiment of the present invention.It should be apparent, however, to those skilled in the art that there is no these specific thin
The present invention also may be implemented in the other embodiments of section.In other cases, it omits to well-known system, device, circuit
And the detailed description of method, in case unnecessary details interferes description of the invention.
Illustrate technical solution of the present invention below by specific embodiment.
Referring to Fig.1, a kind of step process signal of the storage method of pictorial information of one embodiment of the invention is shown
Figure, can specifically include following steps:
S101, image data to be processed is obtained;
It should be noted that image data to be processed is the various pictures for needing to store its information.This reality
The various information of picture can be stored by constructing suffix index by applying example, improve the creation efficiency and search efficiency of index.
S102, semantic analysis is made to the image data, obtains the semantic text for describing the image data;
In embodiments of the present invention, carrying out the result that semantic analysis obtains to picture is usually the language for describing the image content
Sentence, these sentences can be stored separately in a text form in an Attribute domain of suffix index, and the title of the Attribute domain can
With semantics.
Above-mentioned semantic text can be the description for picture itself, other than the things that picture includes, be more
Some descriptions to relationship between things.
In the concrete realization, semantic description sentence directly can be generated according to picture by the method for deep learning.Example
It such as, can be a kind of typical end-to-end mould with Image Caption Generator (image header generator) algorithm, the algorithm
Picture can be converted to vector table first with advantage of the convolutional neural networks CNN in terms of picture high-level characteristic extraction by type
Show, recycles Recognition with Recurrent Neural Network RNN that picture vector is converted to semantic description sentence.
S103, the various features information for extracting the image data, are converted to feature vector for the various features information;
In embodiments of the present invention, the various features information of picture may include local binary patterns feature LBP, direction ladder
Spend at least one of histogram feature HOG, scale invariant feature SIFT or a variety of.
Certainly, in order to more effectively describe pictorial information, the present embodiment includes above-mentioned LBP special with various features information simultaneously
It is introduced for sign, HOG feature, SIFT feature.
LBP (Local Binary Patterns) is characterized in a kind of effective texture description operator, can measure and extract
The texture information of image local.HOG (Histogram of Oriented Gradient) is characterized in one kind in computer vision
With the Feature Descriptor for being used to carry out object detection in image procossing, it is by calculating the gradient side with statistical picture regional area
Carry out constitutive characteristic to histogram.SIFT (Scale-invariant feature transform) feature can be by seeking a width
Characteristic point in figure and its description in relation to size and direction simultaneously carry out Image Feature Point Matching and obtain.
In the concrete realization, corresponding feature extraction algorithm can be respectively adopted and obtain features described above.For example, LBP feature
Extraction algorithm, HOG feature extraction algorithm and SIFT feature extraction algorithm.
It in embodiments of the present invention,, can be by each feature for the ease of storage after obtaining features described above information respectively
Information processing is at feature vector.Features described above vector can be stored respectively as metadata to entitled LBP, HOG and SIFT
In Attribute domain.
S104, the semantic text and the corresponding feature vector of the various features information stored to be built respectively
In multiple Attribute domains of suffix index;
In embodiments of the present invention, it stores by the feature vector being converted in step S103 to before Attribute domain, also
May determine that whether the corresponding feature vector of various features information is binary vector, if the corresponding feature of any feature information to
Amount is not binary vector, then the corresponding feature vector of this feature information can be converted to binary vector.For example, by feature
Vector [10,08,22] is converted to [01010,010000,10110].Then, then respectively with semantic text and various features information
Corresponding binary vector stores above-mentioned metadata into multiple Attribute domains of suffix index to be built as metadata.
Certainly, multiple Attribute domains of suffix index to be built can also include for storing Image ID, picture name, figure
The Attribute domain of the information such as piece date created, picture author, the present embodiment are not construed as limiting this.
As shown in Table 1, be the present embodiment a kind of Attribute domain example.It in this example, include picture I PID,
Picture name picturename, picture date created creationdate, picture author author, picture semantic text
Semantics, multiple Attribute domains such as image content feature LBP, HOG, SIFT.
Table one, Attribute domain example
PID | picturename | creationdate | author | semantics | LBP | HOG | SIFT |
Certainly, according to actual needs, other Attribute domains can also be increased for storing the other information of picture, the present embodiment
This is not construed as limiting.
In the Attribute domain shown in above-mentioned, corresponding metadata can be stored respectively.As shown in Table 2, two width figures are provided
Piece corresponds to the metadata example of each Attribute domain.
Table two, the metadata example of each Attribute domain
PID | picturename | creationdate | author | semantics |
0001 | animal | 2019-5-20 | Bob | A puppy is chasing a butterfly |
0002 | flower | 2019-5-21 | Alice | Blooming roses |
Two (Continued) of table
S105, suffix index is constructed based on the multiple Attribute domain, realizes the storage to the information of the image data.
In embodiments of the present invention, it includes the metadata that the domain particular content is corresponded to for recordable picture that each Attribute domain, which removes,
It outside, further include Suffix array clustering and domain information structure.
In general, metadata is the data object to be stored, based on character string forms.Suffix array clustering has recorded character string
The lexcographical order position of all suffix.And domain information structure can be used for obtaining metadata of the specified file in corresponding Attribute domain,
The specified file is the conditional information of picture to be retrieved.
In embodiments of the present invention, after getting the metadata of some Attribute domain, preset Suffix array clustering can be used
Construction algorithm constructs Suffix array clustering corresponding with metadata in each Attribute domain, and in the domain information knot for determining each Attribute domain
After structure, according to the metadata of each Attribute domain, Suffix array clustering and domain information structure, suffix index is constructed.
It is discussed in detail in order to make it easy to understand, making one to Suffix array clustering and domain information structure separately below.
X=X [0, n] is enabled to be defined in a character string to end up with # on total order character set ∑, wherein the lexicographic order of #
Occur once less than other any characters in ∑ and only in X.X [i, j] is enabled to indicate in character string X with x [i], headed by x [j],
One substring of trailing character, wherein x [i] indicates the character that position is i in character string X, then can claim substring x [i, n] and x [j,
N] it is two suffix that character string X starts in position i and j.
The Suffix array clustering SA [0, n] of character string X is an integer array, and wherein SA [i]=j is indicated with x [j] as initial character
Suffix all suffix in character string X lexicographic order be i-th bit.In the construction algorithm of many SA, ISA can be also used
(Inverse Suffix Array, anti-Suffix array clustering), hasISA [i]=j indicates that meaning is
The lexicographic order of character suffixes all suffix in X is jth position headed by character at the position i in character string X.
It gives character string X=mississippi# (as shown in Table 3), then its SA, ISA, orderly suffix such as four institute of table
Show.
Table three, character string X
Table four, SA, ISA corresponding with three kinds of character strings of table, orderly suffix example
Related terms are defined as follows during Suffix array clustering creation:
1) L type suffix and L type character: if SA [i+1] < A [i], SA [i] they are L type suffix, L type suffix
Initial character is L type character.
2) S type suffix and S type character: if SA [i+1] > A [i], SA [i] they are S type suffix, S type suffix
Initial character is S type character.
3) LMS type suffix and LMS type character: if SA [i] is S type suffix, and SA [i-1] is L type suffix, then
SA [i] is LMS type suffix, and the initial character of LMS type suffix is LMS type character.
4) LMS type substring: for i < j, if SA [i] and SA [j] are LMS type, and do not have in SA [i+1, j-1]
Other LMS characters, then SA [i, j] is LMS type substring.
5) bucket: the suffix section with identical initial character is known as bucket;In a bucket, the character of S type concentrates on most right
The character on side, L type concentrates on Far Left.
In the concrete realization, it can use a kind of SA-IS algorithm (linear session Suffix array clustering memory structures algorithm) progress
The construction of Suffix array clustering, the algorithm are realized using dividing and ruling with recursive induction sort method, can be divided into contraction and be derived two ranks
Section, main thought are as follows:
1) inversely traversal original character string, L type, S type and the LMS type of calculating original character string obtain character type
Type array;
2) character types array is inversely traversed, calculates the position where LMS type character, and move it to the most right of bucket
Side;
3) positive traversal bucket moves it to bucket most if a upper character for LMS type character is L type character
The left side;
4) bucket is inversely traversed, if a upper character for L type character is S type character, moves it to the most right of bucket
Side;
5) positive traversal bucket, obtains all LMS type characters, obtains orderly LMS type substring;
6) relatively and name adjacent LMS type substring, by the LMS type character renamed in original character string into
Row playback;
7) positive traversal bucket obtains L type suffix in front of it by LMS type suffix, and the L type suffix is moved
To the Far Left of bucket;
8) bucket is inversely traversed, S type suffix in front of it is obtained by L type suffix, and the S type suffix is moved to
The rightmost of bucket;
9) scanned, obtain all orderly suffix.
Wherein, the 1 to 6 of above-mentioned steps is the algorithm contraction phase, and step 7 to 9 is the derivation stage.
For the domain picturename in the Attribute domain shown in the table two, the Suffix array clustering of construction can be such as five institute of table
Show.
Table five, Suffix array clustering corresponding with the domain picturename shown in table two
Metadata | animal flower |
Suffix array clustering | 6,4,0,11,7,2,5,8,3,1,9,12,10 |
In embodiments of the present invention, the domain information structure of each Attribute domain may include: the picture stored in the Attribute domain
Number of files fileNum, the metadata size currentSize in the Attribute domain, records each image data in the Attribute domain
File information structure FileInfo.Wherein, FileInfo is used to record the file information structure of each image data in the Attribute domain,
Contain index deletion marker character delete (wherein 0 is does not delete, and 1 is deletion), file corresponds to property content in the Attribute domain
In metadata size size (currentSize is the metadata size summation in current attribute domain, and size is each in the Attribute domain
The metadata size of image data, the adduction of size are exactly currentSize), image data correspond to Attribute domain metadata it is first
The ID number PID etc. of offset offset and image data of the byte in the Attribute domain metadata.
For two width picture shown in table two, the corresponding domain information structure of Attribute domain picturename can be such as table six
It is shown.
Table six, domain information structure corresponding with the domain picturename shown in table two
When carrying out picture retrieval, can then be inquired corresponding by specifying domain to be retrieved and providing data to be retrieved
Domain metadata and Suffix array clustering, obtain offset of the occurrence in the metadata of domain, inquire file information structure according to offset
FileInfo obtains the corresponding PID of image data, and the data of each Attribute domain in designated pictures file are obtained according to PID.
Text data not only can handle using suffix index it can be seen from table two, but also can handle binary data;And
The time complexity of index construction is O (n), and query time complexity is O (logn), wherein n is to construct Suffix array clustering index
String length.Due to the building speed of index be it is linear, compared with traditional indexing means, suffix index advantage is bright
It is aobvious.
In embodiments of the present invention, it after obtaining image data to be processed, by making semantic analysis to image data, obtains
To the semantic text for describing the image data, and extract the various features information of image data, be converted be characterized to
Amount, so as to respectively store semantic text and the corresponding feature vector of various features information to suffix index to be built
In multiple Attribute domains, and then suffix index is constructed based on multiple Attribute domains, realizes the storage to the information of image data.This implementation
Example is by will describe the semantic text of image content and and be used to indicate that the binary vector of picture feature to be stored jointly to suffix
In array, so as to be indexed building with unified technology, the index constructed is met simultaneously based on text
Or the searching method based on image content feature.Also, the building speed by indexing in this present embodiment be it is linear, with tradition
Indexing means comparison, construct more efficient, advantage and become apparent from.
It should be noted that the size of the serial number of each step is not meant that the order of the execution order in above-described embodiment,
The execution sequence of each process should be determined by its function and internal logic, and the implementation process without coping with the embodiment of the present invention, which is constituted, appoints
What is limited.
Referring to Fig. 2, a kind of schematic diagram of the storage device of pictorial information of one embodiment of the invention is shown, specifically may be used
To include following module:
Image data obtains module 201, for obtaining image data to be processed;
Semantic text analysis module 202 is obtained for making semantic analysis to the image data for describing the picture
The semantic text of data;
Characteristic information extracting module 203, for extracting the various features information of the image data, by the various features
Information is converted to feature vector;
Metadata storage module 204, for respectively by the semantic text and the corresponding feature of the various features information
Vector is stored into multiple Attribute domains of suffix index to be built;
Suffix index constructs module 205, for constructing suffix index based on the multiple Attribute domain, realizes to the picture
The storage of the information of data.
In embodiments of the present invention, the various features information includes local binary patterns feature LBP, direction gradient histogram
At least one of figure feature HOG, scale invariant feature SIFT.
In embodiments of the present invention, the metadata storage module 204 can specifically include following submodule:
Binary vector judging submodule, for judge the corresponding feature vector of the various features information whether be two into
Vector processed;
Binary vector transform subblock, if not being binary vector for the corresponding feature vector of any feature information,
The corresponding feature vector of the characteristic information is then converted into binary vector;
Metadata sub-module stored, for respectively with the semantic text and the corresponding binary system of the various features information
Vector stores the metadata into multiple Attribute domains of suffix index to be built as metadata.
In embodiments of the present invention, the suffix index building module 205 can specifically include following submodule:
Suffix array clustering construct submodule, for use preset Suffix array clustering construction algorithm, construction with each Attribute domain in
The corresponding Suffix array clustering of metadata;
Domain information structure determination submodule, for determining the domain information structure of each Attribute domain, the domain information knot
Structure is used to obtain metadata of the specified file in corresponding Attribute domain, and the specified file is that the condition of picture to be retrieved is believed
Breath;
Suffix index constructs submodule, for according to the metadata of each Attribute domain, Suffix array clustering and domain information knot
Structure constructs suffix index.
In embodiments of the present invention, the domain information structure of each Attribute domain may include:
The picture file number fileNum stored in the Attribute domain, the metadata size in the Attribute domain
CurrentSize records the file information structure FileInfo of each image data in the Attribute domain.
For device embodiment, since it is basically similar to the method embodiment, related so describing fairly simple
Place referring to embodiment of the method part explanation.
Referring to Fig. 3, a kind of schematic diagram of server of one embodiment of the invention is shown.As shown in figure 3, the present embodiment
Server 300 include: processor 310, memory 320 and be stored in the memory 320 and can be in the processor
The computer program 321 run on 310.The processor 310 realizes above-mentioned pictorial information when executing the computer program 321
The each embodiment of storage method in step, such as step S101 to S105 shown in FIG. 1.Alternatively, the processor 310 is held
The function of each module/unit in above-mentioned each Installation practice, such as module shown in Fig. 2 are realized when the row computer program 321
201 to 205 function.
Illustratively, the computer program 321 can be divided into one or more module/units, it is one or
Multiple module/the units of person are stored in the memory 320, and are executed by the processor 310, to complete the present invention.Institute
Stating one or more module/units can be the series of computation machine program instruction section that can complete specific function, the instruction segment
It can be used for describing implementation procedure of the computer program 321 in the server 300.For example, the computer program
321, which can be divided into image data, obtains module, semantic text analysis module, characteristic information extracting module, metadata storage
Module and suffix index construct module, and each module concrete function is as follows:
Image data obtains module, for obtaining image data to be processed;
Semantic text analysis module is obtained for making semantic analysis to the image data for describing the picture number
According to semantic text;
Characteristic information extracting module believes the various features for extracting the various features information of the image data
Breath is converted to feature vector;
Metadata storage module, for respectively by the semantic text and the corresponding feature vector of the various features information
It stores into multiple Attribute domains of suffix index to be built;
Suffix index constructs module, for constructing suffix index based on the multiple Attribute domain, realizes to the picture number
According to information storage.
The server 300 can be the calculating such as desktop PC, notebook, palm PC and cloud server and set
It is standby.The server 300 may include, but be not limited only to, processor 310, memory 320.It will be understood by those skilled in the art that
Fig. 3 is only a kind of example of server 300, does not constitute the restriction to server 300, may include more or more than illustrating
Few component perhaps combines certain components or different components, such as the server 300 can also include input and output
Equipment, network access equipment, bus etc..
The processor 310 can be central processing unit (Central Processing Unit, CPU), can also be
Other general processors, digital signal processor (Digital Signal Processor, DSP), 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,
Discrete hardware components etc..General processor can be microprocessor or the processor is also possible to any conventional processor
Deng.
The memory 320 can be the internal storage unit of the server 300, for example, server 300 hard disk or
Memory.The memory 320 is also possible to the External memory equipment of the server 300, such as is equipped on the server 300
Plug-in type hard disk, intelligent memory card (Smart Media Card, SMC), secure digital (Secure Digital, SD) card,
Flash card (Flash Card) etc..Further, the memory 320 can also both include the inside of the server 300
Storage unit also includes External memory equipment.The memory 320 is for storing the computer program 321 and the service
Other programs and data needed for device 300.The memory 320, which can be also used for temporarily storing, have been exported or will be defeated
Data out.
Embodiment described above is merely illustrative of the technical solution of the present invention, rather than its limitations.Although referring to aforementioned reality
Applying example, invention is explained in detail, those skilled in the art should understand that: it still can be to aforementioned each
Technical solution documented by embodiment is modified or equivalent replacement of some of the technical features;And these are modified
Or replacement, the spirit and scope for technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution should all
It is included within protection scope of the present invention.
Claims (10)
1. a kind of storage method of pictorial information characterized by comprising
Obtain image data to be processed;
Semantic analysis is made to the image data, obtains the semantic text for describing the image data;
The various features information is converted to feature vector by the various features information for extracting the image data;
The semantic text and the corresponding feature vector of the various features information are stored to suffix index to be built respectively
Multiple Attribute domains in;
Suffix index is constructed based on the multiple Attribute domain, realizes the storage to the information of the image data.
2. the method according to claim 1, wherein the various features information includes local binary patterns feature
At least one of LBP, histograms of oriented gradients feature HOG, scale invariant feature SIFT.
3. the method according to claim 1, wherein described respectively by the semantic text and the various features
The corresponding feature vector of information is stored includes: into multiple Attribute domains of suffix index to be built
Judge whether the corresponding feature vector of the various features information is binary vector;
If the corresponding feature vector of any feature information is not binary vector, by the corresponding feature vector of the characteristic information
Be converted to binary vector;
Respectively using the semantic text and the corresponding binary vector of the various features information as metadata, by first number
According to storing into multiple Attribute domains of suffix index to be built.
4. according to the method described in claim 3, it is characterized in that, described construct suffix index based on the multiple Attribute domain
Step includes:
Using preset Suffix array clustering construction algorithm, Suffix array clustering corresponding with metadata in each Attribute domain is constructed;
Determine the domain information structure of each Attribute domain, the domain information structure is for obtaining specified file in corresponding Attribute domain
Interior metadata, the specified file are the conditional information of picture to be retrieved;
According to the metadata of each Attribute domain, Suffix array clustering and domain information structure, suffix index is constructed.
5. according to the method described in claim 4, it is characterized in that, the domain information structure of each Attribute domain includes:
The picture file number fileNum stored in the Attribute domain, the metadata size in the Attribute domain
CurrentSize records the file information structure FileInfo of each image data in the Attribute domain.
6. a kind of storage device of pictorial information characterized by comprising
Image data obtains module, for obtaining image data to be processed;
Semantic text analysis module is obtained for making semantic analysis to the image data for describing the image data
Semantic text;
Characteristic information extracting module turns the various features information for extracting the various features information of the image data
It is changed to feature vector;
Metadata storage module, for respectively storing the semantic text and the corresponding feature vector of the various features information
Into multiple Attribute domains of suffix index to be built;
Suffix index constructs module, for constructing suffix index based on the multiple Attribute domain, realizes to the image data
The storage of information.
7. device according to claim 6, which is characterized in that the metadata storage module includes:
Binary vector judging submodule, for judge the corresponding feature vector of the various features information whether be binary system to
Amount;
Binary vector transform subblock will if not being binary vector for the corresponding feature vector of any feature information
The corresponding feature vector of the characteristic information is converted to binary vector;
Metadata sub-module stored, for respectively with the semantic text and the corresponding binary vector of the various features information
As metadata, the metadata is stored into multiple Attribute domains of suffix index to be built.
8. device according to claim 7, which is characterized in that the suffix index constructs module and includes:
Suffix array clustering constructs submodule, for using preset Suffix array clustering construction algorithm, construction and number first in each Attribute domain
According to corresponding Suffix array clustering;
Domain information structure determination submodule, for determining that the domain information structure of each Attribute domain, the domain information structure are used
In obtaining metadata of the specified file in corresponding Attribute domain, the specified file is the conditional information of picture to be retrieved;
Suffix index constructs submodule, for according to the metadata of each Attribute domain, Suffix array clustering and domain information structure, structure
Build suffix index.
9. a kind of server, including memory, processor and storage can transport in the memory and on the processor
Capable computer program, which is characterized in that the processor realizes such as claim 1 to 5 times when executing the computer program
The step of storage method of one pictorial information.
10. a kind of computer readable storage medium, the computer-readable recording medium storage has computer program, and feature exists
In the storage method of realization pictorial information as described in any one of claim 1 to 5 when the computer program is executed by processor
The step of.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910572522.1A CN110390030A (en) | 2019-06-28 | 2019-06-28 | The storage method and device of pictorial information |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910572522.1A CN110390030A (en) | 2019-06-28 | 2019-06-28 | The storage method and device of pictorial information |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110390030A true CN110390030A (en) | 2019-10-29 |
Family
ID=68285943
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910572522.1A Pending CN110390030A (en) | 2019-06-28 | 2019-06-28 | The storage method and device of pictorial information |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110390030A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113407749A (en) * | 2021-06-28 | 2021-09-17 | 北京百度网讯科技有限公司 | Picture index construction method and device, electronic equipment and storage medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101458695A (en) * | 2008-12-18 | 2009-06-17 | 西交利物浦大学 | Mixed picture index construct and enquiry method based on key word and content characteristic and use thereof |
CN103853797A (en) * | 2012-12-07 | 2014-06-11 | 中兴通讯股份有限公司 | Image retrieval method and system based on n-gram image indexing structure |
CN105912684A (en) * | 2016-04-15 | 2016-08-31 | 湘潭大学 | Cross-media retrieval method based on visual features and semantic features |
KR20160127369A (en) * | 2015-04-27 | 2016-11-04 | 삼성에스디에스 주식회사 | System and method for searching image |
CN109284273A (en) * | 2018-09-27 | 2019-01-29 | 中山大学 | A kind of mass small documents querying method and system using Suffix array clustering index |
-
2019
- 2019-06-28 CN CN201910572522.1A patent/CN110390030A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101458695A (en) * | 2008-12-18 | 2009-06-17 | 西交利物浦大学 | Mixed picture index construct and enquiry method based on key word and content characteristic and use thereof |
CN103853797A (en) * | 2012-12-07 | 2014-06-11 | 中兴通讯股份有限公司 | Image retrieval method and system based on n-gram image indexing structure |
KR20160127369A (en) * | 2015-04-27 | 2016-11-04 | 삼성에스디에스 주식회사 | System and method for searching image |
CN105912684A (en) * | 2016-04-15 | 2016-08-31 | 湘潭大学 | Cross-media retrieval method based on visual features and semantic features |
CN109284273A (en) * | 2018-09-27 | 2019-01-29 | 中山大学 | A kind of mass small documents querying method and system using Suffix array clustering index |
Non-Patent Citations (1)
Title |
---|
陈乾明: "基于内容的图像检索系统的设计与实现", 《中国优秀硕士学位论文全文数据库(信息科技辑)》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113407749A (en) * | 2021-06-28 | 2021-09-17 | 北京百度网讯科技有限公司 | Picture index construction method and device, electronic equipment and storage medium |
CN113407749B (en) * | 2021-06-28 | 2024-04-30 | 北京百度网讯科技有限公司 | Picture index construction method and device, electronic equipment and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Peng et al. | An overview of cross-media retrieval: Concepts, methodologies, benchmarks, and challenges | |
CN110399515B (en) | Picture retrieval method, device and system | |
Podlesnaya et al. | Deep learning based semantic video indexing and retrieval | |
US10831820B2 (en) | Content based image management and selection | |
Jiang et al. | Spatial and temporal content-based access to hypervideo databases | |
Kuric et al. | ANNOR: Efficient image annotation based on combining local and global features | |
Hamreras et al. | Content based image retrieval by convolutional neural networks | |
Solanki et al. | Vedl: a novel video event searching technique using deep learning | |
CN110390030A (en) | The storage method and device of pictorial information | |
Rao et al. | Deep learning-based image retrieval system with clustering on attention-based representations | |
Shekhar et al. | An object centric image retrieval framework using multi-agent model for retrieving non-redundant web images | |
Seddati et al. | Triplet networks feature masking for sketch-based image retrieval | |
Belarbi et al. | Indexing video by the content | |
Cao | Review of Image Expression of Content-based Image Retrieval Technology | |
Wang et al. | Exploring statistical correlations for image retrieval | |
Wiesen et al. | Overview of uni-modal and multi-modal representations for classification tasks | |
Doulamis et al. | 3D modelling of cultural heritage objects from photos posted over the Twitter | |
Pinge et al. | A novel video retrieval method based on object detection using deep learning | |
Salih et al. | A novel method for shape representation | |
Yang et al. | Zero-shot transfer learning based on visual and textual resemblance | |
Coustaty et al. | Towards ontology-based retrieval of historical images | |
Johny et al. | A Brief Survey Ondynamic Topic Model for Unsupervised Object Discovery and Localization | |
Om Prakash et al. | Video captioning for proactive video management using deep machine learning | |
Bulysheva et al. | Image database management architecture: logical structure and indexing methods | |
Johny et al. | Knowledge Based Unsupervised Object Discovery Using Probabilistic Randomized Hough Transform (PRHT) With Deep Learning Classification (DLC) |
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: 20191029 |