CN105095215B - Information acquisition device, method and server - Google Patents
Information acquisition device, method and server Download PDFInfo
- Publication number
- CN105095215B CN105095215B CN201410163389.1A CN201410163389A CN105095215B CN 105095215 B CN105095215 B CN 105095215B CN 201410163389 A CN201410163389 A CN 201410163389A CN 105095215 B CN105095215 B CN 105095215B
- Authority
- CN
- China
- Prior art keywords
- images
- recognized
- database
- cluster
- cluster frequency
- 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
It includes: training unit that the embodiment of the present invention, which provides a kind of information acquisition device, method and server, the device, and multiple database images in tranining database obtain the index in relation to cluster frequency；Receiving unit receives images to be recognized；First extraction unit extracts the feature of images to be recognized；First computing unit is distributed according to the cluster frequency of the feature calculation images to be recognized of extraction；Second computing unit is distributed and is indexed according to cluster frequency, calculates images to be recognized at a distance from the cluster frequency distribution of database images；Recognition unit identifies image similar with images to be recognized according to this distance, and obtains the relevant information of images to be recognized；Transmission unit sends the relevant information of images to be recognized.It by the way that the image in place to be identified to as images to be recognized and obtained relevant information, does not need to know the keyword about place, it will be able to fast and accurately obtain information relevant to the place.
Description
Technical field
The present invention relates to field of communication technology more particularly to a kind of information acquisition devices, method and server.
Background technique
With the fast development of smart machine and mobile network, in recent years, location-based service becomes a research
Hot spot.For example, the information in some place (such as building or sight spot) is obtained if necessary when other places travels or goes on business,
The existing method for obtaining relevant information is searched on the internet using keywords such as the titles at the building or sight spot
Rope.
It should be noted that the above description of the technical background be intended merely to it is convenient to technical solution of the present invention carry out it is clear,
Complete explanation, and facilitate the understanding of those skilled in the art and illustrate.Cannot merely because these schemes of the invention
Background technology part is expounded and thinks that above-mentioned technical proposal is known to those skilled in the art.
Summary of the invention
It is above-mentioned existing obtain relevant information method, when due to searching on the internet need input for example building or
The keyword (such as title) in the places such as sight spot is not knowing the title of the building or sight spot or be not available native language
In the case that speech spells out the title, then relevant information can not be obtained by the means scanned in internet.
The embodiment of the present invention provides a kind of information acquisition device, method and server, by using the image in place as
Images to be recognized is identified and is obtained relevant information, does not need to know the keyword about place, it will be able to fast and accurately
Acquisition information relevant to the place.
According to a first aspect of the embodiments of the present invention, a kind of information acquisition device is provided, described device includes:
Training unit, the training unit obtain related cluster frequency for multiple database images in tranining database
The index of rate, wherein the set of the cluster frequency for indexing database images that correspond to each cluster, each；It receives
Unit, the receiving unit is for receiving images to be recognized；First extraction unit, first extraction unit are described for extracting
The feature of images to be recognized；First computing unit, first computing unit are used for according to the feature calculation of extraction wait know
The cluster frequency of other image is distributed；Second computing unit, second computing unit are used for according to the poly- of the images to be recognized
Quefrency distribution, the multiple database images cluster frequency distribution and the index, calculate the images to be recognized with
The distance of the cluster frequency distribution of the multiple database images；Recognition unit, the recognition unit are used for according to described wait know
Other image identifies figure similar with the images to be recognized at a distance from the cluster frequency distribution of the multiple database images
Picture, and obtain the relevant information of the images to be recognized；Transmission unit, the transmission unit is for sending the images to be recognized
Relevant information.
According to a second aspect of the embodiments of the present invention, a kind of server is provided, the server includes real according to the present invention
Apply information acquisition device described in the first aspect of example.
According to a third aspect of the embodiments of the present invention, a kind of information acquisition method is provided, which comprises receive wait know
Other image；Extract the feature of the images to be recognized；The cluster frequency of the images to be recognized according to the feature calculation of extraction point
Cloth；According to the distribution of the cluster frequency of the images to be recognized, the cluster frequency distribution of multiple database images and related cluster
The index of frequency calculates the images to be recognized at a distance from the cluster frequency distribution of the multiple database images；Wherein, institute
Index is stated to obtain by the multiple images in tranining database, it is described to index database that correspond to each cluster, each
The set of the cluster frequency of image；According to the cluster frequency of the images to be recognized and the multiple database images be distributed away from
From identifying image similar with the images to be recognized, and obtain the relevant information of the images to be recognized；Send it is described to
Identify the relevant information of image.
The beneficial effects of the present invention are: by the way that the image in place is identified as images to be recognized and obtains correlation
Information does not need to know the keyword about place, it will be able to fast and accurately obtain information relevant to the place.
Referring to following description and accompanying drawings, only certain exemplary embodiments of this invention is disclosed in detail, specifies original of the invention
Reason can be in a manner of adopted.It should be understood that embodiments of the present invention are not so limited in range.In appended power
In the range of the spirit and terms that benefit requires, embodiments of the present invention include many changes, modifications and are equal.
The feature for describing and/or showing for a kind of embodiment can be in a manner of same or similar one or more
It uses in a other embodiment, is combined with the feature in other embodiment, or the feature in substitution other embodiment.
It should be emphasized that term "comprises/comprising" refers to the presence of feature, one integral piece, step or component when using herein, but simultaneously
It is not excluded for the presence or additional of one or more other features, one integral piece, step or component.
Detailed description of the invention
Included attached drawing is used to provide to be further understood from the embodiment of the present invention, and which constitute one of specification
Point, for illustrating embodiments of the present invention, and come together to illustrate the principle of the present invention with verbal description.Under it should be evident that
Attached drawing in the description of face is only some embodiments of the present invention, for those of ordinary skill in the art, is not paying wound
Under the premise of the property made is laborious, it is also possible to obtain other drawings based on these drawings.In the accompanying drawings:
Fig. 1 is the structural schematic diagram of the information acquisition device of the embodiment of the present invention 1；
Fig. 2 is the structural schematic diagram of the training unit of the embodiment of the present invention 1；
Fig. 3 is that the training unit of the embodiment of the present invention 1 is trained the method to obtain above-mentioned index to database images
Flow chart；
Fig. 4 is the structural schematic diagram of the recognition unit of the embodiment of the present invention 1；
Fig. 5 is the identification image of the embodiment of the present invention 1 and the method flow diagram for obtaining relevant information；
Fig. 6 is the schematic block diagram that the system of the server of the embodiment of the present invention 2 is constituted；
Fig. 7 is the structural schematic diagram of the communication system of the embodiment of the present invention 3；
Fig. 8 is the flow chart of the information acquisition method of the embodiment of the present invention 4.
Specific embodiment
Referring to attached drawing, by following specification, aforementioned and other feature of the invention be will be apparent.In specification
In attached drawing, only certain exemplary embodiments of this invention is specifically disclosed, which show the portions that can wherein use principle of the invention
Divide embodiment, it will thus be appreciated that the present invention is not limited to described embodiments, on the contrary, the present invention includes falling into appended power
Whole modifications, modification and equivalent in the range of benefit requirement.
Embodiment 1
Fig. 1 is the structural schematic diagram of the information acquisition device of the embodiment of the present invention 1.As shown in Figure 1, the device 100 includes:
Training unit 101, receiving unit 102, the first extraction unit 103, the first computing unit 104, the second computing unit 105, identification
Unit 106 and transmission unit 107, wherein
Training unit 101 obtains the index in relation to cluster frequency for multiple database images in tranining database,
In, which corresponds to the set of the cluster frequency of each cluster, each database images；
Receiving unit 102 is for receiving images to be recognized；
First extraction unit 103 is used to extract the feature of the images to be recognized；
First computing unit 104 is used for the cluster frequency distribution according to the feature calculation of the extraction images to be recognized；Second
Computing unit 105 is used to be distributed according to the cluster frequency of the distribution of the cluster frequency of the images to be recognized, multiple database images
And the index, the images to be recognized is calculated at a distance from the cluster frequency distribution of multiple database images；
What recognition unit Unit 106 was used to be distributed according to the cluster frequency of the images to be recognized and multiple database images
Distance identifies image similar with the images to be recognized, and obtains the relevant information of the images to be recognized；
Transmission unit 107 is used to send the relevant information of the images to be recognized.
In the present embodiment, training unit 101 is used for tranining database image, wherein the database diagram seems to deposit in advance
It stores up in the database, which can use existing any method and establish.For example, collecting global every country and ground
The famous buildings in area and the image at sight spot, and these images are classified and numbered.But the embodiment of the present invention is not limited to
This method.In addition, the database can store in the present apparatus, can also separately store, the embodiment of the present invention is not to data
The storage location in library is limited.
In the present embodiment, existing any method can be used to be trained database images for training unit 101.With
Under the training unit of the embodiment of the present invention is trained database images with obtain the method for above-mentioned index carry out it is exemplary
Explanation.
Fig. 2 is the structural schematic diagram of the training unit of the present embodiment.As shown in Fig. 2, the training unit 101 includes: second
Extraction unit 201, third computing unit 202 and the 4th computing unit 203, wherein
Second extraction unit 201 is used to extract the feature of each database images；
Third computing unit 202 is used to be distributed according to the cluster frequency of each database images of feature calculation of extraction；
4th computing unit 203 is used to be distributed according to the cluster frequency of each database images, calculates above-mentioned index.
Fig. 3 is that the training unit of the present embodiment is trained database images to obtain the method flow of above-mentioned index
Figure.As shown in figure 3, this method comprises:
Step 301: extracting the feature of each database images；
Step 302: being distributed according to the cluster frequency of each database images of the feature calculation of extraction；
Step 303: being distributed according to the cluster frequency of each database images, calculate above-mentioned index.
In the present embodiment, the feature that existing any method extracts each database images can be used.For example, can
Use gradient orientation histogram method (Histogram of Oriented Gradient, HOG), scale invariant feature transformation approach
(Scale-invariant Feature Transform, SIFT), fast robust characteristic method (Speeded Up Robust
Feature, SURF) etc. existing methods.But the embodiment of the present invention is not limited to the above method.
In the present embodiment, the characteristic of certain a part of each character representation image of database images, each feature
Position in the picture can be determined by cluster sampling or point of interest.After extracting feature, each database images can
With with a series of character representation, wherein the vector representation of each feature regular length.For example, for database images Xi,
The feature extracted can be indicated with following formula (1):
Wherein, N indicates database images XiCharacteristic Number, i, N are positive integer.
In the present embodiment, after the feature for extracting each database images, existing any method can be used
Calculate the cluster frequency distribution of each database images.For example, can be used C- mean value (C-Means) clustering algorithm by database diagram
All features of picture are clustered, and respectively obtain cluster frequency corresponding to the feature of each database images with this.
For example, each feature of each database images is classified as K cluster using C- means clustering algorithm, can use down
Formula (2) indicates:
C={ C1,…,CK} (2)
Wherein, K is positive integer.
For database images XiN number of feature, using existing any method sorted out to above-mentioned K cluster
In, it can be indicated with following formula (3) and (4):
Fi→Gi={ C2,C1,CK,…,C5} (4)
Wherein, featureBelong to cluster C2, featureBelong to cluster C1, featureBelong to cluster CK... ..., featureBelong to
Cluster C5。
In the present embodiment, after being sorted out each feature of each database images, each number can be calculated
It is distributed according to the cluster frequency of library image.Below to the method for the cluster frequency distribution of each database images of calculating of the present embodiment
Illustratively illustrated.
Such as set 5 clusters, it may be assumed that C={ C1,C2,C3,C4,C5, database images XiWith 10 features, according to
Above formula (4), database images XiThe clusters of 10 features be respectively: Gi={ 2,1,3,5,2,1,1,1,3,3 }, therefore, data
Library image XiFeature in have 4 features belong to cluster C1, there are 2 features to belong to cluster C2, there are 3 features to belong to cluster C3, have
0 feature belongs to cluster C4, there is 1 feature to belong to cluster C5.So, database images XiCluster frequency distribution can calculate
It is as follows: Hi={ 4,2,3,0,1 }/10={ 0.4,0.2,0.3,0,0.1 }.
In the present embodiment, the above method is used to calculate cluster frequency distribution each database images, to obtain
Obtain the cluster frequency distribution of each database images.After obtaining the cluster frequency distribution of each image, cluster frequency is utilized
Rate distribution, calculates the index in relation to cluster frequency, wherein the index corresponds to each cluster, each database images
The set of cluster frequency.The method of the calculating of the present embodiment index is illustratively illustrated below.
For example, equally setting 5 clusters, it may be assumed that C={ C1,C2,C3,C4,C5}.Such as one share 4 database images,
And the cluster frequency distribution of each database images is as follows with following formula (5):
So, the index in relation to cluster frequency can be expressed as follows with following formula (6):
To obtain the set of the cluster frequency of corresponding to each cluster, each database images, that is, obtain
State index.
In the present embodiment, receiving unit 102 can be used existing any method and receive images to be recognized.For example,
The images to be recognized can be received through wireless communication.Wherein, the images to be recognized can be user shooting about certain
The image in a place, for example, the images to be recognized is the image at some building or sight spot.
In the present embodiment, the first extraction unit 103 can be used existing any method and extract the images to be recognized
Feature.For example, this method can be identical as the above method of extraction feature that training unit 101 uses, details are not described herein again.
In the present embodiment, the first computing unit 104 can be used existing any method and calculate the images to be recognized
Cluster frequency distribution.For example, the above method phase for calculating cluster frequency distribution that this method can be used with training unit 101
Together, details are not described herein again.
In the present embodiment, in the cluster frequency distribution that obtains the images to be recognized and each database images and above-mentioned
After index in relation to cluster frequency, the second computing unit 105 is according to the cluster of the images to be recognized and multiple database images
Frequency distribution and the index calculate the images to be recognized at a distance from the cluster frequency distribution of each database images.Pass through
The images to be recognized is calculated at a distance from the cluster frequency distribution of each database images using above-mentioned index, can effectively be mentioned
High calculating speed and image recognition rate, to improve the speed for obtaining relevant information.
Wherein, existing any method can be used to be calculated.The figure to be identified is utilized to the embodiment of the present invention below
The method that picture and the distribution of the cluster frequency of multiple database images and the index calculate above-mentioned distance is illustratively illustrated.
It is, for example, possible to use the cluster frequency that following formula (7) calculate images to be recognized and each database images be distributed away from
From:
Di=D0-Fi-FQ-w|Fi-FQ| (7)
Wherein, DiExpression images to be recognized is at a distance from the cluster frequency distribution of i-th of database images, D0Indicate initial
Distance, FiIndicate the cluster frequency of i-th of database images in above-mentioned index, FQIndicate the cluster frequency of images to be recognized, w
Indicate weight, w > 0 is set according to actual needs.
For example, initial distance D0=2.0, weight w=1, database images have 4, are X respectively1、X2、X3And X4, each number
It is indicated according to the cluster frequency distribution of library image with above formula (5), the cluster frequency distribution of images to be recognized are as follows: HQ=0.4,0.0,
0.6,0.0,0.0 } it, also, using the index that above formula (6) indicate is calculated.But the embodiment of the present invention is not limited to above-mentioned value.
For clustering C1To C5, the poly- of images to be recognized and each database images is successively calculated according to above formula (6) and (7)
The distance of quefrency distribution, also, in calculating process, by the identification image obtained for the cluster calculation and each database
The distance of the cluster frequency distribution of image as calculate next cluster it is each apart from when the initial distance that uses, specific calculating
Process is as follows:
For clustering C1, according to above formula (6) and (7), images to be recognized and database images X1、X2、X3Distance count respectively
It calculates as follows:
Dist_1=2.0-0.4-0.4+ | 0.4-0.4 |=1.2
Dist_2=2.0-0.4-0.1+ | 0.4-0.1 |=1.8
Dist_3=2.0-0.4-0.2+ | 0.4-0.2 |=1.6
And due to database images X4Middle cluster C1Frequency be 0, therefore, images to be recognized and database images X4Away from
From remaining as initial distance, i.e., 2.0.
For clustering C2, will be for clustering C1The images to be recognized being calculated conduct at a distance from each database images
Initial distance, and due in images to be recognized for cluster C2Corresponding frequencies be 0, therefore initial distance is not updated,
Images to be recognized and database images X1、X2、X3、X4Distance be 1.2,1.8,1.6 and 2.0 respectively.
For clustering C3, will be for clustering C2The images to be recognized being calculated conduct at a distance from each database images
Initial distance, according to above formula (6) and (7), images to be recognized and database images X1、X2、X3、X4Distance be respectively calculated as follows:
Dist_1=1.2-0.6-0.3+ | 0.6-0.3 |=0.9
Dist_2=1.8-0.6-0.8+ | 0.6-0.8 |=0.6
Dist_3=1.6-0.6-0.8+ | 0.6-0.8 |=0.4
Dist_4=2.0-0.6-0.2+ | 0.6-0.2 |=1.6
For clustering C4, will be for clustering C3The images to be recognized being calculated conduct at a distance from each database images
Initial distance, and due in images to be recognized for cluster C4Corresponding frequencies be 0, therefore initial distance is not updated,
Images to be recognized and database images X1、X2、X3、X4Distance be 0.9,0.6,0.4 and 1.6 respectively.
For clustering C5, will be for clustering C4The images to be recognized being calculated conduct at a distance from each database images
Initial distance, likewise, due in images to be recognized for cluster C5Corresponding frequencies be 0, therefore not to initial distance carry out
It updates, the images to be recognized and database images X finally obtained1、X2、X3、X4Distance be 0.9,0.6,0.4 and 1.6 respectively.
In the present embodiment, recognition unit 106 is according to the cluster frequency of the images to be recognized and each database images point
The distance of cloth identifies image similar with the images to be recognized, and obtains the relevant information of the images to be recognized.Wherein, may be used
Image similar with the images to be recognized is identified according to above-mentioned distance using existing any method and obtains figure to be identified
The relevant information of picture.It is identified for example, k-NN ballot method can be used and obtains relevant information.Below to the embodiment of the present invention
It identifies image and the method for obtaining relevant information is illustratively illustrated.
Fig. 4 is the structural schematic diagram of the recognition unit of the present embodiment.As described in Figure 4, recognition unit 106 includes: the second mistake
Filter unit 401 and third filter element 402, wherein
Second filter element 401 be used to be distributed according to the cluster frequency of the images to be recognized and each database images away from
From being filtered to each database images, obtain similar with images to be recognized image；
Third filter element 402 is for being filtered image similar with the images to be recognized, obtaining and being somebody's turn to do wait know
The most image category of frequency of occurrence in the similar image of other image, and as the image category of the images to be recognized.
Fig. 5 is the identification image of the present embodiment and the method flow diagram for obtaining relevant information.As shown in figure 5, this method packet
It includes:
Step 501: according to the images to be recognized at a distance from the cluster frequency distribution of each database images, to each number
It is filtered according to library image, obtains image similar with the images to be recognized；
Step 502: image similar with the images to be recognized being filtered, figure similar with the images to be recognized is obtained
The most image category of frequency of occurrence as in, and as the image category of the images to be recognized.
In the present embodiment, such as above-mentioned distance can be less than to the k number of some preset threshold value according to library image work
For image similar with the images to be recognized, then in the k number according to determining the most image category of frequency of occurrence in the image of library,
Image category as the images to be recognized.
In the present embodiment, the image category for obtaining the images to be recognized can also be from data according to the image category
In library obtain respective image other information, for example, in the database images place title.Using the place name, may be used also
Further to search for more information relevant to the place.Wherein, which can carry out in the present apparatus, can also be mutual
It is carried out in networking, the embodiment of the present invention does not limit the position of search.
In the present embodiment, after obtaining the relevant information of images to be recognized, transmission unit 107 is believed for the correlation
Breath, wherein existing any method can be used to send the information.For example, the mode of wireless communication can be used to send the information.
The embodiment of the present invention does not limit the method for sending the information.
In the present embodiment, which can also include: the first filter element 108, be used for according to the images to be recognized
Location information the image in the database is filtered, to obtain multiple database images for calculating above-mentioned distance.
Wherein, the first filter element 108 is selectable unit (SU), is indicated by the dashed box in Fig. 1.
In the present embodiment, which can receive together with the images to be recognized, can also receive respectively, the present invention
Embodiment is limited not to this.Being filtered using the location information to database images can be used existing any side
Method.For example, using the close database images of the position coordinates of position coordinates and images to be recognized as filtered image.
In the present embodiment, filtered database images are used to calculate distance, are capable of the range of downscaled images identification,
Further increase the recognition speed of image.
In the present embodiment, which can also include: cut cells 109, be used for calculated to training unit 101
The cluster frequency for being greater than preset threshold value in the cluster frequency distribution of each database images carries out shear treatment, and to cutting
Cluster frequency distribution after cutting is normalized, and is used to calculate above-mentioned rope for the cluster frequency distribution after normalized
Draw.Wherein, cut cells 109 are selectable unit (SU), are indicated by the dashed box in Fig. 1.
In the present embodiment, which can set according to actual needs.For example, can be by the threshold value
It is set as twice of the average value of the cluster frequency of the database images.The embodiment of the present invention does not limit the value of the threshold value
System.
For example, being distributed H={ 0.2,0.2,0.5,0,0.1 } for cluster frequency, wherein the average value of cluster frequency is
0.2, then the part that the cluster frequency in cluster frequency distribution is more than 0.4 is carried out shear treatment, i.e., it is set to 0.4 by 0.5,
Obtain the result of shear treatment: H={ 0.2,0.2,0.4,0,0.1 }.Then, which is normalized, is returned
Result after one change: H={ 0.22,0.22,0.44,0,0.11 }.
In the present embodiment, the part excessively high for cluster frequency is sheared and is normalized, and will processing
Result afterwards is used to calculate the cluster frequency distribution of database images, can further increase the accuracy of image recognition.
As can be seen from the above embodiments, by being identified the image in place as images to be recognized and obtaining related letter
Breath, does not need to know the keyword about place, it will be able to fast and accurately obtain information relevant to the place.
Embodiment 2
The embodiment of the present invention provides a kind of server, which includes information acquisition device as described in Example 1.
Fig. 6 is the schematic block diagram that the system of the server 600 of the embodiment of the present invention 2 is constituted.As shown in fig. 6, server
600 may include central processing unit 601 and memory 602；Memory 602 is coupled to central processing unit 601.The figure is exemplary
's；Other kinds of structure can also be used, to supplement or replace the structure, to realize telecommunications functions or other function.
As shown in fig. 6, the server 600 can also include: communication module 603, input unit 604, display 605, electricity
Source 606.
In one embodiment, the function of information acquisition device can be integrated into central processing unit 601.Wherein,
Central processing unit 601 can be configured as: receive images to be recognized；Extract the feature of the images to be recognized；According to extraction
The cluster frequency of images to be recognized described in feature calculation is distributed；According to the distribution of the cluster frequency of the images to be recognized, multiple numbers
According to the cluster frequency distribution of library image and the index in relation to cluster frequency, the images to be recognized and the multiple data are calculated
The distance of the cluster frequency distribution of library image；Wherein, the index is obtained by the multiple images in tranining database, described
Index corresponds to the set of the cluster frequency of each cluster, each database images；According to the images to be recognized and institute
The distance for stating the cluster frequency distribution of multiple database images identifies image similar with the images to be recognized, and obtains institute
State the relevant information of images to be recognized；Send the relevant information of the images to be recognized.
Wherein, described to be identified at a distance from the cluster frequency distribution of each database images according to the images to be recognized
Image similar with the images to be recognized and the relevant information for obtaining the images to be recognized, comprising: according to described to be identified
At a distance from the cluster frequency distribution of image and each database images, each database images are filtered, obtain with it is described
The similar image of images to be recognized；Image similar with the images to be recognized is filtered, is obtained and the figure to be identified
As the most image category of frequency of occurrence in similar image, and as the image category of the images to be recognized.
Wherein, central processing unit 601 can be additionally configured to: according to the location information of the images to be recognized, to described
Image in database is filtered, to obtain the multiple database images for calculating the distance.
Wherein, central processing unit 601 can be additionally configured to: multiple database images in tranining database, obtain institute
State index；Wherein, multiple database images in tranining database, obtain the index, comprising: extract each database images
Feature；It is distributed according to the cluster frequency of each database images of the feature calculation of extraction；According to the poly- of each database images
Quefrency distribution, calculates the index.
Wherein, central processing unit 601 can be additionally configured to: to being greater than in the distribution of the cluster frequencies of each database images
The cluster frequency of preset threshold value carries out shear treatment, and the cluster frequency distribution after shearing is normalized.
In another embodiment, information acquisition device can with 601 separate configuration of central processing unit, such as can will
Information acquisition device is configured to the chip connecting with central processing unit 601, is realized at image by the control of central processing unit
Manage the function of device.
Server 600 is also not necessary to include all components shown in Fig. 6 in the present embodiment
As shown in fig. 6, central processing unit 601 be otherwise referred to as controller or operational controls, may include microprocessor or
Other processor devices and/or logic device, central processing unit 601 receive all parts of input and control server 600
Operation.
Memory 602, such as can be buffer, flash memory, hard disk driver, removable medium, volatile memory, non-volatile
One of memory or other appropriate devices or more.Above-mentioned information related with failure can be stored, can additionally be stored
Execute program for information about.And the program of the memory 602 storage can be performed in central processing unit 601, to realize information
Storage or processing etc..The function of other component with it is existing similar, details are not described herein again.Each component of server 600 can pass through
Specialized hardware, firmware, software or its in conjunction with realizing, be made without departing from the scope of the present invention.
As can be seen from the above embodiments, by being identified the image in place as images to be recognized and obtaining related letter
Breath, does not need to know the keyword about place, it will be able to fast and accurately obtain information relevant to the place.
Embodiment 3
Fig. 7 is the structural schematic diagram of the communication system of the embodiment of the present invention 3.As shown in fig. 7, the communication system 700 includes
User equipment 701 and server 702, wherein the structure and function of server is identical as the record in embodiment 2, herein not
It repeats again.
In the present embodiment, user equipment 701 shoots image in some place, and is sent to as images to be recognized
The location information of shooting location can also be sent respectively to server 702 together with image or with image by server 702,
Server 702 is identifying image similar with the images to be recognized, and after obtaining the relevant information of images to be recognized, by the phase
It closes information and is sent to user equipment 701.
In the present embodiment, which is, for example, portable radio communication equipment, including all such as mobile electricity
The equipment of words, smart phone, personal digital assistant (PDA), smart phone etc..It is taken the photograph in addition, the user equipment can also be to have
As the equipment of function, for example, camera, video camera etc..The embodiment of the present invention does not limit the type of user equipment
System.
As can be seen from the above embodiments, by being identified the image in place as images to be recognized and obtaining related letter
Breath, does not need to know the keyword about place, it will be able to fast and accurately obtain information relevant to the place.
Embodiment 4
Fig. 8 is the flow chart of the information acquisition method of the embodiment of the present invention 4, the information acquisition device corresponding to embodiment 1.
As shown in figure 8, this method comprises:
Step 801: receiving images to be recognized；
Step 802: extracting the feature of the images to be recognized；
Step 803: the cluster frequency distribution of the images to be recognized according to the feature calculation of extraction；
Step 804: according to the cluster frequency of the images to be recognized distribution, multiple database images cluster frequency distribution with
And the index in relation to cluster frequency, the images to be recognized is calculated at a distance from the cluster frequency distribution of multiple database images；Its
In, which is obtained by the multiple images in tranining database, which corresponds to each cluster, each database
The set of the cluster frequency of image；
Step 805: according to identified at a distance from the distribution of the cluster frequency of the images to be recognized and multiple database images with
The similar image of the images to be recognized, and obtain the relevant information of the images to be recognized；
Step 806: sending the relevant information of the images to be recognized.
In the present embodiment, this method can also include: according to the location information of the images to be recognized, to the data
Image in library is filtered, to obtain multiple database images for calculating the distance.
In the present embodiment, this method can also include: multiple database images in tranining database, obtain the rope
Draw；Wherein, multiple database images in tranining database, obtain the index, comprising: extract the spy of each database images
Sign；It is distributed according to the cluster frequency of each database images of the feature calculation of extraction；According to the cluster of each database images frequency
Rate distribution, calculates the index.
In the present embodiment, this method can also include: to be greater than in advance in the cluster frequency distribution to each database images
The cluster frequency of the threshold value first set carries out shear treatment, and the cluster frequency distribution after shearing is normalized.
In the present embodiment, the method for tranining database image, the method for carrying out shear treatment and normalized, reception
The method of images to be recognized, the side that the image in the database is filtered according to the location information of the images to be recognized
Method, the method for extracting feature, the method for calculating cluster frequency distribution, the cluster for calculating images to be recognized and each database images
The method of the distance of frequency distribution identifies image similar with images to be recognized according to this distance, and obtains images to be recognized
Relevant information method and send relevant information method it is identical as the record of embodiment 1, details are not described herein again.
As can be seen from the above embodiments, by being identified the image in place as images to be recognized and obtaining related letter
Breath, does not need to know the keyword about place, it will be able to fast and accurately obtain information relevant to the place.
The embodiment of the present invention also provides a kind of computer-readable program, wherein holding when in information acquisition device or server
When row described program, it is as described in example 4 that described program executes computer in the information acquisition device or server
Information acquisition method.
The embodiment of the present invention also provides a kind of storage medium for being stored with computer-readable program, wherein the computer can
Reader makes computer execute information acquisition method as described in example 4 in information acquisition device or server.
The device and method more than present invention can be by hardware realization, can also be by combination of hardware software realization.The present invention
It is related to such computer-readable program, when the program is performed by logical block, the logical block can be made to realize above
The device or component parts, or the logical block is made to realize various method or steps described above.The invention further relates to
For storing the storage medium of procedure above, such as hard disk, disk, CD, DVD, flash memory.
Combining specific embodiment above, invention has been described, it will be appreciated by those skilled in the art that this
A little descriptions are all exemplary, and are not limiting the scope of the invention.Those skilled in the art can be according to the present invention
Spirit and principle various variants and modifications are made to the present invention, these variants and modifications are also within the scope of the invention.
About the embodiment including above embodiments, following note is also disclosed:
It is attached 1, a kind of information acquisition device, described device includes:
Training unit, the training unit obtain related cluster frequency for multiple database images in tranining database
The index of rate, wherein the set of the cluster frequency for indexing database images that correspond to each cluster, each；It receives
Unit, the receiving unit is for receiving images to be recognized；
First extraction unit, first extraction unit are used to extract the feature of the images to be recognized；
First computing unit, first computing unit is for the poly- of the images to be recognized according to the feature calculation of extraction
Quefrency distribution；
Second computing unit, second computing unit are used for according to the distribution of the cluster frequency of the images to be recognized, institute
The cluster frequency for stating multiple database images is distributed and the index, calculates the images to be recognized and the multiple database
The distance of the cluster frequency distribution of image；
Recognition unit, the recognition unit are used for the cluster according to the images to be recognized and the multiple database images
The distance of frequency distribution identifies image similar with the images to be recognized, and obtains the related letter of the images to be recognized
Breath；
Transmission unit, the transmission unit are used to send the relevant information of the images to be recognized.
Note 2, the device according to note 1, wherein described device further include:
First filter element, first filter element is used for the location information according to the images to be recognized, to described
Image in database is filtered, to obtain the multiple database images for calculating the distance.
Note 3, the device according to note 1, wherein the training unit includes:
Second extraction unit, second extraction unit are used to extract the feature of each database images；
Third computing unit, the third computing unit are used for according to the poly- of each database images of feature calculation of extraction
Quefrency distribution；
4th computing unit, the 4th computing unit are used to be distributed according to the cluster frequency of each database images, meter
Calculate the index.
It is attached 4, according to device described in note 3, wherein described device further include:
Cut cells, the cut cells are used to be greater than in the cluster frequency distribution to each database images and preset
Threshold value cluster frequency carry out shear treatment, and to after shearing cluster frequency distribution be normalized.
Note 5, the device according to note 1, wherein the recognition unit includes:
Second filter element, the second filter element are used for the cluster according to the images to be recognized and each database images
The distance of frequency distribution is filtered each database images, obtains image similar with the images to be recognized；
Third filter element, the third filter element was for carrying out image similar with the images to be recognized
Filter obtains the image category that frequency of occurrence is most in image similar with the images to be recognized, and as described wait know
The image category of other image.
Note 6, the device according to note 1, wherein second computing unit using following formula (1) calculate it is described to
Identify image at a distance from the cluster frequency distribution of each database images:
Di=D0-Fi-FQ-w|Fi-FQ| (1)
Wherein, DiExpression images to be recognized is at a distance from the cluster frequency distribution of i-th of database images, D0Indicate initial
Distance, FiIndicate the cluster frequency of i-th of database images in the index, FQIndicate the cluster frequency of images to be recognized；w
Indicate weight, w > 0 is set according to actual needs.
7, a kind of server are attached, including such as 1 to 6 described in any item information acquisition devices of note.
It is attached 8, a kind of information acquisition method, which comprises
Receive images to be recognized；
Extract the feature of the images to be recognized；
The cluster frequency of the images to be recognized according to the feature calculation of extraction is distributed；
It is distributed according to the cluster frequency of the distribution of the cluster frequency of the images to be recognized, multiple database images and related
The index of cluster frequency calculates the images to be recognized at a distance from the cluster frequency distribution of the multiple database images；Its
In, the index is obtained by the multiple images in tranining database, described to index number that correspond to each cluster, each
According to the set of the cluster frequency of library image；
According to being identified at a distance from the distribution of the cluster frequency of the images to be recognized and the multiple database images and institute
The similar image of images to be recognized is stated, and obtains the relevant information of the images to be recognized；
Send the relevant information of the images to be recognized.
It is attached 9, according to method described in note 8, wherein the method also includes:
According to the location information of the images to be recognized, the image in the database is filtered, to be used for
Calculate the multiple database images of the distance.
It is attached 10, according to method described in note 8, wherein the method also includes: multiple data in tranining database
Library image obtains the index；
Wherein, multiple database images in tranining database, obtain the index, comprising:
Extract the feature of each database images；
It is distributed according to the cluster frequency of each database images of the feature calculation of extraction；
It is distributed according to the cluster frequency of each database images, calculates the index.
Note 11, the method according to note 10, wherein the method also includes:
The cluster frequency for being greater than preset threshold value in the cluster frequency distribution of each database images is sheared
Processing, and the cluster frequency distribution after shearing is normalized.
It is attached 12, according to method described in note 8, wherein described according to the images to be recognized and the multiple data
The distance of the cluster frequency distribution of library image identifies image similar with the images to be recognized and obtains the figure to be identified
The relevant information of picture, comprising:
According to the images to be recognized at a distance from the cluster frequency distribution of each database images, to each database diagram
As being filtered, image similar with the images to be recognized is obtained；
Image similar with the images to be recognized is filtered, is obtained in image similar with the images to be recognized
The most image category of frequency of occurrence, and as the image category of the images to be recognized.
Note 13, according to method described in note 8, wherein calculate the images to be recognized and described more using following formula (1)
The distance of the cluster frequency distribution of a database images:
Di=D0-Fi-FQ-w|Fi-FQ| (1)
Wherein, DiExpression images to be recognized is at a distance from the cluster frequency distribution of i-th of database images, D0Indicate initial
Distance, FiIndicate the cluster frequency of i-th of database images in the index, FQIndicate the cluster frequency of images to be recognized, w
Indicate weight, w > 0 is set according to actual needs.
Claims (9)
1. a kind of information acquisition device, described device include:
Training unit, the training unit obtain related cluster frequency for multiple database images in tranining database
Index, wherein the set of the cluster frequency for indexing database images that correspond to each cluster, each；
Receiving unit, the receiving unit is for receiving images to be recognized；
First extraction unit, first extraction unit are used to extract the feature of the images to be recognized；
First computing unit, cluster frequency of first computing unit for the images to be recognized according to the feature calculation of extraction
Rate distribution；
Second computing unit, second computing unit are used to be distributed according to the cluster frequency of the images to be recognized, are described more
The cluster frequency of a database images is distributed and the index, calculates the images to be recognized and the multiple database images
Cluster frequency distribution distance；
Recognition unit, the recognition unit are used for the cluster frequency according to the images to be recognized and the multiple database images
The distance of distribution identifies image similar with the images to be recognized, and obtains the relevant information of the images to be recognized；
Transmission unit, the transmission unit are used to send the relevant information of the images to be recognized,
Wherein, second computing unit calculates the cluster of the images to be recognized and each database images using following formula (1)
The distance of frequency distribution:
Di=D0-Fi-FQ-w|Fi-FQ| (1)
Wherein, DiExpression images to be recognized is at a distance from the cluster frequency distribution of i-th of database images, D0Indicate initial distance,
FiIndicate the cluster frequency of i-th of database images in the index, FQIndicate the cluster frequency of images to be recognized；W indicates power
Value, w > 0 is set according to actual needs.
2. the apparatus according to claim 1, wherein described device further include:
First filter element, first filter element is used for the location information according to the images to be recognized, to the data
Image in library is filtered, to obtain the multiple database images for calculating the distance.
3. the apparatus according to claim 1, wherein the training unit includes:
Second extraction unit, second extraction unit are used to extract the feature of each database images；
Third computing unit, the third computing unit are used for the cluster frequency according to each database images of feature calculation of extraction
Rate distribution；
4th computing unit, the 4th computing unit are used to be distributed according to the cluster frequency of each database images, calculate institute
State index.
4. device according to claim 3, wherein described device further include:
Cut cells, the cut cells are used to be greater than preset threshold in the cluster frequency distribution to each database images
The cluster frequency of value carries out shear treatment, and the cluster frequency distribution after shearing is normalized.
5. the apparatus according to claim 1, wherein the recognition unit includes:
Second filter element, the second filter element are used for the cluster frequency according to the images to be recognized and each database images
The distance of distribution is filtered each database images, obtains image similar with the images to be recognized；
Third filter element, the third filter element are obtained for being filtered to image similar with the images to be recognized
The image category that frequency of occurrence is most in image similar with the images to be recognized is obtained, and as the images to be recognized
Image category.
6. a kind of server, including such as information acquisition device described in any one of claim 1 to 5.
7. a kind of information acquisition method, which comprises
Receive images to be recognized；
Extract the feature of the images to be recognized；
The cluster frequency of the images to be recognized according to the feature calculation of extraction is distributed；
According to the distribution of the cluster frequency of the images to be recognized, the cluster frequency distribution of multiple database images and related cluster
The index of frequency calculates the images to be recognized at a distance from the cluster frequency distribution of the multiple database images；Wherein, institute
Index is stated to obtain by the multiple images in tranining database, it is described to index database that correspond to each cluster, each
The set of the cluster frequency of image；
According to identified at a distance from the distribution of the cluster frequency of the images to be recognized and the multiple database images with it is described to
It identifies the similar image of image, and obtains the relevant information of the images to be recognized；
The relevant information of the images to be recognized is sent,
Wherein, the images to be recognized is calculated at a distance from the cluster frequency distribution of each database images using following formula (1):
Di=D0-Fi-FQ-w|Fi-FQ| (1)
Wherein, DiExpression images to be recognized is at a distance from the cluster frequency distribution of i-th of database images, D0Indicate initial distance,
FiIndicate the cluster frequency of i-th of database images in the index, FQIndicate the cluster frequency of images to be recognized；W indicates power
Value, w > 0 is set according to actual needs.
8. according to the method described in claim 7, wherein, the method also includes:
According to the location information of the images to be recognized, the image in the database is filtered, to obtain for calculating
The multiple database images of the distance.
9. according to the method described in claim 7, wherein, the method also includes: multiple database diagrams in tranining database
Picture obtains the index；
Wherein, multiple database images in tranining database, obtain the index, comprising:
Extract the feature of each database images；
It is distributed according to the cluster frequency of each database images of the feature calculation of extraction；
It is distributed according to the cluster frequency of each database images, calculates the index.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410163389.1A CN105095215B (en) | 2014-04-22 | 2014-04-22 | Information acquisition device, method and server |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410163389.1A CN105095215B (en) | 2014-04-22 | 2014-04-22 | Information acquisition device, method and server |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105095215A CN105095215A (en) | 2015-11-25 |
| CN105095215B true CN105095215B (en) | 2019-05-10 |
Family
ID=54575684
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410163389.1A Active CN105095215B (en) | 2014-04-22 | 2014-04-22 | Information acquisition device, method and server |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105095215B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105930499B (en) * | 2016-05-09 | 2019-11-22 | 深圳市数极科技有限公司 | A kind of image searching method and system |
| CN110912933B (en) * | 2019-12-17 | 2021-04-02 | 中国科学院信息工程研究所 | Equipment identification method based on passive measurement |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103324677A (en) * | 2013-05-24 | 2013-09-25 | 西安交通大学 | Hierarchical fast image global positioning system (GPS) position estimation method |
-
2014
- 2014-04-22 CN CN201410163389.1A patent/CN105095215B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103324677A (en) * | 2013-05-24 | 2013-09-25 | 西安交通大学 | Hierarchical fast image global positioning system (GPS) position estimation method |
Non-Patent Citations (1)
| Title |
|---|
| 基于LSI和软加权的视频语义概念检测;张瑞杰 等;《信息工程大学学报》;20130415;第14卷(第2期);第196-201页 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105095215A (en) | 2015-11-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20210103779A1 (en) | Mobile image search system | |
| EP3028184B1 (en) | Method and system for searching images | |
| CN109284729B (en) | Method, device and medium for acquiring face recognition model training data based on video | |
| CN103946838B (en) | Interactive multi-mode image search | |
| Schroth et al. | Mobile visual location recognition | |
| CN110362677B (en) | Text data category identification method and device, storage medium and computer equipment | |
| CN107633066B (en) | Information display method and device, electronic equipment and storage medium | |
| CN109213882B (en) | Picture ordering method and terminal | |
| CN103927387A (en) | Image retrieval system, method and device | |
| CN106933867B (en) | Image query method and device | |
| CN103793697A (en) | Identity labeling method of face images and face identity recognition method of face images | |
| CN103324677B (en) | Hierarchical fast image global positioning system (GPS) position estimation method | |
| CN102591868A (en) | System and method for automatic generation of photograph guide | |
| CN107562742A (en) | A kind of image processing method and device | |
| Guan et al. | On-device mobile landmark recognition using binarized descriptor with multifeature fusion | |
| WO2012145273A1 (en) | Systems and methods for automatically determining an improved view for a visual query in a mobile search | |
| JP2009258953A (en) | Image processing method, program for executing the method, storage medium, imaging apparatus, and image processing system | |
| CN104520848A (en) | Searching for events by attendants | |
| CN108255915A (en) | File management method and device and machine-readable storage medium | |
| Elharrouss et al. | FSC-set: counting, localization of football supporters crowd in the stadiums | |
| Lee et al. | Fine-Grained Plant Identification using wide and deep learning model | |
| Kim et al. | m CLOVER: mobile content-based leaf image retrieval system | |
| CN110169055A (en) | A kind of method and apparatus generating shot information | |
| Zhang et al. | Dataset-driven unsupervised object discovery for region-based instance image retrieval | |
| CN105095215B (en) | Information acquisition device, method and server |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |