CN104050279B - The method, apparatus and image recognition apparatus of a kind of characteristic matching - Google Patents
The method, apparatus and image recognition apparatus of a kind of characteristic matching Download PDFInfo
- Publication number
- CN104050279B CN104050279B CN201410298545.5A CN201410298545A CN104050279B CN 104050279 B CN104050279 B CN 104050279B CN 201410298545 A CN201410298545 A CN 201410298545A CN 104050279 B CN104050279 B CN 104050279B
- Authority
- CN
- China
- Prior art keywords
- storehouse
- subpattern
- feature
- hash
- round robin
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The applicable field of image recognition of the present invention, there is provided the method, apparatus and image recognition apparatus of a kind of characteristic matching, methods described include：The feature of a time point acquisition is stored with one sub- pattern base, after increasing subpattern storehouse in pattern base, the Feature Mapping by each increased subpattern storehouse is only needed into the corresponding Hash Round Robin data partition of Hash table, when needing to be matched test library with pattern base, the sub- test library newly increased is first mapped to the corresponding Hash Round Robin data partition of Hash table, feature in the Hash Round Robin data partition that the feature of sub- test library is mapped with it again is matched, you can obtains matching result.The present invention, compared with prior art, when being matched every time, it is not necessary to whole pattern base is mapped in the corresponding Hash Round Robin data partition of Hash table, spent seldom in time, be advantageous to improve matching efficiency.Also, matched using hash function, can avoid using the misrecognition problem occurred during cloth dragon wave filter, matching precision is very high.
Description
Technical field
The invention belongs to the method, apparatus and image recognition apparatus of field of image recognition, more particularly to a kind of characteristic matching.
Background technology
All it is often many times that the feature of existing pattern base is all counted in advance during image feature comparison is carried out
Calculate, be then store in a position specified, finally by the feature got in test library and the feature stored in pattern base
It is compared so as to obtain comparison result.Here the pattern base sayed often all is completed or established offline in advance, still
Under appearing in some environmental demands in the application of reality, pattern base is changed with the changes in demand in the external world, when being compared
To amount it is seldom when, can reach the requirement specified based on current hardware and software environment, but when the amount compared
It is very big, when being compared within the specified very short period based on up to ten thousand or more pattern features, current hardware loop
Border and conventional algorithm all can not efficiently solve this problem, if being based only on hash algorithm, only embody the efficient of its comparison
Property, when in specified time section comparison pattern amount increase, be required for substantial amounts of pattern base to be mapped in hash tables every time, when
Between it is upper spend it is more；According to cloth dragon wave filter (bloom filter) although counting tables can be added to realize in table
The deletion action of data, but with pattern base amount increase and in external environment various storehouses change, it is easy to occur
Bloom filter self shortcoming " vacation identification ", is also both misrecognition.
The content of the invention
The embodiments of the invention provide a kind of method, apparatus of characteristic matching and image recognition apparatus, it is intended to solves existing
The method for the characteristic matching that technology provides, compare the problem of spending the time or easily misidentifying.
On the one hand, there is provided a kind of method of characteristic matching, methods described include：
In the feature deposit subpattern storehouse that current point in time is obtained；
The subpattern storehouse is added in pattern base；
The subpattern storehouse is mapped in the corresponding Hash Round Robin data partition of Hash table by hash function；
Receive identical entry matching request；
According to the identical entry matching request, feature and the institute of the sub- test library of current point in time are established by hash function
State the mapping relations of Hash table；
The feature of a sub- pattern base is stored with if mapped in the Hash Round Robin data partition of the Hash table, the subpattern
Storehouse is the affiliated storehouse of the sub- test library of the current point in time；
The feature at least two subpattern storehouses is stored with if mapped in the Hash Round Robin data partition of the Hash table, by described in
The feature of the sub- test library of current point in time and the feature at least two subpatterns storehouse are compared one by one, and described current
The feature identical subpattern storehouse of the sub- test library at time point is the affiliated storehouse of the sub- test library of the current point in time.
Further, after in the pattern base by subpattern storehouse addition, in addition to：
The feature in the subpattern storehouse is divided into p Feature Segmentation, p is the natural number more than or equal to 2；
P Feature Segmentation of the feature in the subpattern storehouse is respectively mapped to the phase of p Hash table by hash function
Answer in Hash Round Robin data partition；
Receive similar item matching request；
According to the similar item matching request, p that the son test planting modes on sink characteristic of current point in time is established by hash function
The mapping relations of Feature Segmentation and the p Hash table；
If k-th of Feature Segmentation maps in the Hash Round Robin data partition of k-th of Hash table the feature point for being stored with subpattern storehouse
Section, then the feature in each subpattern storehouse is obtained one by one, and the characteristic value of each subpattern planting modes on sink characteristic segmentation is tested into Al Kut with son
The characteristic value of sign segmentation compares, if the number of same characteristic features segmentation is more than or equal to the threshold value of setting, the test library
Feature is similar to the feature in the subpattern storehouse, wherein, k is the natural number for being less than or equal to p more than or equal to 1.
Further, described mapped in the subpattern storehouse in the corresponding Hash Round Robin data partition of Hash table by hash function is wrapped
Include：
If there is no other subpattern storehouses to link on the Hash Round Robin data partition that the feature in the subpattern storehouse is mapped, directly use
Chained list is chained up the feature in the subpattern storehouse with the Hash Round Robin data partition mapped；
If there are other subpattern storehouses to link on the Hash Round Robin data partition that the feature in subpattern storehouse is mapped, the subpattern
The feature in storehouse links to mapped Hash Round Robin data partition backmost.
Further, before in the pattern base by subpattern storehouse addition, in addition to：
If the subpattern storehouse stored in pattern base has arrived at maximum quantity, delete and stored most in the pattern base
The subpattern storehouse that early time point obtains；
Hash Round Robin data partition corresponding with the subpattern storehouse of earliest time point acquisition is found by hash function；
Judge the feature in the subpattern storehouse that earliest time point obtains the location of in the Hash Round Robin data partition；
If the subpattern storehouse link for only having earliest time point to obtain on the Hash Round Robin data partition, directly deletes earliest time point
The subpattern storehouse of acquisition and linking for the Hash Round Robin data partition；
If there are other subpattern storehouse links behind the subpattern storehouse that earliest time point obtains, first earliest time
Linking and break between subpattern storehouse that point obtains and the Hash Round Robin data partition, then the subpattern storehouse that earliest time point is obtained and its
Link between the latter subpattern storehouse linked afterwards is broken, and finally, the address pointer that the Hash Round Robin data partition is stored points to
The head in the latter subpattern storehouse.
On the other hand, there is provided a kind of device of characteristic matching, described device include：
Subpattern storehouse creating unit, the feature for current point in time to be obtained are stored in subpattern storehouse；
Unit is put in storage in subpattern storehouse, for the subpattern storehouse to be added in pattern base；
Subpattern storehouse map unit, for the subpattern storehouse to be mapped to the corresponding Hash of Hash table by hash function
In address；
First request reception unit, for receiving identical entry matching request；
Sub- test library map unit, for according to the identical entry matching request, current time to be established by hash function
The feature of the sub- test library of point and the mapping relations of the Hash table；
First matching unit, if being stored with sub- pattern base in Hash Round Robin data partition for mapping to the Hash table
Feature, then the subpattern storehouse is the affiliated storehouse of the sub- test library of the current point in time；
Second matching unit, if being stored with least two subpatterns in Hash Round Robin data partition for mapping to the Hash table
The feature in storehouse, then the feature of the sub- test library of the current point in time and the feature at least two subpatterns storehouse are carried out one
One compares, and is tested with the feature identical subpattern storehouse of the sub- test library of the current point in time for the son of the current point in time
The affiliated storehouse in storehouse.
Further, described device also includes：
Feature Segmentation unit, for the feature in the subpattern storehouse to be divided into p Feature Segmentation, p is more than or equal to 2
Natural number；
Fisrt feature subsection compression unit, for being divided p feature of the feature in the subpattern storehouse by hash function
Section is respectively mapped in the corresponding Hash Round Robin data partition of p Hash table；
Second request reception unit, for receiving similar item matching request；
Second feature subsection compression unit, for according to the similar item matching request, being established by hash function current
The mapping relations of the p Feature Segmentation and the p Hash table of the son test planting modes on sink characteristic at time point；
Similarity matching unit, it is stored with if mapped to for k-th of Feature Segmentation in the Hash Round Robin data partition of k-th of Hash table
The Feature Segmentation in subpattern storehouse, then the feature in each subpattern storehouse, and the spy that each subpattern planting modes on sink characteristic is segmented are obtained one by one
Value indicative and the characteristic value of sub- test library Feature Segmentation compare, if the number of same characteristic features segmentation is more than or equal to the threshold of setting
Value, then the feature of the test library is similar to the feature in the subpattern storehouse, wherein, k be more than or equal to 1 be less than or equal to p oneself
So number.
Further, subpattern storehouse map unit includes：
First mapping block, if not having other submodules on the Hash Round Robin data partition that the feature for the subpattern storehouse is mapped
Formula storehouse links, then directly the feature in the subpattern storehouse is chained up with the Hash Round Robin data partition mapped with chained list；
Second mapping block, if there is other subpattern storehouses chain on the Hash Round Robin data partition that the feature for subpattern storehouse is mapped
Connect, then the feature in the subpattern storehouse is linked to mapped Hash Round Robin data partition backmost.
Further, described device also includes：
Unit is deleted in subpattern storehouse, if having arrived at maximum quantity for the subpattern storehouse stored in pattern base, is deleted
The subpattern storehouse obtained except the earliest time point stored in the pattern base；
Hash Round Robin data partition searching unit, it is corresponding with the subpattern storehouse of earliest time point acquisition for being found by hash function
Hash Round Robin data partition；
Position judgment unit, for judge earliest time point obtain subpattern storehouse feature in the Hash Round Robin data partition institute
The position at place；
Unit is deleted in first mapping, if there was only the subpattern storehouse chain that earliest time point obtains on the Hash Round Robin data partition
Connect, then directly delete the subpattern storehouse of earliest time point acquisition and linking for the Hash Round Robin data partition；
Unit is deleted in second mapping, if there is other subpatterns behind the subpattern storehouse obtained for earliest time point
Storehouse links, then linking and break between the subpattern storehouse first earliest time point obtained and the Hash Round Robin data partition, then will be earliest when
Between put linking between the subpattern storehouse of acquisition and the latter subpattern storehouse linked thereafter and break, finally, the Hash
The address pointer that location is stored points to the head in the latter subpattern storehouse.
The feature of a time point acquisition is stored with the embodiment of the present invention, a sub- pattern base, is increased in pattern base
Behind subpattern storehouse, it is only necessary to map in each increased subpattern storehouse in the corresponding Hash Round Robin data partition of Hash table, compared to existing skill
Art, it is necessary to whole pattern base be mapped in the corresponding Hash Round Robin data partition of Hash table when being matched every time, spend in time very
It is few, be advantageous to improve matching efficiency.Also, matched using hash function, can avoid occurring when using cloth dragon wave filter
Misrecognition problem, matching precision is very high.
Brief description of the drawings
Fig. 1 is the implementation process figure of the method for the characteristic matching that the embodiment of the present invention one provides；
Fig. 2 is the relation in pattern base and subpattern storehouse that the embodiment of the present invention one provides, and the feature in subpattern storehouse and
Mapping relations schematic diagram between the Hash Round Robin data partition of Hash table；
Fig. 3 be the embodiment of the present invention one provide pattern base in add new subpattern storehouse after, the change of pattern base and
Mapping relations schematic diagram between the feature in each subpattern storehouse and the Hash Round Robin data partition of Hash table；
Fig. 4 is the mapping relations signal between sub- test library and Hash table in the test library that the embodiment of the present invention one provides
Figure；
Fig. 5 is the implementation process figure of the method for the characteristic matching that the embodiment of the present invention two provides；
Fig. 6 is being segmented to the feature in sub- test library for the offer of the embodiment of the present invention two, is divided into p Feature Segmentation
Schematic diagram；
Fig. 7 is the mapping of the 1st Feature Segmentation and Hash table 1 in every sub- test library that the embodiment of the present invention two provides
Relation schematic diagram；
Fig. 8 is the mapping of p-th of the Feature Segmentation and Hash table p in every sub- test library that the embodiment of the present invention two provides
Relation schematic diagram；
Fig. 9 be the embodiment of the present invention two provide sub- test library test_ti in the 1st to p Feature Segmentation respectively with Kazakhstan
Table 1 is wished to Hash table p mapping relations schematic diagram；
Figure 10 is the structured flowchart of the device for the characteristic matching that the embodiment of the present invention three provides；
Figure 11 is the structured flowchart of the device for the characteristic matching that the embodiment of the present invention four provides.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, it is right below in conjunction with drawings and Examples
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
In embodiments of the present invention, feature current point in time obtained is stored in subpattern storehouse；By the subpattern storehouse
Add in pattern base；The subpattern storehouse is mapped in the corresponding Hash Round Robin data partition of Hash table by hash function；Receive identical
Item matching request；According to the identical entry matching request, the feature of the sub- test library of current point in time is established by hash function
With the mapping relations of the Hash table；The spy of a sub- pattern base is stored with if mapped in the Hash Round Robin data partition of the Hash table
Sign, then the subpattern storehouse is the affiliated storehouse of the sub- test library of the current point in time；If map to the Kazakhstan of the Hash table
The feature at least two subpattern storehouses is stored with uncommon address, then by the feature of the sub- test library of the current point in time with it is described
The feature at least two subpattern storehouses is compared one by one, the feature identical submodule with the sub- test library of the current point in time
Formula storehouse is the affiliated storehouse of the sub- test library of the current point in time.
It is described in detail below in conjunction with realization of the specific embodiment to the present invention：
Embodiment one
Fig. 1 shows the implementation process of the method for the characteristic matching that the embodiment of the present invention one provides, and details are as follows：
In step S101, the feature that current point in time is obtained is stored in subpattern storehouse.
In the present embodiment, for the feature in pattern base and the feature in test library can use certain character string,
Numeral or symbol describe.
In the present embodiment, feature that the ti moment obtains is represented with Lib_ti (i=1,2,3 ... N).Wherein, ti is represented
Time, Lib_t1 represent the feature that is obtained at moment time t1, the feature that Lib_t2 expressions obtain at moment time t2, i in ti
It is bigger, represent that time experience is longer, wherein each Lib_ti can be represented with a string of characters, numeral or symbol.
The feature of the test library of acquisition is inscribed in Test_ti (i=1,2,3 ... M) expressions in time ti, and i is bigger in ti,
Represent that time experience is longer, wherein each Test_ti can be represented with a string of characters, numeral or symbol.
Wherein, the device of characteristic matching by the characteristic storage that each time point obtains in a sub- pattern base, such as, ti
The feature Lib_ti that moment obtains is stored in the Lib_ti of subpattern storehouse.
In step s 102, the subpattern storehouse is added in pattern base.
In the present embodiment, the device of characteristic matching adds in the subpattern storehouse got in pattern base.Pattern base is used for
Subpattern storehouse is stored, as shown in Fig. 2 the device that the characteristic matching interior for the previous period of current point in time is stored with pattern base obtains
The subpattern storehouse taken, such as, being stored with t1, t2, t3, ti-2, ti-1, ti, ti+1, ti+2, ti+3 moment in pattern base obtains
Subpattern storehouse Lib_t1, Lib_t2, Lib_t3, Lib_ti-2, Lib_ti-1, Lib_ti, Lib_ti+1, Lib_ti+2,
Lib_ti+3。
Preferably, the finite capacity in the subpattern storehouse that can be stored in pattern base, so to delete some moment pair earlier
The subpattern storehouse answered, the feature in the subpattern storehouse that the newest moment obtains is added in pattern base.For example in Fig. 3, delete t1
With the subpattern storehouse at t2 moment, and the subpattern storehouse Lib_ti+4 and Lib_ti+5 at ti+4 moment and ti+5 moment are added to mould
In formula storehouse.
In step s 103, the subpattern storehouse is mapped in the corresponding Hash Round Robin data partition of Hash table by hash function.
In the present embodiment, a new subpattern storehouse is often added in pattern base, the subpattern storehouse and Hash will be established
The mapping relations of table, by the corresponding Hash Round Robin data partition of the Feature Mapping in the subpattern storehouse to Hash table.
Specifically, hash function can be built, by the hash function by the spy in the new subpattern storehouse added in pattern base
Sign is mapped in Hash table.As shown in Fig. 2 the change of the feature in subpattern storehouse in pattern base over time adds accordingly
Timestamp, be so advantageous in subsequent applications for the image recognition on special time period.In fig. 2, by each moment pair
The Feature Mapping in the subpattern storehouse answered into the corresponding Hash Round Robin data partition of Hash table, such as, by subpattern storehouse Lib_ti+3 feature
The 4th Hash Round Robin data partition of Hash table is mapped to, by the second of subpattern storehouse Lib_t2 and Lib_ti Feature Mapping to Hash table
Individual Hash Round Robin data partition.
During due to being mapped using hash function, it may appear that by the Feature Mapping in two or more subpattern storehouses
To the same address of Hash table, that is, collide problem, at this moment needs to use chain address method, will be mapped to using the structure of chained list
Two or more subpattern storehouse of same Hash Round Robin data partition is chained up, as shown in figure 3, when time point reaches ti+3
When, because the subpattern storehouse stored in pattern base has arrived at maximum limitation quantity, so to delete some earlier time point obtain
The subpattern storehouse taken, specifically, the subpattern storehouse that earliest time point obtains first is deleted, the subpattern storehouse that the newest moment is obtained
Feature is added in pattern base, and by the Feature Mapping in the newest subpattern storehouse being added in pattern base to the corresponding Kazakhstan of Hash table
In uncommon address and it is linked to chained list that the Hash Round Robin data partition is stored backmost.
Specific Link Rule is as follows：
1), for the feature in the subpattern storehouse to be deleted, first pass through hash function and find feature with the subpattern storehouse
Corresponding Hash Round Robin data partition, the feature in the subpattern storehouse is then judged the location of in the Hash Round Robin data partition, if the Hash Round Robin data partition
Upper only subpattern storehouse link, then directly delete the subpattern storehouse, if there is other subpattern storehouses behind the subpattern storehouse
Link, at this moment first linking between the subpattern storehouse and the Hash Round Robin data partition is broken, then by the subpattern storehouse with linking thereafter
The latter subpattern storehouse between link break, finally, the address pointer that the Hash Round Robin data partition is stored points to latter height
The head of pattern base；
2), for the feature in the subpattern storehouse for entering new addition pattern base, if the Hash that the feature in subpattern storehouse is mapped
There is no other subpattern storehouses to link on address, then directly with chained list method the feature in the subpattern storehouse and the Hash mapped
Address is chained up；
3), for the feature in the subpattern storehouse for entering new addition pattern base, if the Hash that the feature in subpattern storehouse is mapped
There are other subpattern storehouses to link on address, then the feature in the subpattern storehouse is linked to the Hash Round Robin data partition backmost.
Above rule can illustrate that end product is as shown in Figure 3 with following example：
When ti+3 is reached at time point, the subpattern storehouse stored in pattern base has arrived at maximum limitation, maps
The feature in Lib_t1 storehouses as shown in Figure 2 above, when to time ti+4, is first be mapped to Kazakhstan by the structure into Hash table
Chained list in uncommon table is deleted, from fig. 2 it can be seen that at this time cutting off between Hash (Lib_t1) and Hash (Lib_ti-2)
Link, the address pointer that then currently stored Hash (Lib_t1) Hash Round Robin data partition is stored point to Hash (lib_ti-2),
The new subpattern planting modes on sink characteristic come at the ti+4 moment, labeled as Lib_ti+4, by the mapping of hash function, find
Lib_ti+4 and Lib_ti+3 Hash Round Robin data partition is identical, and Hash (Lib_ti+4) is directly at this moment linked to Hash with chained list
(Lib_ti+3) behind.
When to time point ti+5, the chained list first feature in Lib_t2 storehouses being be mapped in Hash table is deleted, warp
Hash function maps, and finds do not have other chained lists to link on the Hash Round Robin data partition that Lib_ti+5 is mapped, is empty, then Hash
(Lib_ti+5) it is linked to behind the Hash Round Robin data partition.
The feature that the subpattern storehouse to being stored is analogized according to above-mentioned described mode is constantly deleted and inserted, and
Corresponding Hash table is also deleted accordingly and insertion operation.Such as needed in pattern base increase moment ti+4 and
Moment at moment ti+5 obtains the feature obtained, it is necessary to the feature that the deletion t1 and t2 moment obtains in the pattern base shown in Fig. 2
Lib_t1 and Lib_t2, and delete linking between Lib_t1 and Lib_t2 and corresponding Hash Round Robin data partition.
In step S104, identical entry matching request is received.
In the present embodiment, in test phase, within the period of setting, user, which inputs, searches asking for same characteristic features storehouse
Ask, ask to find out the feature identical subpattern with the test library of current point in time in the device slave pattern storehouse of characteristic matching
Storehouse, the affiliated storehouse using the subpattern storehouse as the test library.Wherein, test library includes the son survey of multiple time points acquisitions
Try storehouse, the feature that a time point obtains be stored with every sub- test library, as shown in figure 4, test library include time t1,
T2, t3 ... sub- test library Test_t1, Test_t2 that ti+3, ti+4, ti+5 moment obtain ... Test_ti+3, Test_
ti+4、Test_ti+5。
In step S105, according to the identical entry matching request, the son that current point in time is established by hash function is surveyed
The mapping relations of feature and the Hash table in storehouse are tried, if being stored with a son in mapping to the Hash Round Robin data partition of the Hash table
The feature of pattern base, then the subpattern storehouse is the affiliated storehouse of the sub- test library of the current point in time；If map to described
The feature at least two subpattern storehouses is stored with the Hash Round Robin data partition of Hash table, then by the sub- test library of the current point in time
Feature and the feature at least two subpatterns storehouse are compared one by one, the feature with the sub- test library of the current point in time
Identical subpattern storehouse is the affiliated storehouse of the sub- test library of the current point in time.
In the present embodiment, after the device of characteristic matching receives identical entry matching request, established and worked as by hash function
The feature of the sub- test library at preceding time point is tested with the mapping relations of the Hash table come the son obtained with the current point in time
The feature identical subpattern storehouse in storehouse.
For example in the present embodiment, to be searched in pattern base with the sub- test library Test_ti+5's at moment at time point ti+5
Feature identical subpattern storehouse, then first pass through the corresponding Hash that hash function maps to sub- test library Test_ti+5 Hash table
In address, in the present embodiment, Test_ti+5 maps to the 4th Hash Round Robin data partition of Hash table, as shown in figure 4, being deposited in the address
Hash (Lib_ti+3) and Hash (Lib_ti+4) are contained, then is needed Test_ti+5 feature and Lib_ti+3 and Lib_ti+
4 feature is compared one by one, using the affiliated storehouse with Test_ti+5 feature identical subpattern storehouse as Test_ti+5.
The present embodiment, the feature that a time point obtains is stored with a sub- pattern base, increases subpattern in pattern base
Behind storehouse, it is only necessary to map in each increased subpattern storehouse in the corresponding Hash Round Robin data partition of Hash table, compared with prior art, every time
When being matched, it is not necessary to map to whole pattern base in the corresponding Hash Round Robin data partition of Hash table, spend seldom, have in time
Beneficial to raising matching efficiency.Also, matched using hash function, can avoid using the mistake occurred during cloth dragon wave filter to know
Other problem, matching precision are very high.Found by experiment test, the comparison lower time based on 15000 pattern bases per minute is only
It is only very high in Millisecond, efficiency.The method of this characteristic matching can be applied in image recognition, Large Scale Graphs line matching field.
Can one of ordinary skill in the art will appreciate that realizing that all or part of step in the various embodiments described above method is
To instruct the hardware of correlation to complete by program, corresponding program can be stored in a computer read/write memory medium
In, described storage medium, such as ROM/RAM, disk or CD.
Embodiment two
Fig. 5 shows the implementation process of the method for the characteristic matching that the embodiment of the present invention two provides, and details are as follows：
In step S501, the feature that current point in time is obtained is stored in subpattern storehouse.
In step S502, the subpattern storehouse is added in pattern base.
In step S503, the feature in the subpattern storehouse is divided into p Feature Segmentation.
In step S504, by p hash function by p Feature Segmentation correspondence mappings of the feature in the subpattern storehouse
Into the corresponding Hash Round Robin data partition of p Hash table.
In the present embodiment, the feature in subpattern storehouse is divided into p Feature Segmentation, as shown in Figure 6.Can be with from Fig. 6
See, subpattern storehouse is characterized in character, numeral or symbol, is actually needed and these characters, numeral or symbol are divided
Section, it is divided into P sections here (P is the natural number more than 2).
Each Feature Segmentation of the feature in subpattern storehouse is mapped in different Hash tables, such as, as shown in fig. 7, will
1st Feature Segmentation Lib_1_t1 of subpattern storehouse Lib_t1 feature maps to the 4th Hash Round Robin data partition of Hash table 1；By submodule
P-th of Feature Segmentation Lib_p_t1 of formula storehouse Lib_t1 feature maps to Hash table p the 5th Hash Round Robin data partition, as shown in Figure 8.
In step S505, similar item matching request is received.
In the present embodiment, in test phase, within the period of setting, user, which inputs, searches asking for similar features storehouse
Ask, ask to find out the subpattern similar to the feature of the test library of current point in time in the device slave pattern storehouse of characteristic matching
Storehouse, the affiliated storehouse using the subpattern storehouse as the test library.
In step S506, according to the similar item matching request, by the p of the feature of the sub- test library of current point in time
Feature Segmentation, the p Feature Segmentation and p Hash table of the son test planting modes on sink characteristic of current point in time are established by p hash function
Mapping relations, if k-th of Feature Segmentation maps in the Hash Round Robin data partition of k-th of Hash table the feature for being stored with subpattern storehouse
Segmentation, then the feature in each subpattern storehouse, and the characteristic value that each subpattern planting modes on sink characteristic is segmented and sub- test library are obtained one by one
The characteristic value of Feature Segmentation compares；If the number of same characteristic features segmentation is more than or equal to the threshold value of setting, the test library
Subpattern storehouse of the feature to comparing feature it is similar；If the number of same characteristic features segmentation is less than the threshold value of setting, after
The continuous Hash Round Robin data partition found+1 Feature Segmentation of kth and map to+1 Hash table of kth, has been look for p-th of Feature Segmentation, if
The subpattern storehouse for the threshold value for meeting setting is not found also, then it is assumed that the feature of the sub- test library is not in the pattern base.
In the present embodiment, if the feature of a sub- pattern base of the feature of the test library of current point to finding is similar
Degree is very big, then two features number that same characteristic features are segmented to each other is more, when the number of same characteristic features segmentation is P, explanation
The feature of test library is identical with the feature in subpattern storehouse.Therefore, during practical application, a threshold value is set, when same characteristic features point
When the number of section is more than or equal to the threshold value, represent that the test library of current point is similar to the subpattern storehouse found.Wherein, the threshold
Value is less than p, can be determined according to actual conditions.
In the present embodiment, the sub- test library test_ti of current point in time to be tested feature is also divided into p feature
Segmentation, then, this p Feature Segmentation is respectively mapped in the corresponding Hash Round Robin data partition in Hash table 1 to Hash table p, such as Fig. 9 institutes
Show.
If k-th of Feature Segmentation maps in the Hash Round Robin data partition of k-th of Hash table the feature point for being stored with subpattern storehouse
Section, then obtain the feature in each subpattern storehouse one by one, and by the Feature Segmentation and current point in time in each subpattern storehouse of this feature
The Feature Segmentation of sub- test library compare；If the number of same characteristic features segmentation is more than or equal to the threshold value of setting, the survey
The feature for trying subpattern storehouse of the feature in storehouse to being compared is similar.If the number of same characteristic features segmentation is less than the threshold value of setting,
The Hash Round Robin data partition that+1 Feature Segmentation of kth maps to+1 Hash table of kth is continually looked for, is had been look for p-th of Feature Segmentation,
If the subpattern storehouse for the threshold value for meeting setting is not found also, then it represents that the feature of the test library is not in the pattern base.
Wherein, k is the natural number for being less than or equal to p more than or equal to 1.
Preferably, after step S502, the step S103 to S105 in embodiment one can also be performed, according to feature
The request that the device matched somebody with somebody receives, it can select to perform step S103 to S105 or perform step S503 to S506.
The present embodiment, the feature in the new subpattern storehouse added in pattern base is divided into p Feature Segmentation and passes through Kazakhstan respectively
Then uncommon Function Mapping is established the mapping pass of p Feature Segmentation of the feature of sub- test library to be tested into p Hash table
System, after establishing mapping relations, each Feature Segmentation of sub- test library is compared with each Feature Segmentation in subpattern storehouse,
If the number of same characteristic features segmentation is more than or equal to the threshold value of setting, then it is assumed that, the feature of sub- test library is with being compared
The feature in subpattern storehouse is similar.
Can one of ordinary skill in the art will appreciate that realizing that all or part of step in the various embodiments described above method is
To instruct the hardware of correlation to complete by program, corresponding program can be stored in a computer read/write memory medium
In, described storage medium, such as ROM/RAM, disk or CD.
Embodiment three
Figure 10 shows the concrete structure block diagram of the device for the characteristic matching that the embodiment of the present invention three provides, for the ease of saying
It is bright, it illustrate only the part related to the embodiment of the present invention.The device 10 can be built in computer, can also be built in one
The identification of image is completed in individual special image recognition apparatus, such as, the identification of fingerprint, the device 10 includes：Create in subpattern storehouse
Build unit 101, subpattern storehouse storage unit 102, subpattern storehouse map unit 103, the first request reception unit 104, sub- test
Storehouse map unit 105, the first matching unit 106 and the second matching unit 107.
Wherein, subpattern storehouse creating unit 101, the feature for current point in time to be obtained are stored in subpattern storehouse；
Unit 102 is put in storage in subpattern storehouse, for the subpattern storehouse to be added in pattern base；
Subpattern storehouse map unit 103, for the subpattern storehouse to be mapped into the corresponding of Hash table by hash function
In Hash Round Robin data partition；
First request reception unit 104, for receiving identical entry matching request；
Sub- test library map unit 105, for according to the identical entry matching request, when establishing current by hash function
Between the feature of sub- test library and the mapping relations of the Hash table put；
First matching unit 106, if being stored with a subpattern in Hash Round Robin data partition for mapping to the Hash table
The feature in storehouse, then the subpattern storehouse is the affiliated storehouse of the sub- test library of the current point in time；
Second matching unit 107, if being stored with least two sons in Hash Round Robin data partition for mapping to the Hash table
The feature of pattern base, then the feature of the sub- test library of the current point in time and the feature at least two subpatterns storehouse are entered
Row compares one by one, and the feature identical subpattern storehouse with the sub- test library of the current point in time is the son of the current point in time
The affiliated storehouse of test library.
Specifically, subpattern storehouse map unit 103 includes：
First mapping block, if not having other submodules on the Hash Round Robin data partition that the feature for the subpattern storehouse is mapped
Formula storehouse links, then directly the feature in the subpattern storehouse is chained up with the Hash Round Robin data partition mapped with chained list；
Second mapping block, if there is other subpattern storehouses chain on the Hash Round Robin data partition that the feature for subpattern storehouse is mapped
Connect, then the feature in the subpattern storehouse is linked to mapped Hash Round Robin data partition backmost.
Preferably, described device 10 also includes：
Unit is deleted in subpattern storehouse, if having arrived at maximum quantity for the subpattern storehouse stored in pattern base, is deleted
The subpattern storehouse obtained except the earliest time point stored in the pattern base；
Hash Round Robin data partition searching unit, it is corresponding with the subpattern storehouse of earliest time point acquisition for being found by hash function
Hash Round Robin data partition；
Position judgment unit, for judge earliest time point obtain subpattern storehouse feature in the Hash Round Robin data partition institute
The position at place；
Unit is deleted in first mapping, if there was only the subpattern storehouse chain that earliest time point obtains on the Hash Round Robin data partition
Connect, then directly delete the subpattern storehouse of earliest time point acquisition and linking for the Hash Round Robin data partition；
Unit is deleted in second mapping, if there is other subpatterns behind the subpattern storehouse obtained for earliest time point
Storehouse links, then linking and break between the subpattern storehouse first earliest time point obtained and the Hash Round Robin data partition, then will be earliest when
Between put linking between the subpattern storehouse of acquisition and the latter subpattern storehouse linked thereafter and break, finally, the Hash
The address pointer that location is stored points to the head in the latter subpattern storehouse.
The device of characteristic matching provided in an embodiment of the present invention can be applied in foregoing corresponding embodiment of the method one, in detail
Feelings will not be repeated here referring to the description of above-described embodiment one.
Example IV
Figure 11 shows the concrete structure block diagram of the device for the characteristic matching that the embodiment of the present invention four provides, for the ease of saying
It is bright, it illustrate only the part related to the embodiment of the present invention.The device 11 can be built in computer, can also be built in one
The identification of image is completed in individual special image recognition apparatus, such as, the identification of fingerprint, the device 11 includes：Create in subpattern storehouse
Building unit 111, subpattern storehouse storage unit 112, Feature Segmentation unit 113, fisrt feature subsection compression unit 114, second please
Ask receiving unit 115, second feature subsection compression unit 116 and Similarity matching unit 117.
Wherein, subpattern storehouse creating unit 111, the feature for current point in time to be obtained are stored in subpattern storehouse；
Unit 112 is put in storage in subpattern storehouse, for the subpattern storehouse to be added in pattern base；
Feature Segmentation unit 113, for the feature in the subpattern storehouse to be divided into p Feature Segmentation, p be more than or equal to
2 natural number；
Fisrt feature subsection compression unit 114, for by hash function by p feature of the feature in the subpattern storehouse
It is respectively mapped in the corresponding Hash Round Robin data partition of p Hash table；
Second request reception unit 115, for receiving similar item matching request；
Second feature subsection compression unit 116, for according to the similar item matching request, being established and being worked as by hash function
The mapping relations of the p Feature Segmentation and the p Hash table of the son test planting modes on sink characteristic at preceding time point；
Similarity matching unit 117, deposited if mapped to for k-th of Feature Segmentation in the Hash Round Robin data partition of k-th of Hash table
The Feature Segmentation in subpattern storehouse is contained, then obtains the feature in each subpattern storehouse one by one, and each subpattern planting modes on sink characteristic is segmented
Characteristic value and the characteristic value of sub- test library Feature Segmentation compare, if same characteristic features segmentation number be more than or equal to setting
Threshold value, then the feature of the test library is similar to the feature in the subpattern storehouse, wherein, k is less than or equal to p's more than or equal to 1
Natural number.
Preferably, subpattern storehouse map unit 103, the first request reception unit 104, son can also be included in the device 11
Test library map unit 105, the first matching unit 106 and the second matching unit 107, to realize the matching of feature identical entry.Tool
Body selects to perform identical entry matching or the matching of similar item, is determined according to the content of the request instruction of user's transmission.
The device of characteristic matching provided in an embodiment of the present invention can be applied in foregoing corresponding embodiment of the method two, in detail
Feelings will not be repeated here referring to the description of above-described embodiment two.
It is worth noting that, in said apparatus embodiment, included unit is simply drawn according to function logic
Point, but above-mentioned division is not limited to, as long as corresponding function can be realized；In addition, each functional unit is specific
Title is also only to facilitate mutually distinguish, the protection domain being not intended to limit the invention.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all essences in the present invention
All any modification, equivalent and improvement made within refreshing and principle etc., should be included in the scope of the protection.
Claims (9)
- A kind of 1. method of characteristic matching, it is characterised in that methods described includes：In the feature deposit subpattern storehouse that current point in time is obtained；The subpattern storehouse is added in pattern base, the pattern base is used to store subpattern storehouse, is stored with the pattern base The subpattern storehouse of the acquisition interior for the previous period of current point in time；The subpattern storehouse is mapped in the corresponding Hash Round Robin data partition of Hash table by hash function, the submodule in the pattern base The change of the feature in formula storehouse over time adds corresponding timestamp；Receive identical entry matching request；According to the identical entry matching request, the feature of the sub- test library of current point in time and the Kazakhstan are established by hash function The mapping relations of uncommon table；The feature of a sub- pattern base is stored with if mapped in the Hash Round Robin data partition of the Hash table, the subpattern storehouse is The affiliated storehouse of the sub- test library of the current point in time；The feature at least two subpattern storehouses is stored with if mapped in the Hash Round Robin data partition of the Hash table, will be described current The feature of the sub- test library at time point and the feature at least two subpatterns storehouse are compared one by one, with the current time The feature identical subpattern storehouse of the sub- test library of point is the affiliated storehouse of the sub- test library of the current point in time.
- 2. the method as described in claim 1, it is characterised in that after in the pattern base by subpattern storehouse addition, Also include：The feature in the subpattern storehouse is divided into p Feature Segmentation, p is the natural number more than or equal to 2；By hash function by p Feature Segmentation correspondence mappings of the feature in the subpattern storehouse to the corresponding Kazakhstan of p Hash table In uncommon address；Receive similar item matching request；According to the similar item matching request, the son that current point in time is established by hash function tests p feature of planting modes on sink characteristic Segmentation and the mapping relations of the p Hash table；If k-th of Feature Segmentation maps in the Hash Round Robin data partition of k-th of Hash table the Feature Segmentation for being stored with subpattern storehouse, The feature in each subpattern storehouse is obtained one by one, and the characteristic value of each subpattern planting modes on sink characteristic segmentation and the feature of sub- test library are divided The characteristic value of section compares, if the number of same characteristic features segmentation is more than or equal to the threshold value of setting, the feature of the test library It is similar to the feature in the subpattern storehouse, wherein, k is the natural number for being less than or equal to p more than or equal to 1.
- 3. the method as described in claim 1, it is characterised in that described that the subpattern storehouse is mapped to by Kazakhstan by hash function The corresponding Hash Round Robin data partition of uncommon table includes：If there is no other subpattern storehouses to link on the Hash Round Robin data partition that the feature in the subpattern storehouse is mapped, chained list is directly used The feature in the subpattern storehouse is chained up with the Hash Round Robin data partition mapped；If there are other subpattern storehouses to link on the Hash Round Robin data partition that the feature in subpattern storehouse is mapped, the subpattern storehouse Feature links to mapped Hash Round Robin data partition backmost.
- 4. the method as described in claim 1, it is characterised in that before in the pattern base by subpattern storehouse addition, Also include：If the subpattern storehouse stored in pattern base has arrived at maximum quantity, delete stored in the pattern base it is earliest when Between put the subpattern storehouse of acquisition；Hash Round Robin data partition corresponding with the subpattern storehouse of earliest time point acquisition is found by hash function；Judge the feature in the subpattern storehouse that earliest time point obtains the location of in the Hash Round Robin data partition；If the subpattern storehouse link for only having earliest time point to obtain on the Hash Round Robin data partition, directly delete earliest time point and obtain Subpattern storehouse and the Hash Round Robin data partition link；If there is other subpattern storehouse links behind the subpattern storehouse that earliest time point obtains, first earliest time point is obtained Linking between the subpattern storehouse taken and the Hash Round Robin data partition breaks, then the subpattern storehouse that earliest time point is obtained and chain thereafter Link between the latter subpattern storehouse connect is broken, and finally, the address pointer that the Hash Round Robin data partition is stored points to latter The head of individual sub- pattern base.
- 5. a kind of device of characteristic matching, it is characterised in that described device includes：Subpattern storehouse creating unit, the feature for current point in time to be obtained are stored in subpattern storehouse；Unit is put in storage in subpattern storehouse, and for the subpattern storehouse to be added in pattern base, the pattern base is used to store subpattern Storehouse, the subpattern storehouse of the acquisition interior for the previous period of current point in time is stored with the pattern base；Subpattern storehouse map unit, for the subpattern storehouse to be mapped to the corresponding Hash Round Robin data partition of Hash table by hash function In, the change of the feature in the subpattern storehouse in the pattern base over time adds corresponding timestamp；First request reception unit, for receiving identical entry matching request；Sub- test library map unit, for according to the identical entry matching request, current point in time to be established by hash function The feature of sub- test library and the mapping relations of the Hash table；First matching unit, if being stored with the spy of a sub- pattern base in Hash Round Robin data partition for mapping to the Hash table Sign, then the subpattern storehouse is the affiliated storehouse of the sub- test library of the current point in time；Second matching unit, if being stored with least two subpattern storehouses in Hash Round Robin data partition for mapping to the Hash table Feature, then the feature of the sub- test library of the current point in time and the feature at least two subpatterns storehouse are compared one by one Right, the feature identical subpattern storehouse with the sub- test library of the current point in time is the sub- test library of the current point in time Affiliated storehouse.
- 6. device as claimed in claim 5, it is characterised in that described device also includes：Feature Segmentation unit, for the feature in the subpattern storehouse to be divided into p Feature Segmentation, p is the nature more than or equal to 2 Number；Fisrt feature subsection compression unit, for being divided p Feature Segmentation of the feature in the subpattern storehouse by hash function Do not map in the corresponding Hash Round Robin data partition of p Hash table；Second request reception unit, for receiving similar item matching request；Second feature subsection compression unit, for according to the similar item matching request, current time to be established by hash function The mapping relations of the p Feature Segmentation and the p Hash table of the son test planting modes on sink characteristic of point；Similarity matching unit, submodule is stored with if mapped to for k-th of Feature Segmentation in the Hash Round Robin data partition of k-th of Hash table The Feature Segmentation in formula storehouse, then the feature in each subpattern storehouse, and the characteristic value that each subpattern planting modes on sink characteristic is segmented are obtained one by one Compared with the characteristic value of sub- test library Feature Segmentation, if the number of same characteristic features segmentation is more than or equal to the threshold value of setting, The feature of the test library is similar to the feature in the subpattern storehouse, wherein, k is the natural number for being less than or equal to p more than or equal to 1.
- 7. device as claimed in claim 5, it is characterised in that subpattern storehouse map unit includes：First mapping block, if not having other subpattern storehouses on the Hash Round Robin data partition that the feature for the subpattern storehouse is mapped Link, then be directly chained up the feature in the subpattern storehouse with the Hash Round Robin data partition mapped with chained list；Second mapping block, if there is other subpattern storehouses to link on the Hash Round Robin data partition that the feature for subpattern storehouse is mapped, The feature in the subpattern storehouse is then linked to mapped Hash Round Robin data partition backmost.
- 8. device as claimed in claim 5, it is characterised in that described device also includes：Unit is deleted in subpattern storehouse, if having arrived at maximum quantity for the subpattern storehouse stored in pattern base, deletes institute State the subpattern storehouse that the earliest time point stored in pattern base obtains；Hash Round Robin data partition searching unit, for finding breathe out corresponding with the subpattern storehouse of earliest time point acquisition by hash function Uncommon address；Position judgment unit, the feature in the subpattern storehouse obtained for judging earliest time point are residing in the Hash Round Robin data partition Position；Unit is deleted in first mapping, if the subpattern storehouse link for only having earliest time point to obtain on the Hash Round Robin data partition, Directly delete the subpattern storehouse of earliest time point acquisition and linking for the Hash Round Robin data partition；Unit is deleted in second mapping, if there is other subpattern storehouses chain behind the subpattern storehouse obtained for earliest time point Connect, then linking and break between the subpattern storehouse first earliest time point obtained and the Hash Round Robin data partition, then by earliest time point Linking between the subpattern storehouse of acquisition and the latter subpattern storehouse linked thereafter breaks, finally, the Hash Round Robin data partition institute The address pointer of storage points to the head in the latter subpattern storehouse.
- 9. a kind of image recognition apparatus, it is characterised in that described image identification equipment is included such as any one of claim 5 to 8 institute The device for the characteristic matching stated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410298545.5A CN104050279B (en) | 2014-06-27 | 2014-06-27 | The method, apparatus and image recognition apparatus of a kind of characteristic matching |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410298545.5A CN104050279B (en) | 2014-06-27 | 2014-06-27 | The method, apparatus and image recognition apparatus of a kind of characteristic matching |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104050279A CN104050279A (en) | 2014-09-17 |
| CN104050279B true CN104050279B (en) | 2018-03-06 |
Family
ID=51503111
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410298545.5A Active CN104050279B (en) | 2014-06-27 | 2014-06-27 | The method, apparatus and image recognition apparatus of a kind of characteristic matching |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104050279B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104572080B (en) * | 2014-12-16 | 2018-01-19 | 广东欧珀移动通信有限公司 | A kind of photographic method and device |
| CN110084590A (en) * | 2019-03-27 | 2019-08-02 | 努比亚技术有限公司 | Method for controlling mobile terminal, mobile terminal and computer readable storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101840578A (en) * | 2009-03-17 | 2010-09-22 | 鸿富锦精密工业（深圳）有限公司 | Camera device and dynamic detection method thereof |
| CN102314485A (en) * | 2011-07-27 | 2012-01-11 | 中国科学院计算机网络信息中心 | Method and device for adding, searching and deleting hash table |
| CN102364463A (en) * | 2011-09-19 | 2012-02-29 | 浪潮电子信息产业股份有限公司 | Hash-based method for searching CAM (central address memory) |
| CN103514293A (en) * | 2013-10-09 | 2014-01-15 | 北京中科模识科技有限公司 | Method for video matching in video template library |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002117407A (en) * | 2000-10-10 | 2002-04-19 | Satake Corp | Dynamic image retrieval method and device thereof |
| US8189925B2 (en) * | 2009-06-04 | 2012-05-29 | Microsoft Corporation | Geocoding by image matching |
-
2014
- 2014-06-27 CN CN201410298545.5A patent/CN104050279B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101840578A (en) * | 2009-03-17 | 2010-09-22 | 鸿富锦精密工业（深圳）有限公司 | Camera device and dynamic detection method thereof |
| CN102314485A (en) * | 2011-07-27 | 2012-01-11 | 中国科学院计算机网络信息中心 | Method and device for adding, searching and deleting hash table |
| CN102364463A (en) * | 2011-09-19 | 2012-02-29 | 浪潮电子信息产业股份有限公司 | Hash-based method for searching CAM (central address memory) |
| CN103514293A (en) * | 2013-10-09 | 2014-01-15 | 北京中科模识科技有限公司 | Method for video matching in video template library |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104050279A (en) | 2014-09-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101055585B (en) | System and method for clustering documents | |
| US7133860B2 (en) | Device and method for automatically classifying documents using vector analysis | |
| CN105589938A (en) | Image retrieval system and retrieval method based on FPGA | |
| Liu | Study on application of apriori algorithm in data mining | |
| WO2018015848A2 (en) | Finding k extreme values in constant processing time | |
| CN107463665A (en) | A kind of data correlation rule mining algorithms | |
| CN106326475A (en) | High-efficiency static hash table implement method and system | |
| CN106126730B (en) | A kind of method and device of Mass production warning information | |
| CN104050279B (en) | The method, apparatus and image recognition apparatus of a kind of characteristic matching | |
| CN103246663A (en) | Searching method and communication terminal | |
| CN107807793B (en) | The storage of data copy isomery and access method in distributed computer storage system | |
| Srinadh | Evaluation of Apriori, FP growth and Eclat association rule mining algorithms | |
| US7197498B2 (en) | Apparatus, system and method for updating a sorted list | |
| CN107506370A (en) | Multi-medium data depth method for digging, storage medium and electronic equipment | |
| CN106934033A (en) | A kind of bent plate robot data indexing means and device | |
| CN103823834B (en) | Device and method for data transmission among Hash join operators | |
| CN101655844A (en) | Device, method and equipment for information data interaction for processing information data | |
| CN106911969B (en) | A kind of focus searching method, device and electronic equipment | |
| CN107273509B (en) | Neural network data memory, data storage method and data search method | |
| CN101354722B (en) | Method, apparatus and system for controlling version | |
| Dash et al. | Efficient reservoir sampling for transactional data streams | |
| KR20220099745A (en) | A spatial decomposition-based tree indexing and query processing methods and apparatus for geospatial blockchain data retrieval | |
| CN106528590A (en) | A query method and apparatus | |
| US10372917B1 (en) | Uniquely-represented B-trees | |
| Bulysheva et al. | Visual database design: indexing methods |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |