CN105243060B - A kind of method and device of retrieving image - Google Patents
A kind of method and device of retrieving image Download PDFInfo
- Publication number
- CN105243060B CN105243060B CN201410239452.5A CN201410239452A CN105243060B CN 105243060 B CN105243060 B CN 105243060B CN 201410239452 A CN201410239452 A CN 201410239452A CN 105243060 B CN105243060 B CN 105243060B
- Authority
- CN
- China
- Prior art keywords
- picture
- retrieved
- characteristic information
- terminal
- cluster
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Abstract
The present disclosure discloses a kind of method and devices of retrieving image, belong to field of computer technology.The described method includes: obtaining the characteristic information that terminal is sent, the characteristic information includes at least one of characteristics of image, face characteristic or label characteristics;Retrieval and the characteristic information picture that match degree is greater than the preset threshold in picture library, and the picture retrieved is determined as search result;The picture for meeting preset condition is being chosen as query expansion in cluster where the picture retrieved as a result, cluster where the picture retrieved is the corresponding cluster of type of the characteristic information;The search result and the query expansion result are returned into the terminal.The disclosure passes through using the picture retrieved as search result, and it chooses in the corresponding cluster of picture retrieved and meets other pictures of preset condition as query expansion result, can retrieve be more in line with user requirement as a result, improving the accuracy and efficiency of retrieval.
Description
Technical field
This disclosure relates to field of computer technology, in particular to a kind of method and device of retrieving image.
Background technique
With the development of internet, cloud space storage has been spread in the life and work of user, and user can be by hand
Various pictures in machine, PC are stored in cloud space, are retrieved and are got from cloud space by network whenever and wherever possible
The picture of oneself, is not limited by regions.
Inevitably there is the case where searching picture, search picture under normal circumstances in user when using cloud space
Mode are as follows: user can be by inputting keyword, and search has the picture of the Keyword Tag, and will exactly match keyword
Picture as search result;Or user chooses picture to be detected, by using the mode of face characteristic or characteristics of image
Retrieval is greater than the picture of certain threshold value with the presence or absence of similarity, and using the picture retrieved as search result.
Inventor find the relevant technologies the prior art has at least the following problems:
For the mode of keyword, it is only capable of retrieving the picture for exactly matching the keyword;For face characteristic or image
Feature is known otherwise, is also only capable of retrieving the picture that similarity is greater than certain threshold value.The limit of above two retrieval mode
System is excessive, and the picture retrieved can not show the Search Requirement of user completely.
Summary of the invention
In order to solve the problems, such as that the relevant technologies, the embodiment of the present disclosure provide a kind of method and device of retrieving image.Institute
It is as follows to state technical solution:
According to the first aspect of the embodiments of the present disclosure, a kind of method of retrieving image is provided, comprising:
The characteristic information that terminal is sent is obtained, the characteristic information includes in characteristics of image, face characteristic or label characteristics
At least one;
Retrieval and the characteristic information picture that match degree is greater than the preset threshold in picture library, and retrieved described
Picture is determined as search result;
The picture for meeting preset condition is being chosen as query expansion as a result, institute in cluster where the picture retrieved
Cluster where stating the picture retrieved is the corresponding cluster of type of the characteristic information;
The search result and the query expansion result are returned into the terminal.
According to the second aspect of an embodiment of the present disclosure, a kind of device of retrieving image is provided, comprising:
First obtains module, and for obtaining the characteristic information of terminal transmission, the characteristic information includes characteristics of image, face
At least one of feature or label characteristics;
Retrieval module, for the retrieval in picture library and the characteristic information picture that match degree is greater than the preset threshold, and
The picture retrieved is determined as search result;
First chooses module, for choosing the picture work for meeting preset condition in cluster where the picture retrieved
For query expansion as a result, cluster where the picture retrieved is the corresponding cluster of type of the characteristic information;
Return module, for the search result and the query expansion result to be returned to the terminal.
The technical scheme provided by this disclosed embodiment can include the following benefits: by making the picture retrieved
For search result, and chooses in the corresponding cluster of picture retrieved and meet other pictures of preset condition as query expansion knot
Fruit, can retrieve be more in line with user requirement as a result, improving the accuracy and efficiency of retrieval.
It should be understood that above general description and following detailed description be only it is exemplary and explanatory, not
The disclosure can be limited.
Detailed description of the invention
The drawings herein are incorporated into the specification and forms part of this specification, and shows the implementation for meeting the disclosure
Example, and together with specification for explaining the principles of this disclosure.
Fig. 1 is the method flow diagram that the disclosure is a kind of retrieving image shown according to an exemplary embodiment;
Fig. 2 is the method flow diagram that the disclosure is a kind of retrieving image shown according to an exemplary embodiment;
Fig. 3 is the method flow diagram that the disclosure is a kind of retrieving image shown according to an exemplary embodiment;
Fig. 4 is the method flow diagram that the disclosure is a kind of retrieving image shown according to an exemplary embodiment;
Fig. 5 is the apparatus structure schematic diagram that the disclosure is a kind of retrieving image shown according to an exemplary embodiment;
Fig. 6 is the apparatus structure schematic diagram that the disclosure is a kind of retrieving image shown according to an exemplary embodiment.
Specific embodiment
To keep the purposes, technical schemes and advantages of the disclosure clearer, below in conjunction with attached drawing to disclosure embodiment party
Formula is described in further detail.
Embodiment one
The embodiment of the present disclosure provides a kind of method of retrieving image, and referring to Fig. 1, method flow includes:
In a step 101, the characteristic information that terminal is sent is obtained, characteristic information includes characteristics of image, face characteristic or mark
At least one of feature is signed, the method for the present embodiment is in server;
In a step 102, retrieval and the characteristic information picture that match degree is greater than the preset threshold in picture library, and will retrieval
To picture be determined as search result, the method for the present embodiment is in server;
In step 103, the picture for meeting preset condition is being chosen where the picture retrieved in cluster as extension inspection
Rope is as a result, cluster where the picture retrieved is characterized the corresponding cluster of type of information, and the method for the present embodiment is for servicing
In device.
At step 104, search result and query expansion result are returned into terminal, the method for the present embodiment is for taking
It is engaged in device.
Wherein, it before executing step 101 and starting to carry out the process of retrieving image, needs in advance in same data source
Picture row cluster will belong to same category of picture and be gathered in a cluster.In the embodiments of the present disclosure for clustering method
Can include but is not limited to following manner: the cluster based on characteristics of image, the cluster based on face characteristic are poly- based on label
Class.
Cluster based on characteristics of image will belong to matching by calculating the feature vector of key point in picture and being compared
The picture that degree is greater than certain threshold value is confirmed as a cluster, belongs in the same cluster that can be all stored between each picture
With degree.It is extracted in the embodiments of the present disclosure by using SIFT (Scale-invariant feature transform) algorithm
The key point of picture, the algorithm are the algorithms for detecting local feature, and each key point obtains in picture after being calculated by SIFT algorithm
The feature vector of the SIFT feature vector tieed up to one 128, final all key points constitutes characteristics of image.When final comparison,
Similarity determination measurement by the Euclidean distance between key point feature vector as key point in two images.Optional degree
Amount mode can also include cosine similarity amount, not define herein.
Cluster based on face characteristic will be believed by extracting the feature of face in picture comprising the feature of the same face
The picture of breath but thinks a cluster, is confirmed as in the cluster of the same face, that can be all stored between each picture
With degree.The feature structure of final face constitutes the face characteristic of picture.
Cluster based on label can be divided into two kinds:
The first are as follows: it is clustered, will be belonged to certain according to the temporal information and/or position information tag that include in picture
Time range, the picture of certain position range are determined as a cluster.Wherein time range, position range do not define.
In the first case, temporal information and location information are obtained by extracting the exif information in picture.Believe eventually by the time
Breath and/or location information obtain the label characteristics of picture.
Second are as follows: clustered according to the text information in the label characteristics for the user's calibration for including in picture.Wherein,
In second situation, text information be can include but is not limited to: characters name, scenery, place name etc., through user in picture
It is demarcated to obtain.The text information of picture is obtained in label characteristics eventually by user's calibration.
Further, there is also similarities between picture each in each cluster, wherein for characteristics of image, face
In the cluster of feature, the similarity between each picture can be carried out by characteristics of image or face characteristic;For label characteristics
Cluster in, then can be by the time span in the temporal information in label characteristics, the distance in location information is each to calculate
Similarity between a picture.
In the embodiments of the present disclosure, in server there are above-mentioned a variety of clusters obtained based on different cluster modes, so as to
User finds more accurate search result by sending different characteristic informations.
The embodiment of the present disclosure passes through the picture that will be retrieved as search result, and it is corresponding poly- to choose the picture retrieved
Meet other pictures of preset condition in class as query expansion as a result, can retrieve be more in line with user requirement as a result,
Improve the accuracy and efficiency of retrieval.
Embodiment two
The embodiment of the present disclosure provides a kind of method of retrieving image, and referring to fig. 2, method flow includes:
In step 201, the characteristic information that terminal is sent is obtained, characteristic information includes characteristics of image, the side of the present embodiment
Method is in server.
Wherein, in the embodiments of the present disclosure, user chooses picture to be retrieved in the terminal, determines that retrieval is to be retrieved with this
The similar picture of picture.At this point, the SIFT feature vector of each key point in the picture is calculated by SIFT algorithm in the terminal,
The characteristics of image is obtained, and the characteristics of image is sent to server.
Optionally, after user chooses picture to be retrieved in terminal, the picture to be retrieved directly can also be issued into clothes
Business device, after server receives the picture to be retrieved, carries out the process for obtaining the characteristic information of the picture to be retrieved.Its
In, server obtains the characteristic information of the picture to be retrieved, including characteristics of image and/or face characteristic.
In step 202, the picture and terminal stored in other picture libraries according to the mark of terminal, acquisition terminal exists
The corresponding characteristic information of the picture stored in other picture libraries, the method for the present embodiment is in server.
Wherein, the mark of terminal can be the contents such as account, by the mark to other corresponding picture libraries of terminal
Send the request for obtaining picture.
Wherein, other picture libraries can be associated in advance, be previously stored with other picture libraries in such situation
Address;Or terminal carries other picture libraries in the characteristic information of transmission or the retrieval request of picture to be retrieved
Address.
In step 203, the corresponding characteristic information of the picture stored in other picture libraries according to terminal, to terminal at it
The picture that he stores in picture library is clustered and is stored in the picture library of itself, and the method for the present embodiment is used for server
In.
Wherein, the corresponding characteristic information of the picture stored in other picture libraries is obtained, may include: characteristics of image, face
At least one of feature or label characteristics.Wherein, for characteristics of image, face characteristic can by calculating process as implement
In example one, do not repeat herein.It for label characteristics, can be got in the attribute of picture, such as: temporal information, position letter
Breath, text information etc..
Further, the process clustered in the process and embodiment one clustered according to characteristic information to picture is identical,
It does not repeat herein.
In step 204, when characteristic information is characteristics of image, it is corresponding with picture every in picture library to calculate characteristics of image
Characteristics of image between matching degree, and choosing the picture that match degree is greater than the preset threshold is the picture retrieved, and will retrieval
To picture be determined as search result, the method for the present embodiment is in server.
Wherein, the matching degree between picture is calculated by the mode of the Euclidean distance mode of selection or cosine similarity amount.
Matching degree can numerical value between 0-100.It is more similar that numerical value more Gao Ze represents two pictures, and sets preset threshold as phase
Like the judgment criteria of picture.
In step 205, the picture for meeting preset condition is being chosen where the picture retrieved in cluster as extension inspection
Rope is as a result, cluster where the picture retrieved is characterized the corresponding cluster of type of information, and the method for the present embodiment is for servicing
In device.
Wherein, other picture conducts are further chosen in the corresponding cluster of characteristics of image that the picture retrieved belongs to
Query expansion as a result, the similarity according to the mode chosen in the picture retrieved and the cluster between other pictures be greater than it is pre-
If similarity, determine choose the picture as query expansion result.
Such as: the matching degree of the picture E to be retrieved that user the chooses and picture A retrieved is 90, is belonged to together with picture A
One cluster includes B, C, D.Wherein, the matching degree between A is 90,80,70 respectively by B, C, D.So E and B, C, between D
Matching degree is respectively as follows: 90- (100-90) i.e. 80,90- (100-80) i.e. 70,90- (100-70) i.e. 60, wherein default similarity is
70, then picture of the picture similarity chosen and retrieved greater than 70 is as query expansion result.
The sequence of arrangement is generated according to matching degree of other pictures in spreading result, this can sequentially be pressed in terminal display
It is shown according to sequence from high to low.
Therefore, step 205 can be further are as follows:
In the similarity between picture in cluster, chosen and retrieved where the picture retrieved greater than default similarity
Picture as query expansion result.
In step 206, search result and query expansion result are returned into terminal, the method for the present embodiment is for taking
It is engaged in device.
It has been more than the display model of current screen if picture is excessive during terminal display for the picture that retrieves
It encloses, is then shown that other do not show according to the part that can be shown in terminal screen that puts in order in query expansion result
The picture shown is in order according to the carry out sequence displaying of the sliding action of user.
The embodiment of the present disclosure passes through the picture that will be retrieved as search result, and it is corresponding poly- to choose the picture retrieved
Meet other pictures of preset condition in class as query expansion as a result, can retrieve be more in line with user requirement as a result,
Improve the accuracy and efficiency of retrieval.
Embodiment three
The embodiment of the present disclosure provides a kind of method of retrieving image, and referring to Fig. 3, method flow includes:
In step 301, the characteristic information that terminal is sent is received, characteristic information includes face characteristic, the side of the present embodiment
Method is in server.
Wherein, in the embodiments of the present disclosure, user chooses the face in picture to be retrieved in the terminal, determine retrieval with
The similar picture of face in the picture to be retrieved.At this point, it is to be retrieved to calculate this by face recognition algorithms in the terminal
Face characteristic in picture, and the face characteristic is sent to server.
Optionally, after user chooses picture to be retrieved in terminal, the picture to be retrieved directly can also be issued into clothes
Business device, after server receives the picture to be retrieved, carries out the process for obtaining the characteristic information of the picture to be retrieved.Its
In, server obtains the characteristic information of the picture to be retrieved, including characteristics of image and/or face characteristic.
In step 302, the picture and terminal stored in other picture libraries according to the mark of terminal, acquisition terminal exists
The corresponding characteristic information of the picture stored in other picture libraries, the method for the present embodiment is in server.
Wherein, the mark of terminal can be the contents such as account, by the mark to other corresponding picture libraries of terminal
Send the request for obtaining picture.
Wherein, other picture libraries can be associated in advance, be previously stored with other picture libraries in such situation
Address;Or terminal carries other picture libraries in the characteristic information of transmission or the retrieval request of picture to be retrieved
Address.
In step 303, the corresponding characteristic information of the picture stored in other picture libraries according to terminal, to terminal at it
The picture that he stores in picture library is clustered and is stored in the picture library of itself, and the method for the present embodiment is used for server
In.
Wherein, the corresponding characteristic information of the picture stored in other picture libraries is obtained, may include: characteristics of image, face
At least one of feature or label characteristics.Wherein, for characteristics of image, face characteristic can by calculating process as implement
In example one, do not repeat herein.It for label characteristics, can be got in the attribute of picture, such as: temporal information, position letter
Breath, text information etc..
Further, the process clustered in the process and embodiment one clustered according to characteristic information to picture is identical,
It does not repeat herein.
In step 304, when characteristic information is face characteristic, the people of every picture in face characteristic and picture library is calculated
Matching degree between face feature, and choosing the picture that match degree is greater than the preset threshold is the picture retrieved, and will retrieve
Picture is determined as search result, and the method for the present embodiment is in server.
It wherein, can be with for the calculating process of the matching degree of face characteristic are as follows:
Calculate the matching between each characteristic point in each characteristic point in face characteristic and picture library in every picture
Degree.
In step 305, the picture for meeting preset condition is being chosen where the picture retrieved in cluster as extension inspection
Rope is as a result, cluster where the picture retrieved is characterized the corresponding cluster of type of information, and the method for the present embodiment is for servicing
In device.
Wherein, other picture conducts are further chosen in the corresponding cluster of face characteristic that the picture retrieved belongs to
Query expansion as a result, the similarity according to the mode chosen in the picture retrieved and the cluster between other pictures be greater than it is pre-
If similarity, determine choose the picture as query expansion result.
Therefore, step 305 can be further are as follows:
In the similarity between picture in cluster, chosen and retrieved where the picture retrieved greater than default similarity
Picture as query expansion result.
Within step 306, search result and query expansion result are returned into terminal, the method for the present embodiment is for taking
It is engaged in device.
It has been more than the display model of current screen if picture is excessive during terminal display for the picture that retrieves
It encloses, is then shown that other do not show according to the part that can be shown in terminal screen that puts in order in query expansion result
The picture shown is in order according to the carry out sequence displaying of the sliding action of user.
The embodiment of the present disclosure passes through the picture that will be retrieved as search result, and it is corresponding poly- to choose the picture retrieved
Meet other pictures of preset condition in class as query expansion as a result, can retrieve be more in line with user requirement as a result,
Improve the accuracy and efficiency of retrieval.
Example IV
The embodiment of the present disclosure provides a kind of method of retrieving image, and referring to fig. 4, method flow includes:
In step 401, the characteristic information that terminal is sent is received, characteristic information includes label characteristics, the side of the present embodiment
Method is in server.
Wherein, in the embodiments of the present disclosure, user inputs label to be detected in the terminal, determines that the retrieval label is similar
Picture.At this point, the label characteristics are sent to server in terminal.
Optionally, after user chooses picture to be retrieved in terminal, the picture to be retrieved directly can also be issued into clothes
Business device, after server receives the picture to be retrieved, carries out the process for obtaining the characteristic information of the picture to be retrieved.Its
In, server obtains the characteristic information of the picture to be retrieved, including characteristics of image and/or face characteristic.
In step 402, the picture and terminal stored in other picture libraries according to the mark of terminal, acquisition terminal exists
The corresponding characteristic information of the picture stored in other picture libraries, the method for the present embodiment is in server.
Wherein, the mark of terminal can be the contents such as account, by the mark to other corresponding picture libraries of terminal
Send the request for obtaining picture.
Wherein, other picture libraries can be associated in advance, be previously stored with other picture libraries in such situation
Address;Or terminal carries other picture libraries in the characteristic information of transmission or the retrieval request of picture to be retrieved
Address.
In step 403, the corresponding characteristic information of the picture stored in other picture libraries according to terminal, to terminal at it
The picture that he stores in picture library is clustered and is stored in the picture library of itself, and the method for the present embodiment is used for server
In.
Wherein, the corresponding characteristic information of the picture stored in other picture libraries is obtained, may include: characteristics of image, face
At least one of feature or label characteristics.Wherein, for characteristics of image, face characteristic can by calculating process as implement
In example one, do not repeat herein.It for label characteristics, can be got in the attribute of picture, such as: temporal information, position letter
Breath, text information etc..
Further, the process clustered in the process and embodiment one clustered according to characteristic information to picture is identical,
It does not repeat herein.
In step 404, when characteristic information is label characteristics, it is corresponding with picture every in picture library that label characteristics are calculated
Label characteristics between matching degree, and choosing the picture that match degree is greater than the preset threshold is the picture retrieved, and will retrieval
To picture be determined as search result, the method for the present embodiment is in server.
Wherein, for temporal information, location information and the text information in label characteristics, it is greater than in advance choosing matching degree
If when the picture of threshold value, the picture that label characteristics are consistent completely with the label characteristics received being chosen in picture library and is used as retrieval
As a result.
Such as: label characteristics are " location information ", such as longitude and latitude, search the picture for belonging to the longitude and latitude in location information,
And the picture found is determined as to the picture retrieved;
Alternatively, label characteristics are temporal information " 2010 " and location information " Beijing ", chosen position is believed in picture library
It include that the picture that Beijing and shooting time are 2010 is determined as the picture retrieved in breath;
Alternatively, label characteristics are text information " Beijing ", after semantics recognition, determine that " Beijing " is location information, and
The longitude and latitude of the location information is obtained, searches the picture for belonging to the longitude and latitude in location information, and the picture found is determined
For the picture retrieved;
Alternatively, label characteristics are text information " Zhang San ", the label comprising the name is searched, and the picture found is true
It is set to the picture retrieved.
Correspondingly, the embodiment of the present disclosure can be further for step 404 are as follows:
Calculate the temporal information in the temporal information label characteristics corresponding with picture every in picture library in label characteristics
Matching degree, and choosing the picture that match degree is greater than the preset threshold is the picture retrieved;Alternatively,
Calculate the location information in the location information label characteristics corresponding with picture every in picture library in label characteristics
Matching degree, and choosing the picture that match degree is greater than the preset threshold is the picture retrieved.
In step 405, the picture for meeting preset condition is being chosen where the picture retrieved in cluster as extension inspection
Rope is as a result, cluster where the picture retrieved is characterized the corresponding cluster of type of information, and the method for the present embodiment is for servicing
In device.
Wherein, other picture conducts are further chosen in the corresponding cluster of label characteristics that the picture retrieved belongs to
Query expansion as a result, the similarity according to the mode chosen in the picture retrieved and the cluster between other pictures be greater than it is pre-
If similarity, determine choose the picture as query expansion result.
Therefore, step 405 can be further are as follows:
In the similarity between picture in cluster, chosen and retrieved where the picture retrieved greater than default similarity
Picture as query expansion result.
In a step 406, choose that the picture retrieved belongs to the corresponding cluster of characteristics of image or face characteristic is corresponding poly-
Other pictures in class are as query expansion as a result, the method for the present embodiment is in server.
Such as: label characteristics be location information " the Forbidden City ", found in the picture demarcated by label it is all belong to " therefore
After the picture in palace ", it is understood that there may be not by label demarcate include " the Forbidden City " scenery picture.At this point, being looked by searching for this
The picture for belonging to " the Forbidden City " scenery is got in the corresponding cluster of characteristics of image belonging to the picture found, to get more
Accurately search result.
Again for example: label characteristics are text information " Zhang San ", and all belong to is found in the picture demarcated by label
After the picture of " Zhang San ", it is understood that there may be not by label demarcate include " Zhang San " personage picture.At this point, by searching for this
The picture for belonging to " Zhang San " personage is got in the corresponding cluster of face characteristic belonging to the picture found, to get more
Add accurately search result.
In step 407, search result and query expansion result are returned into terminal, the method for the present embodiment is for taking
It is engaged in device.
It has been more than the display model of current screen if picture is excessive during terminal display for the picture that retrieves
It encloses, is then shown that other do not show according to the part that can be shown in terminal screen that puts in order in query expansion result
The picture shown is in order according to the carry out sequence displaying of the sliding action of user.
The embodiment of the present disclosure passes through the picture that will be retrieved as search result, and it is corresponding poly- to choose the picture retrieved
Meet other pictures of preset condition in class as query expansion as a result, can retrieve be more in line with user requirement as a result,
Improve the accuracy and efficiency of retrieval.
Embodiment five
The embodiment of the present disclosure provides a kind of device of retrieving image, and referring to Fig. 5, which includes:
First obtains module 501, and for obtaining the characteristic information of terminal transmission, characteristic information includes characteristics of image, face
At least one of feature or label characteristics;
Retrieval module 502, for the retrieval in picture library and the characteristic information picture that match degree is greater than the preset threshold, and will
The picture retrieved is determined as search result;
First chooses module 503, for choosing the picture work for meeting preset condition in cluster where the picture retrieved
For query expansion as a result, cluster is characterized the corresponding cluster of type of information where the picture retrieved;
Return module 504, for search result and query expansion result to be returned to terminal.
Wherein, the first acquisition module 501 is used for:
The picture to be detected that terminal is sent is received, and characteristic information is calculated in picture to be detected.
Wherein, device further include:
Second obtains module 505, for the mark according to terminal, obtain the picture that terminal stores in other picture libraries with
And the corresponding characteristic information of picture that terminal stores in other picture libraries;
Cluster module 506, the corresponding characteristic information of picture for being stored in other picture libraries according to terminal, to terminal
The picture stored in other picture libraries clustered and be stored in itself picture library in.
Wherein, retrieval module 502, comprising:
First retrieval unit, for calculating every figure in characteristics of image and picture library when characteristic information is characteristics of image
Matching degree between the corresponding characteristics of image of piece, and choosing the picture that match degree is greater than the preset threshold is the picture retrieved.
Wherein, retrieval module 502, comprising:
Second retrieval unit, for calculating in face characteristic and picture library every when characteristic information is face characteristic
Matching degree between the face characteristic of picture, and choosing the picture that match degree is greater than the preset threshold is the picture retrieved.
Wherein, the second retrieval unit is used for:
Calculate the matching between each characteristic point in each characteristic point in face characteristic and picture library in every picture
Degree.
Wherein, retrieval module 502, comprising:
Third retrieval unit, for calculating in label characteristics and picture library every when characteristic information is label characteristics
Matching degree between the corresponding label characteristics of picture, and choosing the picture that match degree is greater than the preset threshold is the picture retrieved.
Wherein, third retrieval unit, comprising:
First retrieval subelement, for calculating the mark corresponding with picture every in picture library of the temporal information in label characteristics
The matching degree of the temporal information in feature is signed, and choosing the picture that match degree is greater than the preset threshold is the picture retrieved;
Second retrieval subelement, for calculating the mark corresponding with picture every in picture library of the location information in label characteristics
The matching degree of the location information in feature is signed, and choosing the picture that match degree is greater than the preset threshold is the picture retrieved.
Wherein, device further include:
Second chooses module 507, belongs to the corresponding cluster of characteristics of image or face characteristic for choosing the picture retrieved
Other pictures in corresponding cluster are as query expansion result.
Wherein, the first selection module 503 is used for:
In the similarity between picture in cluster, chosen and retrieved where the picture retrieved greater than default similarity
Picture as query expansion result.
The embodiment of the present disclosure passes through the picture that will be retrieved as search result, and it is corresponding poly- to choose the picture retrieved
Meet other pictures of preset condition in class as query expansion as a result, can retrieve be more in line with user requirement as a result,
Improve the accuracy and efficiency of retrieval.
Embodiment six
Fig. 6 is a kind of block diagram of device 1900 for retrieving image shown according to an exemplary embodiment.For example, dress
Setting 1900 may be provided as a server.Referring to Fig. 6, it further comprises one that device 1900, which includes processing component 1922,
Or multiple processors and memory resource represented by a memory 1932, it can holding by processing component 1922 for storing
Capable instruction, such as application program.The application program stored in memory 1932 may include one or more each
A module for corresponding to one group of instruction.In addition, processing component 1922 is configured as executing instruction, to execute above-mentioned retrieving image
Method.
Device 1900 can also include that a power supply module 1926 be configured as the power management of executive device 1900, and one
Wired or wireless network interface 1950 is configured as device 1900 being connected to network and input and output (I/O) interface
1958.Device 1900 can be operated based on the operating system for being stored in memory 1932, such as Windows ServerTM, Mac
OS XTM, UnixTM, LinuxTM, FreeBSDTM or similar.
The embodiment of the present disclosure passes through the picture that will be retrieved as search result, and it is corresponding poly- to choose the picture retrieved
Meet other pictures of preset condition in class as query expansion as a result, can retrieve be more in line with user requirement as a result,
Improve the accuracy and efficiency of retrieval.
Above-mentioned embodiment of the present disclosure serial number is for illustration only, does not represent the advantages or disadvantages of the embodiments.
Those of ordinary skill in the art will appreciate that realizing that all or part of the steps of above-described embodiment can pass through hardware
It completes, relevant hardware can also be instructed to complete by program, the program can store in a kind of computer-readable
In storage medium, storage medium mentioned above can be read-only memory, disk or CD etc..
The foregoing is merely the preferred embodiments of the disclosure, not to limit the disclosure, all spirit in the disclosure and
Within principle, any modification, equivalent replacement, improvement and so on be should be included within the protection scope of the disclosure.
Claims (20)
1. a kind of method of retrieving image, which is characterized in that be applied in server, which comprises
The picture to be retrieved that terminal is sent is obtained, the characteristic information of the picture to be retrieved, the characteristic information packet are obtained
It includes a variety of in characteristics of image, face characteristic or label characteristics;
Retrieval and the characteristic information picture that match degree is greater than the preset threshold in picture library, and by the picture retrieved
It is determined as search result;
It in cluster, is choosing where the picture retrieved and is being greater than default phase with the similarity between the picture retrieved
Like the picture of degree as query expansion as a result, cluster is corresponding for the type of the characteristic information where the picture retrieved
It clusters, the similarity in each cluster between each picture is stored in the server;
The search result and the query expansion result are returned into the terminal.
2. the method according to claim 1, wherein the characteristic information for obtaining terminal and sending, comprising:
The picture to be detected that terminal is sent is received, and characteristic information is calculated in the picture to be detected.
3. the method according to claim 1, wherein described retrieve in picture library matches with the characteristic information
Degree is greater than before the picture of preset threshold, the method also includes:
According to the mark of the terminal, picture and the terminal that the terminal stores in other picture libraries are obtained at other
The corresponding characteristic information of the picture stored in picture library;
The corresponding characteristic information of picture stored in other picture libraries according to the terminal, to the terminal in other picture libraries
The picture of middle storage is clustered and is stored in the picture library of itself.
4. the method according to claim 1, wherein described retrieve in picture library matches with the characteristic information
Degree is greater than the picture of preset threshold, comprising:
When the characteristic information is characteristics of image, it is special to calculate described image feature image corresponding with picture every in picture library
Matching degree between sign, and choosing the picture that match degree is greater than the preset threshold is the picture retrieved.
5. the method according to claim 1, wherein described retrieve in picture library matches with the characteristic information
Degree is greater than the picture of preset threshold, comprising:
When the characteristic information is face characteristic, calculate every picture in the face characteristic and picture library face characteristic it
Between matching degree, and choosing the picture that match degree is greater than the preset threshold is the picture retrieved.
6. according to the method described in claim 5, it is characterized in that, described calculate every figure in the face characteristic and picture library
Matching degree between the face characteristic of piece, comprising:
Calculate the matching between each characteristic point in each characteristic point in the face characteristic and picture library in every picture
Degree.
7. the method according to claim 1, wherein described retrieve in picture library matches with the characteristic information
Degree is greater than the picture of preset threshold, comprising:
When the characteristic information is label characteristics, it is special to calculate label characteristics label corresponding with picture every in picture library
Matching degree between sign, and choosing the picture that match degree is greater than the preset threshold is the picture retrieved.
8. the method according to the description of claim 7 is characterized in that described calculate every figure in the label characteristics and picture library
Matching degree between the corresponding label characteristics of piece, and choosing the picture that match degree is greater than the preset threshold is the picture retrieved, packet
It includes:
Calculate the temporal information in the temporal information label characteristics corresponding with picture every in picture library in the label characteristics
Matching degree, and choosing the picture that match degree is greater than the preset threshold is the picture retrieved;Alternatively,
Calculate the location information in the location information label characteristics corresponding with picture every in picture library in the label characteristics
Matching degree, and choosing the picture that match degree is greater than the preset threshold is the picture retrieved.
9. the method according to the description of claim 7 is characterized in where the picture retrieved described in the selection in cluster
After other pictures are as query expansion result, the method also includes:
The picture retrieved described in selection belongs to other figures in the corresponding cluster of characteristics of image or the corresponding cluster of face characteristic
Piece is as query expansion result.
10. a kind of device of retrieving image, which is characterized in that be applied in server, described device includes:
First obtains module, and for obtaining the characteristic information of terminal transmission, the characteristic information includes characteristics of image, face characteristic
Or at least one of label characteristics;
Retrieval module, for the retrieval in picture library and the characteristic information picture that match degree is greater than the preset threshold, and by institute
It states the picture retrieved and is determined as search result;
First chooses module, in cluster, choosing between the picture retrieved where the picture retrieved
Similarity be greater than the picture of default similarity as query expansion as a result, cluster where the picture retrieved is the spy
The corresponding cluster of type of reference breath, the similarity in each cluster between each picture are stored in the server;
Return module, for the search result and the query expansion result to be returned to the terminal.
11. device according to claim 10, which is characterized in that the first acquisition module is used for:
The picture to be detected that terminal is sent is received, and characteristic information is calculated in the picture to be detected.
12. device according to claim 10, which is characterized in that described device further include:
Second acquisition module obtains the picture that the terminal stores in other picture libraries for the mark according to the terminal
And the corresponding characteristic information of picture that the terminal stores in other picture libraries;
Cluster module, the corresponding characteristic information of picture for being stored in other picture libraries according to the terminal, to the end
Hold the picture that stores in other picture libraries to be clustered and be stored in itself picture library in.
13. device according to claim 10, which is characterized in that the retrieval module, comprising:
First retrieval unit, it is every in described image feature and picture library for calculating when the characteristic information is characteristics of image
Matching degree between the corresponding characteristics of image of picture, and choosing the picture that match degree is greater than the preset threshold is the figure retrieved
Piece.
14. device according to claim 10, which is characterized in that the retrieval module, comprising:
Second retrieval unit, for calculating in the face characteristic and picture library when the characteristic information is face characteristic
Matching degree between the face characteristic of every picture, and choosing the picture that match degree is greater than the preset threshold is the picture retrieved.
15. device according to claim 14, which is characterized in that second retrieval unit is used for:
Calculate the matching between each characteristic point in each characteristic point in the face characteristic and picture library in every picture
Degree.
16. device according to claim 14, which is characterized in that the retrieval module, comprising:
Third retrieval unit, for calculating in the label characteristics and picture library when the characteristic information is label characteristics
Matching degree between the corresponding label characteristics of every picture, and choosing the picture that match degree is greater than the preset threshold is the figure retrieved
Piece.
17. device according to claim 16, which is characterized in that the third retrieval unit, comprising:
First retrieval subelement, for calculating the mark corresponding with picture every in picture library of the temporal information in the label characteristics
The matching degree of the temporal information in feature is signed, and choosing the picture that match degree is greater than the preset threshold is the picture retrieved;
Second retrieval subelement, for calculating the mark corresponding with picture every in picture library of the location information in the label characteristics
The matching degree of the location information in feature is signed, and choosing the picture that match degree is greater than the preset threshold is the picture retrieved.
18. device according to claim 16, which is characterized in that described device further include:
Second chooses module, and for choosing, the picture retrieved belongs to the corresponding cluster of characteristics of image or face characteristic is corresponding
Cluster in other pictures as query expansion result.
19. a kind of server, which is characterized in that the server includes processor and memory, is stored in the memory
Program code, said program code are loaded by the processor and are executed to realize such as the described in any item inspections of claim 1-9
The method of rope picture.
20. a kind of storage medium, which is characterized in that be stored with instruction on the storage medium, described instruction is executed by processor
Method to complete the described in any item retrieving images of claim 1-9.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410239452.5A CN105243060B (en) | 2014-05-30 | 2014-05-30 | A kind of method and device of retrieving image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410239452.5A CN105243060B (en) | 2014-05-30 | 2014-05-30 | A kind of method and device of retrieving image |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105243060A CN105243060A (en) | 2016-01-13 |
CN105243060B true CN105243060B (en) | 2019-11-08 |
Family
ID=55040712
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410239452.5A Active CN105243060B (en) | 2014-05-30 | 2014-05-30 | A kind of method and device of retrieving image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105243060B (en) |
Families Citing this family (24)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10891019B2 (en) * | 2016-02-29 | 2021-01-12 | Huawei Technologies Co., Ltd. | Dynamic thumbnail selection for search results |
CN106021704A (en) * | 2016-05-17 | 2016-10-12 | 京东方科技集团股份有限公司 | Dressing case, control method of dressing case and makeup system |
CN105956186A (en) * | 2016-06-03 | 2016-09-21 | 捷开通讯科技(上海)有限公司 | Picture search system and method |
CN106446108B (en) * | 2016-09-14 | 2019-11-19 | 维沃移动通信有限公司 | A kind of sharing method of picture, mobile terminal and server |
CN106874513A (en) * | 2017-03-07 | 2017-06-20 | 维沃移动通信有限公司 | A kind of photo lookup method and mobile terminal |
CN108664526B (en) * | 2017-04-01 | 2022-04-29 | 华为技术有限公司 | Retrieval method and device |
CN108108711B (en) * | 2017-12-29 | 2019-12-17 | 深圳云天励飞技术有限公司 | Face control method, electronic device and storage medium |
CN108921034A (en) | 2018-06-05 | 2018-11-30 | 北京市商汤科技开发有限公司 | Face matching process and device, storage medium |
CN109241314A (en) * | 2018-08-27 | 2019-01-18 | 维沃移动通信有限公司 | A kind of selection method and device of similar image |
CN109308299B (en) * | 2018-09-12 | 2020-01-14 | 北京字节跳动网络技术有限公司 | Method and apparatus for searching information |
CN111078924B (en) * | 2018-10-18 | 2024-03-01 | 深圳云天励飞技术有限公司 | Image retrieval method, device, terminal and storage medium |
CN109711298B (en) * | 2018-12-14 | 2021-02-12 | 南京甄视智能科技有限公司 | Method and system for efficient face characteristic value retrieval based on faiss |
CN110083732B (en) * | 2019-03-12 | 2021-08-31 | 浙江大华技术股份有限公司 | Picture retrieval method and device and computer storage medium |
CN110134815A (en) * | 2019-04-15 | 2019-08-16 | 深圳壹账通智能科技有限公司 | Image processing method, device, computer equipment and storage medium |
CN110750661B (en) * | 2019-09-04 | 2022-09-16 | 成都华为技术有限公司 | Method, device, computer equipment and storage medium for searching image |
CN110851634B (en) * | 2019-11-18 | 2022-07-19 | 广东三维家信息科技有限公司 | Picture retrieval method and device and electronic equipment |
CN111881322B (en) * | 2020-09-28 | 2020-12-25 | 成都睿沿科技有限公司 | Target searching method and device, electronic equipment and storage medium |
CN112232290B (en) * | 2020-11-06 | 2023-12-08 | 四川云从天府人工智能科技有限公司 | Data clustering method, server, system and computer readable storage medium |
CN112182262B (en) * | 2020-11-30 | 2021-03-19 | 江西师范大学 | Image query method based on feature classification |
CN113286040B (en) * | 2021-05-17 | 2024-01-23 | 广州三星通信技术研究有限公司 | Method for selecting image and apparatus therefor |
CN113360698A (en) * | 2021-06-30 | 2021-09-07 | 北京海纳数聚科技有限公司 | Picture retrieval method based on image-text semantic transfer technology |
CN115168637B (en) * | 2022-06-17 | 2023-03-17 | 杭州晶彩数字科技有限公司 | Method, system and storage medium for adding label to picture |
CN114880514B (en) * | 2022-07-05 | 2022-11-01 | 人民中科(北京)智能技术有限公司 | Image retrieval method, image retrieval device and storage medium |
CN115455061B (en) * | 2022-11-14 | 2023-04-18 | 广东电网有限责任公司佛山供电局 | User characteristic quick retrieval method based on heterogeneous computing |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101859326A (en) * | 2010-06-09 | 2010-10-13 | 南京大学 | Image searching method |
CN102842042A (en) * | 2011-06-07 | 2012-12-26 | 埃森哲环球服务有限公司 | Biometric authentication technology |
CN103312730A (en) * | 2012-03-07 | 2013-09-18 | 腾讯科技(深圳)有限公司 | Picture storage method, server, terminal and picture storage system |
CN103678661A (en) * | 2013-12-24 | 2014-03-26 | 中国联合网络通信集团有限公司 | Image searching method and terminal |
-
2014
- 2014-05-30 CN CN201410239452.5A patent/CN105243060B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101859326A (en) * | 2010-06-09 | 2010-10-13 | 南京大学 | Image searching method |
CN102842042A (en) * | 2011-06-07 | 2012-12-26 | 埃森哲环球服务有限公司 | Biometric authentication technology |
CN103312730A (en) * | 2012-03-07 | 2013-09-18 | 腾讯科技(深圳)有限公司 | Picture storage method, server, terminal and picture storage system |
CN103678661A (en) * | 2013-12-24 | 2014-03-26 | 中国联合网络通信集团有限公司 | Image searching method and terminal |
Non-Patent Citations (1)
Title |
---|
基于内容的图象检索系统;唐立军等;《计算机应用研究》;20011231(第7期);41—45 * |
Also Published As
Publication number | Publication date |
---|---|
CN105243060A (en) | 2016-01-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105243060B (en) | A kind of method and device of retrieving image | |
US10606887B2 (en) | Providing relevant video scenes in response to a video search query | |
US20210103779A1 (en) | Mobile image search system | |
US10885100B2 (en) | Thumbnail-based image sharing method and terminal | |
US10268703B1 (en) | System and method for associating images with semantic entities | |
US7917514B2 (en) | Visual and multi-dimensional search | |
KR101377389B1 (en) | Visual and multi-dimensional search | |
US20180365489A1 (en) | Automatically organizing images | |
US20170279840A1 (en) | Automated event id field analysis on heterogeneous logs | |
US9652534B1 (en) | Video-based search engine | |
WO2015101289A1 (en) | Image management method, apparatus and system | |
TW201322021A (en) | Image search method and image search apparatus | |
JP2017182628A (en) | Augmented reality user interface application device and control method | |
US11810177B2 (en) | Clothing collocation | |
CN104850600B (en) | A kind of method and apparatus for searching for the picture comprising face | |
US20170255691A1 (en) | Information processing system, information processing method, and program | |
US20130332824A1 (en) | Embedded font processing method and device | |
JP2006350655A (en) | Method, device and program for classifying and presenting result of web image search, and storage medium with the program stored therein | |
KR20150101846A (en) | Image classification service system based on a sketch user equipment, service equipment, service method based on sketch and computer readable medium having computer program recorded therefor | |
CN103886031A (en) | Method and equipment for image browsing | |
US10956493B2 (en) | Database comparison operation to identify an object | |
US8943074B2 (en) | Generating and displaying an image tour in response to a user search query | |
JP2003281094A (en) | Management system, method for extracting unauthorized candidate access data by the management system, and program for extracting unauthorized candidate access data | |
CN104484381B (en) | For searching for the method and system of multiple strings | |
CN112948328A (en) | Retrieval method, device, equipment and medium of log data |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | 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 |