CN103020252B - A kind of remote sensing image demand fusion method based on demand characteristic association - Google Patents

A kind of remote sensing image demand fusion method based on demand characteristic association Download PDF

Info

Publication number
CN103020252B
CN103020252B CN201210558202.9A CN201210558202A CN103020252B CN 103020252 B CN103020252 B CN 103020252B CN 201210558202 A CN201210558202 A CN 201210558202A CN 103020252 B CN103020252 B CN 103020252B
Authority
CN
China
Prior art keywords
demand
remote sensing
sensing image
correlation
strong
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
Application number
CN201210558202.9A
Other languages
Chinese (zh)
Other versions
CN103020252A (en
Inventor
呙维
范利伟
朱欣焰
李铭
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Wuhan joint space time Mdt InfoTech Ltd
Original Assignee
Wuhan University WHU
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Wuhan University WHU filed Critical Wuhan University WHU
Priority to CN201210558202.9A priority Critical patent/CN103020252B/en
Publication of CN103020252A publication Critical patent/CN103020252A/en
Application granted granted Critical
Publication of CN103020252B publication Critical patent/CN103020252B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Landscapes

  • Image Analysis (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The invention discloses a kind of remote sensing image demand fusion method based on demand characteristic association, comprise the association of remote sensing image demand characteristic, strong relevant subset division and the large step of Intelligent Fusion three.Remote sensing image demand characteristic associated steps mainly concentrates the correlativity between each demand to carry out modeling to demand according to original remote sensing image demand, thus obtains remote sensing image demand correlation model; Strong relevant subset partiting step is, according to remote sensing image demand correlation model, original remote sensing demand collection is divided into multiple demand subset to be fused; Intelligent Fusion step analyzes the feature of demand in each demand subset to be fused, picks out the applicable sensor that each demand subset to be fused is corresponding, complete Intelligent Fusion.The inventive method can be used for the Intelligent Fusion between multiple remote sensing image demand.

Description

A kind of remote sensing image demand fusion method based on demand characteristic association
Technical field
The invention belongs to remote sensing image to produce and process field, particularly relate to a kind of remote sensing image demand fusion method based on demand characteristic association.
Background technology
When user orders remote sensing image, propose demand by modes such as webpage order system, mail, fax or telephone counselings, data, services side generally first can check in product library whether there is this image, if existed, is then directly supplied to user; If there is no, be generally produce for each order.And in fact a lot of order has similarity in time, space and image parameters etc.Image demand as user A is: the multispectral image in Hubei Province in June, 2011 to October, spatial resolution requirements more than 10 meters; The image demand of user B is: Wuhan City's in August, 2011 image, spatial resolution requirements 5 meters.If directly produced according to order, planning twice may be needed, produce twice.In fact these two demands are through merger fusion calculation, can be described as a demand: the multispectral image of Hubei Province's in August, 2011 spatial resolution 5 meters, this image can meet the demand of user A and B simultaneously.After image has been produced, directly can be supplied to party A-subscriber and use; Calculate through spatial dimension, the image within the scope of Wuhan City is supplied to party B-subscriber.
Current needs control fusion is less, and especially remote sensing image demand fusion correlative study is almost blank, and only optimizing field at computer inquery has similar correlative study, mainly comprises query optimization and multi-user's parallel query.Query optimization refers to multiple different inquiry request to merge according to attribute similarity, but the characteristic item that current merging process adopts is only one dimension attribute information, and is not suitable for the remote sensing image demand with two-dimension time-space feature.Multi-user's parallel query is used for multiple online user search of database field, less SQL(Structured Query Language (SQL) is generated by carrying out Cluster merging to the relevant information in relation table simultaneously) inquiry, improve parallel processing efficiency, because object-oriented is different with application scenario, and be not suitable for the fusion of remote sensing image demand.
To set forth domestic and international correlative study present situation below.
1, the present Research of query optimization
Multiple queries processing is decomposed into two subtasks by dual stage process.First stage is single query optimization stage, and it for input, is single query execution plan that its generation is optimized with single query request.Subordinate phase is merging phase, and inquiry plan more than is merged in all single query optimization plan that the first stage generates by primary responsibility.First stage can utilize traditional directory optimizer to realize, and need not make any amendment.The search of subordinate phase can based on Cost Model, also can based on heuristic rule.Document [1]have studied as how dual stage process process two multi-join queries in multi-processor environment, share and parallelism in pipelining to utilize.Document [2]the inquiry plan of the optimization generated the first stage resolves into subquery, then arranges their execution sequence according to the various relations between inquiry, finally the result of each subquery is assembled into net result.Document [3]discuss and how to find in a complex query and to extract common subexpression, its basic step is, Optimizing Queries first in a conventional manner, generates the inquiry plan optimized; Then the inquiry plan of analysis optimization, by extracting common subexpression as instantaneous view and the operation utilizing instantaneous view rewritten query to eliminate wherein redundancy, and estimates its Executing Cost according to Cost Model, to determine finally whether adopt this plan.Document [4]what adopt is two benches dynamic optimization, first genesis sequence inquiry plan, then in scheduler task process, passes through heuritic approach, tries to find out and utilizes the operations such as Select, join, sort shared between different inquiry.The major advantage of dual stage process implements relatively simply.First stage is actually as multi-query optimization reduction search volume, and many inquiry plans are only searched for, to reduce the complicacy of optimization in the single inquiry plan optimized.But this method may lose some good prioritization schemes, because the many inquiry plans optimized not necessarily are made up of the single inquiry plan optimized.Another shortcoming of dual stage process is, if subordinate phase is based on cost, many work of first stage may be futile.
One-phase method and dual stage process cut both ways, and some papers start to explore modification method for this reason.As improved dual stage process, in the first phase, after obtaining single inquiry plan of the optimization of an inquiry, this single inquiry plan is converted, generate multiple single inquiry plan, using multiple single inquiry plan as the input of subordinate phase, expand the search volume of subordinate phase, thus add the possibility obtaining optimal planning.The method does not obviously inherently overcome the defect of dual stage process, the policy space simultaneously produced is unconfined to a great extent, the many strategies consequently do not optimized in a large number are retained, and searching algorithm is also had in the face of very large search volume.Also have and adopt the method for dividing and rule to carry out multiple queries processing.The policy space inquired about is divided into two parts by the method more, single query strategy of the local optimum that space is generated by traditional optimizer forms, and obtains the many inquiry plans of candidate by the single inquiry plan merging the local optimum generated by traditional directory optimizer; Another space is by all inquiries integrally, special alternative many query strategies for the generation of this entirety: finally with many inquiry plans that searching algorithm selects cost minimum from above-mentioned two spaces.
2, multi-user's Parallel query optimization present Research
The principal focal point of the research of parallel database query optimization is: how the relational database query having the more complicated that multiple connection (JOIN) operates is optimized.And the research of this query optimization is mainly manifested in following two aspects: how the model of query execution plan designs; Which algorithm query optimization has.For nothing-shared architecture pattern, take into full account that communication overhead provides cost estimation model and adopts two-phase optimization method, first sequential optimization is carried out to query tree according to cost estimation model, and a kind of new Two-phrase query optimisation strategy is proposed, parallelization is carried out to the inquiry plan of sequential optimization, takes full advantage of polyprocessor concurrency.
In recent years, along with the maturation with cheap and powerful multiple processor system of improving of parallel algorithms, make to adopt multiple processor system to carry out parallel processing OLAP(on-line analytical processing) inquire about the one preferred technique becoming current effective raising OLAP query handling property.Document [11]give the effective ways of the connection of a kind of parallel processing multidimensional and aggregation operator, but do not relate to the query aspects of OLAP.Document [12]discuss the formal definitions of OLAP query and the many OLAP query collection prioritization scheme based on semanteme in detail.Document [15-16]in indicate that multidimensional is expressed (MDX) and provided interface for OLAP multi-retrieving.Multidimensional Expressions the execution of multi-retrieving is optimized by various greedy algorithm and approximate data with usual.Document [14]propose three kinds of new attended operation symbols: the shared scanning for hash distance join is connected with the shared index for hash distance join, shared to solve multiple subtasks inquiry in OLAP.
Perhaps on the sunny side, Feng Yucai etc. have studied Semantic Query Optimization Method for Parallel [7].Semantic optimization refers to the inquiry problem of specifying, writes out different query statements, therefrom chooses DBMS(data base management system (DBMS)) statement of most efficient implementation method can be found for it [8].According to statistics, the conceptual schema of the regular Database Systems of may standing of semantic integrity describes 80% of content.A query transformation is become the inquiry of or several semantic equivalence by semantic query method, and then finds and perform an inquiry with better implementation strategy.
Can find out according to status both at home and abroad above, the user that existing research mainly lays particular emphasis on traditional database and online Web service field inquires about, and has a small amount of application.But remote sensing image demand is strongly professional as one, relate to time and space factor, baroque user's request simultaneously, pay close attention to less at present, also have nothing to do utility system temporarily that merge in remote sensing image demand.Therefore, groundwork of the present invention is by the incidence relation in analysis semantic frame between each characteristic item and similarity, realizes the intelligent method for fusing between multi-semantic meaning framework.
The list of references related in literary composition is as follows:
1.Kina-LeeTan,HonunLu:Workload Seheduling of Multiple Query Proeessing.IPL55(5):251-257,1995.
2.TimosK.Sellis，MultiPle-Query Optimization，ACM Trnasaction Database System，Vol.13，No.1，March1998，PP.23-52
3.SubbuN.Subrmanaina，ShivkaumarVenkatarmaan，Cost-Based OPtimiZation of DecisionSupport Queries using Trnasient-Views，SIGMOD1998，PP.319-330.
4.MnaishMehta,ValeyrSoloviev，dvaidJ.Dewitt,Batch Seheduling Parallel DatabaseSystems，gthInetationalConefreneeon Data Engineering，1993，PP.400-410.
5.Harald Kosch.Managing the operator ordering problem in parallel databases[J].FutureGeneration Computer Systems,2000,16:665-676.
6.Brunie L,Kosch H,Wohner W.From the modeling of parallel relational query processing toquery optimization and simulation[J].Parallel Processing Letters,2002,8(1):2-14.
7. permitted on the sunny side, Feng Yucai, Wang Yuanzhen. the application of semantic integrity in query optimization [J]. small-sized microcomputer system, 1996,17 (10): 35-39.
8. permitted on the sunny side. query semantics optimization [J]. micro computer and application, 1998 (3): 53-57.
9.Harald Kosch.Managing the operator ordering problem in parallel databases[J].FutureGeneration Computer Systems,2000,16:665-676.
10.Brunie L,Kosch H,Wohner W.From the modeling of parallel relational query processing toquery optimization and simulation[J].Parallel Processing Letters,2002,8(1):2-14.
11. Xue's immortality, Huang Zhenhua, Duan Jiangjiao, etc. the effective ways [J] of a kind of parallel processing multidimensional connection and aggregation operator. Journal of Computer Research and Development, 2004,41 (10): 1661-1669.
12. Yang Ke China. improve the research [D] of the some gordian techniquies of on-line analytical processing (OLAP) performance. Nanjing: Southeast China University, 2006
13. is high, Chen Rongguo, Zhao Yanqing, Yan Xun. spatial data accessing is integrated with distributed spatial data sources Object Query [J]. Earth Information Science journal, 2010, (4): 532-540.
14.Kalnis,Panos;Papadias,Dimitris.Multi-query optimization for on-line analyticalprocessing[J].Information System,2003(28):457-473.
15.Y.Zhao,P.M.Deshpande,J.F.Naughton,A.Shukla,Simultaneous optimization andevaluation ofmultiple dimension queries,Proceedings ofthe ACM-SIGMOD,1998.
16.W.Liang,M.E.Orlowska,J.X.Yu,Optimizing multiple dimensional queriessimultaneously in multidimensional databases,VLDB J.8(2000).
Summary of the invention
For the deficiency that prior art exists, the present invention is directed to structurized remote sensing image demand list, based on the demand characteristic association between multiple remote sensing image demand, propose a kind of fusion method of remote sensing image demand.
The present invention is by the relation between each linked character item of each demand entity in the remote sensing image demand list of analytical structure and similarity, and according to remote sensing image feature, select the linked character being used for Remote Sensing Image Fusion, realize the association of remote sensing image demand characteristic, thus complete the fusion of remote sensing image demand.
The technical solution used in the present invention is as follows:
Based on a remote sensing image demand fusion method for demand characteristic association, comprise step:
Step one, concentrates the correlativity between each demand and each demand to build demand correlation model according to original remote sensing image demand;
Step 2, original remote sensing demand collection is divided into multiple strong joint demand subset by correlation model according to demand, i.e. demand subset to be fused, in described strong joint demand subset, any demand is to being strong correlation demand pair, strong correlation demand is to the related needs pair being greater than setting threshold value for strength of correlation, and the strength of correlation that demand is right obtains specified remote sensing image area of space area and position relationship based on related needs;
Step 3, merges respectively to each strong joint demand subset.
Above-mentioned steps one comprises following sub-step further:
Based on demand characteristic, 1.1 judge that original remote sensing image demand concentrates each demand whether to be correlated with between any two, and to related needs to execution step 2), described demand characteristic comprises attribute sensor, remote sensing image shooting time and remote sensing image area of space;
1.2 obtain the right strength of correlation of related needs based on related needs to specified remote sensing image area of space area and position relationship;
1.3 build demand correlation model based on original remote sensing image demand collection and strength of correlation that wherein related needs is right.
Described attribute sensor comprises title, type, resolution, wave band, orbital fashion, imaging mode attribute further.
The strength of correlation ρ that described related needs is right s(a, b) is:
When related needs is adjacent to specified remote sensing image area of space, to intersect or when comprising mutually, the strength of correlation ρ that related needs is right s(a, b):
ρ s ( a , b ) = D ( A a + A b ) A smbr ,
Wherein,
D is constant, sets according to actual conditions;
A aand A brepresent the remote sensing image area of space area specified by related needs centering two demand respectively;
A smbrrepresent the outline area that the area of space specified by related needs centering two demand intersects;
When related needs to specified remote sensing image area of space from time, the strength of correlation ρ that related needs is right s(a, b):
ρ s(a,b)＝C(1-d/T)，
Wherein,
C is constant, sets according to actual conditions;
D represents the minimum distance between the area of space specified by related needs centering two demand;
The threshold value of the spacing of the area of space of T specified by related needs centering two demand.
T value can remote sensing image spatial resolution specified according to demand obtain, and is specially:
Because specified spatial resolution is a scope, T value is the intermediate value of spatial resolution scope common factor and the product of m kilometer of demand centering two demand, m is the estimated value according to actual conditions, wants demand to specified area of space as far as possible separately, the desirable smaller value of m.
Sub-step 1.3 is specially:
Concentrate each demand for node with original remote sensing image demand, representing line between the right node of strong correlation demand, and representing with the thickness of line the strength of correlation size that strong correlation demand is right, described strong correlation demand is to the related needs pair being greater than setting threshold value for strength of correlation.
Adopt greedy algorithm to carry out strong joint demand subset division to original remote sensing demand collection in step 2, comprise step:
The original remote sensing demand strong correlation demand of concentrating current relevance intensity maximum is to as the finite element in strong relevant subset; The strength of correlation in other demands and strong relevant subset between demand is concentrated according to original remote sensing demand, the demand meeting preset requirement is concentrated to be added in strong relevant subset original remote sensing demand, namely obtain a strong relevant subset, in the strong relevant subset of gained, any demand is to being strong correlation demand pair; Demand in strong for gained relevant subset is concentrated from original remote sensing demand and deletes, and concentrate remaining demand again to carry out strong joint demand subset division to original remote sensing demand, until the demand that original remote sensing demand is concentrated divides complete.
The one adopting greedy algorithm to carry out strong joint demand subset division is embodied as:
1) judge whether original remote sensing demand collection is empty, if it is empty, terminate strong relevant subset and divide; If not empty, then concentrate the strong correlation demand pair finding strength of correlation maximum from original demands, by this demand to adding current strong relevant subset, and concentrating this demand pair of deletion from original remote sensing demand, then performing step 2); If original remote sensing demand is concentrated and be can not find strong correlation demand pair, then remain node and form a strong relevant subset respectively, terminate strong relevant subset and divide;
2) concentrate to find from original remote sensing demand and form requirements set T with the demand of the equal strong correlation of all nodes in current strong relevant subset, perform step 3); If the requirements set T formed is empty, current strong relevant subset is now the strong relevant subset of gained, performs step 4) and carries out the strong relevant subset division of the next one;
3) to demand each in requirements set T, obtain the strength of correlation sum of all demands in itself and current strong relevant subset, demand maximum for strength of correlation sum in requirements set T is added current strong relevant subset, and concentrates this demand of deletion from original remote sensing demand, circulation performs step 2).
Fusion in above-mentioned steps three carries out based on the attribute sensor in strong relevant subset specified by demand, remote sensing image shooting time and remote sensing image area of space, be specially: according to demand each in strong relevant subset to the requirement of sensor, the sensor met the demands is picked out from sensor database, and according to the requirement of each demand to remote sensing image shooting time and remote sensing image area of space, obtain shooting time and the shooting area of space of sensor, then complete fusion.
The inventive method comprises the association of remote sensing image demand characteristic, strong relevant subset divides and the large step of Intelligent Fusion three.Remote sensing image demand characteristic associated steps mainly concentrates the correlativity between each demand to carry out modeling to demand according to original remote sensing image demand, thus obtains remote sensing image demand correlation model; Strong relevant subset partiting step is, according to remote sensing image demand correlation model, original remote sensing demand collection is divided into multiple demand subset to be fused; Intelligent Fusion step analyzes the feature of demand in each demand subset to be fused, picks out the applicable sensor that each demand subset to be fused is corresponding, complete Intelligent Fusion.
The association of remote sensing image demand characteristic is mainly by analyzing remote sensing image demand linked character, and computation requirement, to strength of correlation, sets up remote sensing image demand correlation model then, and its core is the strength of correlation that related needs is right.Sensor, remote sensing image shooting time and spatial dimension three aspects that correlativity between each demand is mainly asked from each demand judge, the correlativity namely between each demand comprises sensor correlativity, remote sensing image shooting time correlativity and area of space correlativity.Sensor correlativity comprises the correlativitys such as title, type, imaging mode, orbital fashion, resolution, wave band further.
Uncorrelated between the demand had in remote sensing image demand correlation model, in order to ensure that very positively related demand can merge, the present invention carries out strong relevant subset division to remote sensing image demand model, obtain strong joint demand subset to be fused.In strong joint demand subset, any demand is to related, and demand larger for strength of correlation is divided in a subset as far as possible.Ask the optimum solution of graph structure to be a np complete problem, adopt greedy algorithm to complete division here.
According to remote sensing image demand each in strong joint demand subset to the requirement of sensor, optimal sensor is picked out from sensor knowledge base, and according to the requirement in Time and place region, obtain the shooting time of sensor and coverage and number of times, then complete Intelligent Fusion work.
The present invention can be used for the Intelligent Fusion between multiple remote sensing image demand.Use the inventive method same or analogous original remote sensing image demand can be merged the new demand that merger is negligible amounts, meet new demand and just can meet all original remote sensing image demands simultaneously.The inventive method is used for supporting that remote sensing image planning is produced, and effectively can reduce production cost, reduce the wasting of resources, enhance productivity.
Accompanying drawing explanation
Fig. 1 is the inventive method process flow diagram;
Fig. 2 is remote sensing image demand correlation model schematic diagram;
Fig. 3 is that strong relevant subset divides schematic diagram.
Embodiment
To be described further the specific embodiment of the present invention below.
Input item in this concrete enforcement is structurized remote sensing image demand list, the original remote sensing image demand collection namely described in literary composition.Structurized remote sensing image demand list is a bivariate table, each record is a demand, the attribute of each demand comprises the attribute sensor, shooting time scope, area of space etc. of image, and attribute sensor also comprises the attributes such as title, type, imaging mode, orbital fashion, resolution, wave band further.
After adopting the remote sensing image demand list of the inventive method to input to merge, then export the remote sensing image demand after merging, and with the mapping relations of original remote sensing image demand of input.
Step one, remote sensing image demand characteristic associates.
With the formalized description that the nonoculture of structurized remote sensing image demand schedule is remote sensing image demand, using the attribute sensor specified by demand, shooting time and area of space as primary association feature, wherein, attribute sensor comprises the contents such as title, type, resolution, wave band, orbital fashion, imaging mode further.Study correlativity between each remote sensing image demand and strength of correlation based on above-mentioned linked character, to divide for follow-up strong relevant subset and Intelligent Fusion provides basic.
One) demand between correlativity judge
This sub-step needs all to carry out correlativity judgement to demand between two any in remote sensing image demand list.Below by for the arbitrary demand in remote sensing image demand list to (a, b), illustrate the right correlativity of demand based on every linked character and judge:
1, the correlativity of attribute sensor judges
Client, when proposing remote sensing image demand, can specify the remote sensing image captured by the sensor needing to have particular community.For the attribute sensor that client specifies, carry out correlativity judgement by the following method.
To simplify the process, in the correlativity of attribute sensor judges, if judged result is 0, then represent that demand is to the attribute sensor mutual exclusion of (a, b), namely attribute sensor is uncorrelated; If judged result is 1, then represent that demand is correlated with to the attribute sensor of (a, b).The correlativity of attribute sensor mainly judges from several angles such as sensor name, sensor type, imaging mode, orbital fashion, resolution and wave bands.Sensor type is divided into panchromatic, multispectral, EO-1 hyperion and synthetic aperture radar (SAR).
1) correlativity of sensor name judges
If demand specifies different sensor name to (a, b), then two demands can not be completed by same sensor, then demand is to the sensor name mutual exclusion of (a, b); Otherwise demand is correlated with to the sensor name of (a, b).If demand is to there being the non-specified sensor title of at least one demand in (a, b), also think that demand is correlated with to the sensor name of (a, b).
2) correlativity of sensor type judges
If demand does not exist common factor to the sensor type that (a, b) specifies, then demand is to the sensor type mutual exclusion of (a, b); Otherwise demand is correlated with to the sensor type of (a, b).If demand is to there being the non-specified sensor type of at least one demand in (a, b), also think that demand is correlated with to the sensor type of (a, b).
3) correlativity of imaging mode judges
Sensor imaging mode comprises push-broom type, sweep type etc., if demand specifies different imaging modes to (a, b), then demand is to the imaging mode mutual exclusion of (a, b); Otherwise demand is correlated with to the imaging mode of (a, b).If demand does not specify imaging mode to there being at least one demand in (a, b), also think that demand is correlated with to the imaging mode of (a, b).
4) correlativity of orbital fashion judges
The orbital fashion of sensor mainly contains sun synchronization and Geo-synchronous two kinds, if demand specifies different sensor track modes to (a, b), then demand is to the orbital fashion mutual exclusion of (a, b); Otherwise demand is correlated with to the orbital fashion of (a, b).If demand is to there being the non-track designation mode of at least one demand in (a, b), also think that demand is correlated with to the orbital fashion of (a, b).
In addition, according to orbital motion mode, for synthetic-aperture radar SAR, its orbital fashion can be divided into rail lift or fall rail.When the sensor type of specifying in demand is SAR, the orbital fashion correlativity of demand to (a, b) can be judged according to the orbital fashion of SAR.
5) correlativity of resolution judges
The resolution of sensor comprises spatial resolution, temporal resolution and spectral resolution, will illustrate respectively below.
A) correlativity of spatial resolution judges
Occur simultaneously if demand exists the spatial resolution scope specified by (a, b), then demand is correlated with to the spatial resolution of (a, b); Otherwise demand is to the spatial resolution mutual exclusion of (a, b).
Spatial discrimination rate dependence between demand judges to adopt following formula to represent:
κ r ( a , b ) = 1 a rv ∩ b rv = 1 0 - - - ( 1 )
Wherein,
κ r(a, b) represents the correlativity judged result of demand to the spatial resolution of (a, b), κ r(a, b)=1, then demand is correlated with to the spatial resolution of (a, b); κ r(a, b)=0, then demand is to the spatial resolution mutual exclusion of (a, b);
A rvexpression demand is to the spatial resolution scope specified by demand a in (a, b);
B rvexpression demand is to the spatial resolution scope specified by demand b in (a, b);
A rv∩ b rv=1 represents that the spatial resolution scope specified by demand a and b intersects.
If demand is to there being the non-designated space resolution of at least one demand in (a, b), also think that demand is correlated with to the spatial resolution of (a, b).
B) time resolution rate dependence judges
Occur simultaneously if demand exists the temporal resolution scope specified by (a, b), then demand is correlated with to the temporal resolution of (a, b); Otherwise demand is to the temporal resolution mutual exclusion of (a, b).
Time resolution rate dependence between demand judges to adopt following formula to represent:
κ rt ( a , b ) = 1 a tv ∩ b tv = 1 0 - - - ( 2 )
Wherein,
κ rt(a, b) represents the correlativity judged result of demand to the temporal resolution of (a, b), κ rt(a, b)=1, then demand is correlated with to the temporal resolution of (a, b); κ rt(a, b)=0, then demand is to the temporal resolution mutual exclusion of (a, b);
A tvexpression demand is to the temporal resolution scope specified by demand a in (a, b);
B tvexpression demand is to the temporal resolution scope specified by demand b in (a, b);
A tv∩ b tv=1 represents that the temporal resolution scope specified by demand a and b intersects.
If demand is to there being the non-fixed time resolution of at least one demand in (a, b), also think that demand is correlated with to the temporal resolution of (a, b).
C) spectrally resolved rate dependence judges
Spectral resolution is significant for EO-1 hyperion sensor; For not high spectrum sensor, meaning is less.
Occur simultaneously if demand exists the spectral resolution scope specified by (a, b), then demand is correlated with to the spectral resolution of (a, b); Otherwise, the spectral resolution mutual exclusion that this demand is right.
The spectrally resolved rate dependence that demand is right judges to adopt following formula to represent:
κ p ( a , b ) = 1 a pv ∩ b pv = 1 0 - - - ( 3 )
Wherein,
κ p(a, b) represents the correlativity judged result of demand to the spectral resolution of (a, b), κ p(a, b)=1, then demand is correlated with to the spectral resolution of (a, b); κ p(a, b)=0, then demand is to the spectral resolution mutual exclusion of (a, b);
A pvexpression demand is to the spectral resolution scope specified by demand a in (a, b);
B pvexpression demand is to the spectral resolution scope specified by demand b in (a, b);
A pv∩ b pv=1 represents that the spectral resolution scope specified by demand a and b intersects.
If demand does not specify spectral resolution to there being at least one demand in (a, b), also think that demand is correlated with to the spectral resolution of (a, b).
6) wave band correlativity judges
If remote sensing image demand to the same image that specified wave band can be taken by same sensor cover, show that these two demands can be met by same sensor, otherwise, if there is no any one sensor can cover the wavelength band of these two demands simultaneously, then these two demands are uncorrelated on wave band.
According to above-mentioned analysis, demand judges to adopt following formula to represent to the wave band correlativity of (a, b):
Wherein,
κ b(a, b) represents the wave band correlativity judged result of demand to (a, b), κ b(a, b)=1, then demand is correlated with to the wave band of (a, b); κ p(a, b)=0, then demand is to the wave band mutual exclusion of (a, b);
A waand a wbrepresent a object and the wave band specified by b object respectively;
W representative sensor;
represent that existence sensor can cover the wavelength band specified by demand a and b in sensor database.Here sensor database is database that built, that comprise various types of sensor relevant information (such as title, type, resolution etc.).
If demand is to there being the non-designated band of at least one demand in (a, b), also think that demand is correlated with to the wave band of (a, b).
2, the correlativity of remote sensing image shooting time judges
Some remote sensing application occasion to the requirement of filming image time and out of true limit, in certain time range, the image of shooting all can meet the demands, therefore the time ambiguity of this image demand can be utilized, multiple image demand to be fused is carried out merger according to its tolerable time span, and the demand that realizes merges.Tolerable time range refers to the acceptable time range of user, and the scope that may more originally than user specify is little, but still can meet consumers' demand.
First, the present invention will be divided into the time year, half a year, season, the moon, week and day six time scales.The different shooting times specified by image demand may there are differences on yardstick, and different time scales often reflects that the application target of demand a and b to image is different.
When the time scale that demand is right is identical or adjacent, right shooting time yardstick is correlated with to think demand; Otherwise, uncorrelated.As demand a needs the image within month, and demand b needs the image within a year, and the time scale of demand a and b is uncorrelated.To be divided into the time in the present invention year, half a year, season, the moon, week and day six time scales, for " half a year " yardstick, year, season and half a year belong to adjacent time yardstick.When two tolerable shooting time scopes of demand intersect, and both time scales are identical or adjacent, then think that this is correlated with to the shooting time of demand.
Based on above-mentioned analysis, demand judges to adopt following formula to represent to the shooting time correlativity of (a, b):
Wherein,
κ t(a, b) represents the shooting time correlativity judged result of demand to (a, b), κ t(a, b)=1, then demand is correlated with to the shooting time of (a, b); κ t(a, b)=0, then demand is to the shooting time mutual exclusion of (a, b);
A tvand b tvrepresent the shooting time scope specified by demand a and b respectively;
A tsand b tsrepresent the shooting time yardstick specified by demand a and b respectively;
A tv∩ b tvwhether the shooting time scope represented specified by demand a and b intersects, if intersect, then a tv∩ b tv=1;
| a ts-b ts|≤1 to represent demand a identical or adjacent with the shooting time yardstick specified by b.
3, area of space correlativity judges
Judgement demand between area of space correlativity time, need to consider space scale factor, the difference between the area of space area namely specified by demand.Such as, when user specifies the whole nation and the image in certain county, often different to the requirement of sensor.
This is concrete implement in based on space scale judge demand between spatial coherence, two area of space that general area discrepancy is less, spatial resolution requirements is often more similar, thinks that its space scale is more relevant.Demand between space scale is relevant adopts following formula to judge:
ρ ( a , b ) = 1 S a / S b ≥ E 0 - - - ( 6 )
Wherein:
ρ (a, b) represents the multi-scale spatial relationship judged result of demand to (a, b), if ρ (a, b)=1, then demand is correlated with to the space scale of (a, b); If ρ (a, b)=0, then the space scale of demand to (a, b) is uncorrelated;
Sa and Sb is respectively the area of demand to the area of space specified by demand a and b in (a, b);
E represents the threshold value of the business of the area of space area specified by demand a and b, and when Sa/Sb is more than or equal to E, then demand is correlated with to the space scale of (a, b), otherwise, uncorrelated.E, according to actual conditions value, is taken as 50 in this concrete enforcement.
In actual applications, the demand in remote sensing image demand list not necessarily comprises above-mentioned all linked characters completely, to demand to (a, b), as long as wherein at least one demand does not comprise certain feature above-mentioned, then give tacit consent to demand and this feature of (a, b) is correlated with.
When demand is to (a, b) when above-mentioned all features (sensor name, type, imaging mode, orbital fashion, temporal resolution, spatial resolution, spectral resolution, wave band, shooting time, area of space) are related, then demand is related needs pair to (a, b).
Two) the spatial coherence tolerance that related needs is right
Last sub-step carries out correlativity judgement respectively to the right attribute sensor of demand, remote sensing image shooting time, area of space, and when all correlation properties are related, then this demand is to relevant.This sub-step is then carry out relativity measurement to obtain the quantitative description of related needs to the strength of correlation of (a, b) to related needs to (a, b).
It is considered herein that the spatial coherence tolerance of related needs to (a, b) should meet two rules below:
1) distance dependency rule: the spatial object that distance is larger, strength of correlation is lower;
2) area dependency rule: when apart from time close, the strength of correlation between spatial object and the larger object of area is larger.
Based on above-mentioned rule, the present invention adopts following formula to measure the right spatial coherence size of related needs:
ρ s ( a , b ) = D ( A a + A b ) A smbr - - - ( 7 )
Wherein,
ρ s(a, b) represents that related needs is to the spatial coherence size between (a, b), and namely related needs is to the strength of correlation of (a, b);
D is constant, and in this concrete enforcement, D gets 0.5;
A arepresent the area of related needs to the area of space specified by demand a in (a, b);
A brepresent the area of related needs to the area of space specified by demand b in (a, b);
A smbrrepresent the outline area that related needs intersects the area of space specified by demand a and demand b in (a, b).
When two area of space are apart from time far, think that they are space-independent, therefore, need for ρ s(a, b) value specifies a threshold value, works as ρ swhen (a, b) is less than threshold value, uncorrelated in two demand spaces.But, due to ρ s(a, b) value relates to area and the distance of area of space, and this threshold value is difficult to define, and namely allows to determine a threshold value, this formula be also not suitable for measuring from the space correlation relation of two area of space.Such as, two at a distance of nearer little area of space, the spatial coherence size calculated is the same with the spatial coherence size in two large space regions apart from each other.Usually we think that nearer little area of space correlativity should be greater than large space region apart from each other apart.Therefore, two area of space from time be not suitable for using above-mentioned computing formula (7).
For this reason, when area of space is adjacent, intersect or when comprising, use above-mentioned formula (7) to calculate related needs to the spatial coherence size of (a, b); When area of space from time, use below formula (8) calculate related needs to the spatial coherence size of (a, b):
ρ s(a,b)＝C(1-d/T)（8）
Wherein:
ρ s(a, b) represents that related needs is to the spatial coherence size between (a, b), and namely related needs is to the strength of correlation of (a, b);
C is constant, and in this concrete enforcement, C gets 0.5;
D represents that related needs is to (a, minimum distance between the area of space b) specified by demand a and b, specified area of space can regard polygon as, each searching node on two polygon outer contours, make this shortest to the distance of node, this bee-line is the minimum distance between two area of space;
The threshold value of the spacing of the area of space of T specified by demand a and b.
T value can be specified according to demand remote sensing image spatial resolution calculate, be specially:
Because specified spatial resolution is a scope, T value is the intermediate value of spatial resolution scope common factor and the product of m kilometer of demand centering two demand, m is the estimated value according to actual conditions, wants demand to specified area of space as far as possible separately, the desirable smaller value of m.
The value of T can adopt formula to be described below:
For demand to (a, b), if the spatial resolution scope that demand a specifies is [a 1, a 2], the spatial resolution scope that demand b specifies is [b 1, b 2], then T=(min (a 2, b 2)+max (a 1, b 1))/2*m kilometer.
In formula (7) and (8), the value of C and D will keep a balance, namely wants to reflect that area of space is adjacent, intersects, comprises the correlativity with separating situation, guarantee area of space is adjacent, intersect, correlativity when comprising be greater than from correlativity.Therefore, C and D value can according to actual conditions reasonable value, and in this concrete enforcement, C and D value all gets 0.5.
Three) remote sensing image demand correlation model is built
Based on the demand between correlativity judged result and the right spatial coherence intensity of related needs, build remote sensing image demand correlation model.Constructed demand correlation model can be graph model.Node in graph model represents demand entity, the line of connected node represents the correlativity between demand, when two demands are uncorrelated or spatial coherence intensity is less than setting threshold value, wireless connections between two demands, namely, just line between two demands only with certain strength of correlation, hereinafter, will the relevant and strength of correlation demand that is greater than setting threshold value to referred to as " strong correlation demand to "; The line of employing different thicknesses distinguishes the strength of correlation size between demand.Originally the remote sensing image demand correlation model constructed by concrete enforcement can see Fig. 1.
Remote sensing image demand correlation model G also can formalization representation as follows:
Demand correlation model G is a non-directed graph, is made up of, is designated as: G=(V, E) set V and E;
Wherein:
V is that the finite nonempty set of summit Node in demand associated diagram model closes, and namely represents the set of each demand entity;
E=(a, b), be the finite aggregate of summit couple (being called limit) in V, namely represent the set that strong correlation demand is right;
P (E) represents the limit weight of strong correlation demand to (a, b), and its value is ρ s(a, b), ρ s(a, b) is for related needs is to the strength of correlation of (a, b).
Step 2, strong relevant subset divides
When demand to be fused is more, a demand may occur with N number of demand related, and exists uncorrelated between this N number of demand.Therefore, in order to improve the efficiency of follow-up fusion, original demands set is divided into multiple strong relevant subset by the demand correlation model that the present invention is based on constructed by step one, i.e. demand subset to be fused, in strong relevant subset, any pair demand is strong correlation demand pair.
If H is original demands collection, the strong relevant subset of Q non-overlapping copies can be divided into, be expressed as:
H＝G 1∪G 2∪...∪G Q
G m∩ G n=Φ m=1,2 ..., Q, n=1,2 ..., Q, and m ≠ n
Based on remote sensing image demand correlation model in this concrete enforcement, greedy algorithm is utilized to carry out strong relevant subset division.First, using strong correlation demand maximum for current relevance intensity in original demands set to as the finite element in strong relevant subset; Then, according to the incidence relation of the demand in strong relevant subset in demand correlation model, by correlativity judged result and strength of correlation, in strong relevant subset, add satisfactory node (i.e. demand), until add fashionable without node, a strong relevant subset can be obtained.Said process is repeated to the unmet demand not being added to strong relevant subset in original demands set, finally completes strong relevant subset and divide.Strong relevant subset division result in this concrete enforcement can see Fig. 2.
Elaborate to the strong relevant subset partition process of this step below.
This process is with step one gained remote sensing image demand correlation model G(V, E) be input, after dividing, export the set Z be made up of each strong relevant subset.Detailed process is as follows:
1) judge demand G(V, E) in node be whether empty, if it is empty, then terminate strong relevant subset and divide; If not empty, then the strong correlation node finding limit weight maximum from set G is to N 1and N 2, by this node to adding current strong relevant subset Q, and deleting this node pair from set G, then performing step 2), if the strong correlation node that in set G, limit weight is maximum is to more than a pair, then optionally wherein add strong relevant subset Q for a pair; If can not find strong correlation node pair in set G, then remain node and form a strong relevant subset respectively, terminate strong relevant subset and divide;
2) forming node set T from gathering G to find with the node gathering the equal strong correlation of all nodes in Q, performing step 3); If the node set T formed is empty, then perform step 4);
3) to node each in node set T, obtain itself and the limit weight sum of all nodes in set Q, node maximum for limit weight sum in node set T is added set Q, and delete this node from set G, circulation performs step 2);
4) Q set is added set Z, and delete all limits that is end points with Q interior joint in G, perform step 1).
Step 3, remote sensing image demand Intelligent Fusion
This step first with remote sensing image demand correlativity for Main Basis, on room and time, there is overlap according to remote sensing image demand, sensor parameters requires to there is the features such as compatible, mechanism, process and correlation processing technique that research remote sensing image demand merges, from space, time and sensor different angles realize demand and merge.
The fusion of this step carries out for the strong relevant subset of step 2 gained, and this step mainly comprises following sub-step:
1, using the remote sensing image demand in strong relevant subset as a collective, its requirement to area of space, shooting time and sensor is analyzed.By analyzing the feature of the feature of remote sensing image and the display requirement of remote sensing image request and implicit requirement, in this embodiment, each strong relevant subset merges as follows according to the requirement of area of space, remote sensing image shooting time and sensor:
1) fusion of area of space
The union of the area of space of the area of space after fusion specified by original demands each in strong relevant subset.Such as, two demands in strong relevant subset, user UA and UB asks the remote sensing image in Hubei Province and Hunan Province respectively, the fusion results of area of space is the union in Hubei Province and Hunan Province, like this, the remote sensing image after shooting, through cutting, can be supplied to user UA and UB respectively.
2) fusion of shooting time
The common factor of the shooting time scope of the time range after fusion specified by original demands each in strong relevant subset.Such as, two demands in strong relevant subset, user UA asks 2012 6, the image in July, user UB asks 2012 7, the image in August, both time ranges are correlated with, yardstick is compatible, and the time range after fusion is in July, 2012, and shooting time is the demand that the remote sensing image in July, 2012 can meet user UA and UB simultaneously.
3) fusion of spatial resolution
The common factor of the spatial resolution of the spatial resolution after fusion specified by original demands each in strong relevant subset.Such as, two demands in strong relevant subset, user UA request spatial resolution is the image of 1-10 rice, user UB request spatial resolution is the image of 5-15 rice, the then remote sensing image of the fusion results of above-mentioned spatial resolution to be spatial resolution be 5-10 rice, then take the image that resolution is 5-10 rice, the requirement of user UA and UB to spatial resolution can be met simultaneously.
4) fusion of spectral resolution
The common factor of the spectral resolution of the spectral resolution after fusion specified by original demands each in strong relevant subset.Such as, two demands in strong relevant subset, user UA needs spectral resolution to be the image of 1000-2000nm, user UB needs the image of 1000-1400nm, the then remote sensing image of the fusion results of above-mentioned spectral resolution to be spectral resolution be 1000-1400nm, then the spectral resolution of taking is that the image of 1000-1400nm can be supplied to user UA and UB and use.
5) fusion of temporal resolution
The common factor of the temporal resolution of the temporal resolution after fusion specified by original demands each in strong relevant subset.Such as, two demands in strong relevant subset, user UA needs to take weekly an image, and user UB needs within 5-10 days, to take an image, then can meet user UA and UB when the temporal resolution of filmed image is 5-7 days simultaneously.
6) fusion of wave band
The union of the wave band specified by original demands each in strong relevant subset of the wave band after fusion.Such as, user UA ask image at least comprise near infrared and middle-infrared band, user UB ask image at least comprise short-wave infrared and far infrared, then shooting after image at least should comprise near infrared, in infrared, short-wave infrared and these four wave bands of far infrared.
7) fusion of sensor type
The common factor of the sensor type of the sensor type after fusion specified by original demands each in consistent subset.Such as, user UA asks panchromatic or multispectral image, and user UB asks image that is multispectral or EO-1 hyperion, then their common factor multispectral image can meet user UA and UB simultaneously.
8) fusion of orbital fashion and imaging mode
For orbital fashion and imaging mode, can learn according to sensor as aforementioned Attribute Correlation judgment rule, sensor name in same strong relevant subset specified by each original demands, orbital fashion and imaging mode are the same certainly, so, sensor name, orbital fashion and the fusion results of imaging mode and identical specified by original remote sensing image demand.
2, sensor is selected according to fusion results
According to fusion results, from sensor database, pick out the sensor that can meet fusion results, be the sensor that remote sensing image that actual photographed meets fusion results uses.So far, Intelligent Fusion process of the present invention is complete.Here sensor database is database that built, that comprise various types of sensor relevant information (such as title, type, resolution etc.).

Claims (6)

1., based on a remote sensing image demand fusion method for demand characteristic association, it is characterized in that, comprise step:
Step one, build demand correlation model according to the correlativity between each demand of original remote sensing image requirements set and each demand, this step comprises following sub-step further:
Based on demand characteristic, 1.1 judge that original remote sensing image demand concentrates each demand whether to be correlated with between any two, and to related needs to execution step 1.2, described demand characteristic comprises attribute sensor, remote sensing image shooting time and remote sensing image area of space;
1.2 obtain the right strength of correlation of related needs based on related needs to specified remote sensing image area of space area and position relationship; The strength of correlation ρ that described related needs is right s(a, b) is:
When related needs is adjacent to specified remote sensing image area of space, to intersect or when comprising mutually, the strength of correlation ρ that related needs is right s(a, b):
ρ s ( a , b ) = D ( A a + A b ) A s m b r ,
Wherein,
D is constant, sets according to actual conditions;
A aand A brepresent the remote sensing image area of space area specified by related needs centering two demand respectively;
A smbrrepresent the area of the union of the area of space specified by related needs centering two demand;
When related needs to specified remote sensing image area of space from time, the strength of correlation ρ that related needs is right s(a, b):
ρ s(a,b)＝C(1-d/T)，
Wherein,
C is constant, sets according to actual conditions;
D represents the minimum distance between the area of space specified by related needs centering two demand;
The threshold value of the spacing of the area of space of T specified by related needs centering two demand;
1.3 build demand correlation model based on original remote sensing image demand collection and strength of correlation that wherein related needs is right;
Step 2, original remote sensing demand collection is divided into multiple strong joint demand subset by correlation model according to demand, i.e. demand subset to be fused, in described strong joint demand subset, any demand is to being strong correlation demand pair, strong correlation demand is to the related needs pair being greater than setting threshold value for strength of correlation, and the strength of correlation that demand is right obtains specified remote sensing image area of space area and position relationship based on related needs;
Step 3, merges respectively to each strong joint demand subset.
2., as claimed in claim 1 based on the remote sensing image demand fusion method of demand characteristic association, it is characterized in that:
Described attribute sensor comprises title, type, resolution, wave band, orbital fashion, imaging mode attribute further.
3., as claimed in claim 1 based on the remote sensing image demand fusion method of demand characteristic association, it is characterized in that:
Described T value can remote sensing image spatial resolution specified according to demand obtain, and is specially:
T value is the intermediate value of spatial resolution scope common factor and the product of m kilometer of demand centering two demand, and m is the estimated value according to actual conditions.
4., as claimed in claim 1 based on the remote sensing image demand fusion method of demand characteristic association, it is characterized in that:
Described build demand correlation model based on original remote sensing image demand collection and strength of correlation that wherein related needs is right, be specially:
Concentrate each demand for node with original remote sensing image demand, representing line between the right node of strong correlation demand, and representing with the thickness of line the strength of correlation size that strong correlation demand is right, described strong correlation demand is to the related needs pair being greater than setting threshold value for strength of correlation.
5., as claimed in claim 1 based on the remote sensing image demand fusion method of demand characteristic association, it is characterized in that:
Adopt greedy algorithm to carry out strong joint demand subset division to original remote sensing demand collection in step 2, comprise step:
The original remote sensing demand strong correlation demand of concentrating current relevance intensity maximum is to as the finite element in strong relevant subset; The strength of correlation in other demands and strong relevant subset between demand is concentrated according to original remote sensing demand, the demand meeting preset requirement is concentrated to be added in strong relevant subset original remote sensing demand, namely obtain a strong relevant subset, in the strong relevant subset of gained, any demand is to being strong correlation demand pair; Demand in strong for gained relevant subset is concentrated from original remote sensing demand and deletes, and concentrate remaining demand again to carry out strong joint demand subset division to original remote sensing demand, until the demand that original remote sensing demand is concentrated divides complete.
6., as claimed in claim 1 based on the remote sensing image demand fusion method of demand characteristic association, it is characterized in that:
Described fusion carries out based on the attribute sensor in strong relevant subset specified by demand, remote sensing image shooting time and remote sensing image area of space;
According in fusion results to the requirement of sensor, the sensor met the demands is picked out from sensor database, and according in fusion results to the requirement of remote sensing image shooting time and remote sensing image area of space, obtain shooting time and the shooting area of space of sensor, then complete fusion.
CN201210558202.9A 2012-12-20 2012-12-20 A kind of remote sensing image demand fusion method based on demand characteristic association Active CN103020252B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201210558202.9A CN103020252B (en) 2012-12-20 2012-12-20 A kind of remote sensing image demand fusion method based on demand characteristic association

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201210558202.9A CN103020252B (en) 2012-12-20 2012-12-20 A kind of remote sensing image demand fusion method based on demand characteristic association

Publications (2)

Publication Number Publication Date
CN103020252A CN103020252A (en) 2013-04-03
CN103020252B true CN103020252B (en) 2015-09-02

Family

ID=47968855

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201210558202.9A Active CN103020252B (en) 2012-12-20 2012-12-20 A kind of remote sensing image demand fusion method based on demand characteristic association

Country Status (1)

Country Link
CN (1) CN103020252B (en)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103336715B (en) * 2013-07-03 2016-04-27 邢立宁 Based on the imaging satellites scheduling problem model construction method of body
CN104516975B (en) * 2014-12-29 2019-03-22 中国科学院电子学研究所 Automatic correlation method towards multivariate data
CN105243145B (en) * 2015-10-16 2018-07-27 武汉大学 Task based access control and the remote sensing image of sensor Typical Disposition find method and system
CN109308300B (en) * 2018-09-27 2021-11-12 上海达梦数据库有限公司 Logic operation processing method and device, conversion plug-in and storage medium
CN113610406A (en) * 2021-08-11 2021-11-05 北京市遥感信息研究所 Remote sensing satellite demand fusion processing method
CN117522023A (en) * 2023-11-07 2024-02-06 中国人民解放军战略支援部队航天工程大学 Remote sensing image demand fusion method based on graph model

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1770201A (en) * 2004-11-05 2006-05-10 北京师范大学 Adjustable remote sensing image fusion method based on wavelet transform
CN102314677A (en) * 2010-07-07 2012-01-11 中国科学院地理科学与资源研究所 Classification-based high time definition and high space definition remote sensing data quantitative fusing method

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5808916A (en) * 1994-08-04 1998-09-15 City Of Scottsdale Method for monitoring the environment

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1770201A (en) * 2004-11-05 2006-05-10 北京师范大学 Adjustable remote sensing image fusion method based on wavelet transform
CN102314677A (en) * 2010-07-07 2012-01-11 中国科学院地理科学与资源研究所 Classification-based high time definition and high space definition remote sensing data quantitative fusing method

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
冷猛.卫星对地观测需求分析方法及其应用研究.《中国优秀硕士学位论文全文数据库电子期刊》.2012, *
冷猛等.基于完全图的对地电磁探测需求融合分析方法.《航天电子对抗》.2011,第27卷(第6期), *
刘异等.一种基于云计算模型的遥感处理服务模式研究与实现.《计算机应用研究》.2009,第26卷(第9期), *

Also Published As

Publication number Publication date
CN103020252A (en) 2013-04-03

Similar Documents

Publication Publication Date Title
CN103020252B (en) A kind of remote sensing image demand fusion method based on demand characteristic association
Tan et al. Enabling query processing across heterogeneous data models: A survey
EP3740880A1 (en) Pick and applicator for use with a stringed instrument
Mami et al. A survey of view selection methods
US10120902B2 (en) Apparatus and method for processing distributed relational algebra operators in a distributed database
Umbrich et al. Comparing data summaries for processing live queries over linked data
Gidófalvi et al. Mining long, sharable patterns in trajectories of moving objects
Huang et al. Query optimization of distributed pattern matching
CA2534788C (en) Dml statements for densifying data in a relational database system
CN104699698A (en) Graph query processing method based on massive data
JP2017539012A (en) Multiple query optimization in SQL-on-Hadoop system
Zimányi et al. MobilityDB: A mainstream moving object database system
CN115237937A (en) Distributed collaborative query processing system based on interplanetary file system
Ghazal et al. Adaptive optimizations of recursive queries in teradata
Bordogna et al. A flexible framework to cross-analyze heterogeneous multi-source geo-referenced information: The J-CO-QL proposal and its implementation
Alsubaiee et al. Asterix: scalable warehouse-style web data integration
Raghavan et al. Skyline and mapping aware join query evaluation
CN115630964A (en) Construction method of high-dimensional private data-oriented correlation data transaction framework
CN112883063A (en) SPARQL query processing method on partition-based distributed RDF system
Pagare et al. Recommendation system using bloom filter in mapreduce
Troumpoukis et al. A Geospatial Join Optimization for Federated GeoSPARQL querying.
Camacho-Rodríguez et al. PigReuse: A Reuse-based Optimizer for Pig Latin
Papaioannou et al. Outer and anti joins in temporal-probabilistic databases
Schneider et al. SimDataMapper: An Architectural Pattern to Integrate Declarative Similarity Matching into Database Applications.
Zhu et al. Hydb: Access optimization for data-intensive service

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
C41 Transfer of patent application or patent right or utility model
TR01 Transfer of patent right

Effective date of registration: 20170227

Address after: The East Lake new technology development zone of Hubei Province, Wuhan City, 430075 East Road No. 4 Building 19 layer 1 WITHUB space

Patentee after: Wuhan joint space time Mdt InfoTech Ltd

Address before: 430072 Hubei Province, Wuhan city Wuchang District of Wuhan University Luojiashan

Patentee before: Wuhan University