CN109543612A - A kind of Remote Sensing Data Processing method based on more granularities - Google Patents
A kind of Remote Sensing Data Processing method based on more granularities Download PDFInfo
- Publication number
- CN109543612A CN109543612A CN201811400783.7A CN201811400783A CN109543612A CN 109543612 A CN109543612 A CN 109543612A CN 201811400783 A CN201811400783 A CN 201811400783A CN 109543612 A CN109543612 A CN 109543612A
- Authority
- CN
- China
- Prior art keywords
- data
- screening
- remote sensing
- degree
- fine grain
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Remote Sensing (AREA)
- Astronomy & Astrophysics (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
A kind of Remote Sensing Data Processing method based on more granularities, comprising the following steps: 1, together by time-sequencing unified integration by remote sensing, and label is added according to data pick-up source；2, input Expert Rules data carry out Granule Computing sifting sort as granulation standard；3, label is counted, coarse grain data are divided by number of tags according to data；4, to the data in step 3 using existing clear remotely-sensed data as Rules Filtering；5, secondary Granule Computing sifting sort data are obtained from step 4 screening, turn the screening rule in step 4, carry out repeating screening in the case where not changing data；6, the convenient data screening next time of screening rule according to Fine Grain Data feature, in amendment step 5.Existing a large amount of remotely-sensed datas are split as to obtain categorizing information based on grain classification method, and all remaining valuable information are obtained by Granule Computing after finding useful remote sensing information, with in other classification to homogeneous data, facilitate the processing of follow-up data.
Description
Technical field
The present invention is Remote Sensing Data Processing method, and in particular to a kind of Remote Sensing Data Processing method based on more granularities.
Background technique
In recent ten years, Hi-spatial resolution remote sensing image is widely used for agricultural, forestry, ocean and environmental monitoring etc.
Field has huge economic value and social benefit.However, due to the scale of construction (Volume) of Hi-spatial resolution remote sensing image
Greatly, data type (Variety) is more, abundant information, and interpretation analysis process is complicated, and it is accurately and efficiently right to be also difficult to so far
Hi-spatial resolution remote sensing image carries out automatic terrain classification.How atural object point is carried out to high spatial resolution remote sense big data
Class becomes one of the technological difficulties for influencing its large-scale application and bottleneck.Compared with middle low resolution remote sensing images, high spatial point
Resolution Remote Sensing Image Texture is more abundant, shape is more obvious, and spatial relationship is more complicated.Existing technology usually uses light
The different characteristics of spectrum, shape and textural characteristics to describe in Hi-spatial resolution remote sensing image class.However, these are characterized in bottom
Layer feature, it is difficult to describe the geometry and structural information of atural object in high spatial resolution images comprehensively.In recent years, text analyzing and
Word packet model (Bag-of-Word, BOW) and topic model (Topic model) in scene understanding are introduced into remote sensing fields.This
A little methods extract the statistical information or semantic information of local feature by word packet model, and analyze high spatial resolution accordingly
Theme in remote sensing images, to achieve the purpose that classification.Existing feature extracting method is statistical nature mostly, it is difficult to accurate
The essential information of ground description ground class, it is difficult to realize Hi-spatial resolution remote sensing image automatic interpretation.How for the more of sensor
The complexity of sample, the variability of image-forming condition and ground target extracts atural object in high spatial resolution remote sense big data
Deep structure information imperfectly describes ground class feature as far as possible, is the terrain classification in high spatial resolution remote sense big data
It is crucial.In order to excavate better feature, people, which have to put into a large amount of energy, goes one good feature of research.And good feature
Exploitation generally requires the understanding for having very deep to problem, needs to grope repeatedly.Therefore it is required to automatically generate suitable spy instantly
Sign.
Summary of the invention
The present invention be directed to existing remotely-sensed data data volume is big, it is difficult to find useful information and one kind for providing be based on it is more
Existing a large amount of remotely-sensed datas can be split as obtaining sorting out letter based on grain classification method by the Remote Sensing Data Processing method of granularity
Breath, and all remaining valuable information are obtained by Granule Computing after finding useful remote sensing information, and according to algorithm
Statistics facilitates the processing of follow-up data with into other classification of homogeneous data.
A kind of Remote Sensing Data Processing method based on more granularities, comprising the following steps:
M1 together by time-sequencing unified integration by remotely-sensed data, and adds label according to data pick-up source；
M2 inputs Expert Rules data as granulation standard, carries out a Granule Computing sifting sort；
M3 is that each data block adds the specific label screened every time by screening；
M4 counts label included by coarse grain, and from the classification data of step M3, divides coarse grain number by number of tags according to data
According to；
M5, to the data in step M4 using existing clear remotely-sensed data as Rules Filtering；
M6 is screened to obtain secondary Granule Computing sifting sort data from step M5, is turned the screening rule in step M5, do not change number
It carries out repeating screening in the case where；
M7 repeats step M5, M6 several times, the coarse grain data after being refined, and according to the secondary grain in final step M6
Calculating sifting classification data, determines Fine Grain Data；
M8, the convenient data screening next time of screening rule according to Fine Grain Data feature, in amendment step M6.
Wherein the Fine Grain Data Feature Selection mode of step M8 is Effective selection reference mode, can be similar remote sensing number
There is provided a screening easy way according to screening, can mode simultaneously and concurrently carry out the calculating of step M5 accelerating step M6
Journey, while step M6 is calculated as to the granulated key of data, and data constraint can be accelerated in other way,
Only it can also be accumulated by folk prescription formula data and obtain Effective selection mode.
Preferably, in the step M2, including following sub-step: MA1 sets P Expert Rules, each expert
Rule all includes target property U, with property feature V；
Target property U is split as r gradient and is collected (U1, U2, U3, U4 ... Ur), similarly obtains property feature V's by MA2
Collect (V1, V2, V3, V4 ... Vc)
MA3 obtains set N=∑ijnijWherein N is that the total characteristic of single Expert Rules and total characteristic type close i and belong to r, and j belongs to
In c；
MA4 is calculatedWherein ai characteristic Ui is the sum of the concentration of V in feature, and bj is characterized
Sum of the Vj in the concentration that feature is U
MA5 calculates target property function according to transfer operator R obtained in step MA4
MA6, according to transfer operator R estimated performance characteristic function obtained in step MA4
MA7 classifies to data according to the function of step MA5 and step MA6.
Preferably, in the step M3, including following sub-step:
MB1 resettles data mapping tables according to the result in step M2；
MB 2 adds repeatable label in the mapping table for corresponding block data collection；
MB 3 completes to carry out Data Integration according to map section cumulative data after screening；
MB 4, splitting size according to data is that the thin block of correspondence in each data set stamps the whole labels belonged in collection；
MB 5 adds data by browsing degree according to by label number.
Preferably, in the step B5, including following sub-step:
C1 establishes different degree matrix according to based tab navigation degree is planned as a whole；
C2, replaces data source or switching number source mode sensor obtains secondary data；
C3 recalculates step MB 1 to MB 4 with step c2 data, and obtained result is updated the matrix in step c1；
C4 repeats c3 until several source modules are empty or number source module is not with multiple types；
All matrix multiples obtained in step c3 are obtained observing matrix by c5；
C6 maps observing matrix and source data segment relationship.
Preferably, in the step M7, including following sub-step:
D1 establishes Fine Grain Data reinforcing degree calculated value U with Fine Grain Dataio:
Wherein UiFine Grain Data access degree when for the time being t, Ui-1The Fine Grain Data access degree of last moment, Δ t is between the time
Every T is that block accumulates access degree, and a is to be manually set to obscure value；
D2, according to Fine Grain Data reinforcing degree, total value come statistics accumulation reinforcing degree whether be more than setting value if it exceeds setting value then
Jump procedure d3, otherwise jump procedure d4
D3 continues to divide Granule Computing screening；
D4 is stepped up and obscures value, until reinforcing degree total value is more than setting value and records maximum and obscures value amax；
D5 completes this sifting sort, and amaxGive over to the basic value of screening next time.
Substantial effect of the invention is to be split as existing a large amount of remotely-sensed datas using method of the invention
Categorizing information is obtained based on grain classification method, and is obtained by Granule Computing after finding useful remote sensing information all remaining
Valuable information, and the processing of follow-up data is facilitated with into other classification of homogeneous data according to algorithm statistics.
Specific embodiment
Below by specific embodiment, technical scheme of the present invention will be further explained in detail.
Embodiment 1
A kind of shown, described Remote Sensing Data Processing method based on more granularities, comprising the following steps:
M1 together by time-sequencing unified integration by remotely-sensed data, and adds label according to data pick-up source；
M2 inputs Expert Rules data as granulation standard, carries out a Granule Computing sifting sort；
M3 is that each data block adds the specific label screened every time by screening；
M4 counts label included by coarse grain, and from the classification data of step M3, divides coarse grain number by number of tags according to data
According to；M5, to the data in step M4 using existing clear remotely-sensed data as Rules Filtering；
M6 is screened to obtain secondary Granule Computing sifting sort data from step M5, is turned the screening rule in step M5, do not change number
It carries out repeating screening in the case where；
M7 repeats step M5, M6 several times, the coarse grain data after being refined, and according to the secondary grain in final step M6
Calculating sifting classification data, determines Fine Grain Data；
M8, the convenient data screening next time of screening rule according to Fine Grain Data feature, in amendment step M6.
Wherein the Fine Grain Data Feature Selection mode of step M8 is Effective selection reference mode, can be similar remote sensing number
There is provided a screening easy way according to screening, can mode simultaneously and concurrently carry out the calculating of step M5 accelerating step M6
Journey, while step M6 is calculated as to the granulated key of data, and data constraint can be accelerated in other way,
Only it can also be accumulated by folk prescription formula data and obtain Effective selection mode.
In the step M2, including following sub-step: MA1 sets P Expert Rules, and each Expert Rules all wrap
Target property U is included, with property feature V；
Target property U is split as r gradient and is collected (U1, U2, U3, U4 ... Ur), similarly obtains property feature V's by MA2
Collect (V1, V2, V3, V4 ... Vc)
MA3 obtains set N=∑ijnijWherein N is that the total characteristic of single Expert Rules and total characteristic type close i and belong to r, and j belongs to
In c；
MA4 is calculatedWherein ai characteristic Ui is the sum of the concentration of V in feature, and bj is characterized
Sum of the Vj in the concentration that feature is U
MA5 calculates target property function according to transfer operator R obtained in step MA4
MA6, according to transfer operator R estimated performance characteristic function obtained in step MA4
MA7 classifies to data according to the function of step MA5 and step MA6.
In the step M3, including following sub-step:
MB1 resettles data mapping tables according to the result in step M2；
MB 2 adds repeatable label in the mapping table for corresponding block data collection；
MB 3 completes to carry out Data Integration according to map section cumulative data after screening；
MB 4, splitting size according to data is that the thin block of correspondence in each data set stamps the whole labels belonged in collection；
MB 5 adds data by browsing degree according to by label number.
Preferably, in the step B5, including following sub-step:
C1 establishes different degree matrix according to based tab navigation degree is planned as a whole；
C2, replaces data source or switching number source mode sensor obtains secondary data；
C3 recalculates step MB 1 to MB 4 with step c2 data, and obtained result is updated the matrix in step c1；
C4 repeats c3 until several source modules are empty or number source module is not with multiple types；
All matrix multiples obtained in step c3 are obtained observing matrix by c5；
C6 maps observing matrix and source data segment relationship.
In the step M7, including following sub-step:
D1 establishes Fine Grain Data reinforcing degree calculated value U with Fine Grain Dataio:
Wherein UiFine Grain Data access degree when for the time being t, Ui-1The Fine Grain Data access degree of last moment, Δ t is between the time
Every T is that block accumulates access degree, and a is to be manually set to obscure value；
D2, according to Fine Grain Data reinforcing degree, total value come statistics accumulation reinforcing degree whether be more than setting value if it exceeds setting value then
Jump procedure d3, otherwise jump procedure d4
D3 continues to divide Granule Computing screening；
D4 is stepped up and obscures value, until reinforcing degree total value is more than setting value and records maximum and obscures value amax；
D5 completes this sifting sort, and amaxGive over to the basic value of screening next time.
Claims (5)
1. a kind of Remote Sensing Data Processing method based on more granularities, which comprises the following steps:
M1 together by time-sequencing unified integration by remotely-sensed data, and adds label according to data pick-up source；
M2 inputs Expert Rules data as granulation standard, carries out a Granule Computing sifting sort；
M3 is that each data block adds the specific label screened every time by screening；
M4 counts label included by coarse grain, and from the classification data of step M3, divides coarse grain number by number of tags according to data
According to；
M5, to the data in step M4 using existing clear remotely-sensed data as Rules Filtering；
M6 is screened to obtain secondary Granule Computing sifting sort data from step M5, is turned the screening rule in step M5, do not change number
It carries out repeating screening in the case where；
M7 repeats step M5, M6 several times, the coarse grain data after being refined, and according to the secondary grain in final step M6
Calculating sifting classification data, determines Fine Grain Data；
M8, the convenient data screening next time of screening rule according to Fine Grain Data feature, in amendment step M6.
2. a kind of Remote Sensing Data Processing method based on more granularities according to claim 1, which is characterized in that the step
In rapid M2, including following sub-step:
MA1 sets P Expert Rules, and each Expert Rules all include target property U, with property feature V；
Target property U is split as r gradient and is collected (U1, U2, U3, U4 ... Ur), similarly obtains property feature V's by MA2
Collect (V1, V2, V3, V4 ... Vc)
MA3 obtains set N=∑ijnijWherein N is that the total characteristic of single Expert Rules and total characteristic type close i and belong to r, and j belongs to
c；
MA4 is calculatedWherein ai characteristic Ui is the sum of the concentration of V in feature, and bj is characterized Vj
In the sum for the concentration that feature is U
MA5 calculates target property function according to transfer operator R obtained in step MA4
MA6, according to transfer operator R estimated performance characteristic function obtained in step MA4
MA7 classifies to data according to the function of step MA5 and step MA6.
3. a kind of Remote Sensing Data Processing method based on more granularities according to claim 1, which is characterized in that the step
In rapid M3, including following sub-step:
MB1 resettles data mapping tables according to the result in step M2；
MB 2 adds repeatable label in the mapping table for corresponding block data collection；
MB 3 completes to carry out Data Integration according to map section cumulative data after screening；
MB 4, splitting size according to data is that the thin block of correspondence in each data set stamps the whole labels belonged in collection；
MB 5 adds data by browsing degree according to by label number.
4. a kind of Remote Sensing Data Processing method based on more granularities according to claim 3, which is characterized in that the step
In rapid MB 5, including following sub-step:
C1 establishes different degree matrix according to based tab navigation degree is planned as a whole；
C2, replaces data source or switching number source mode sensor obtains secondary data；
C3 recalculates step MB 1 to MB 4 with step c2 data, and obtained result is updated the matrix in step c1；
C4 repeats c3 until several source modules are empty or number source module is not with multiple types；
All matrix multiples obtained in step c3 are obtained observing matrix by c5；
C6 maps observing matrix and source data segment relationship.
5. a kind of Remote Sensing Data Processing method based on more granularities according to claim 1, which is characterized in that the step
In rapid M7, including following sub-step:
D1 establishes Fine Grain Data reinforcing degree calculated value U with Fine Grain Dataio:
Wherein UiFine Grain Data access degree when for the time being t, Ui-1The Fine Grain Data access degree of last moment, Δ t is between the time
Every T is that block accumulates access degree, and a is to be manually set to obscure value；
D2, according to Fine Grain Data reinforcing degree, total value come statistics accumulation reinforcing degree whether be more than setting value if it exceeds setting value then
Jump procedure d3, otherwise jump procedure d4
D3 continues to divide Granule Computing screening；
D4 is stepped up and obscures value, until reinforcing degree total value is more than setting value and records maximum and obscures value amax；
D5 completes this sifting sort, and amaxGive over to the basic value of screening next time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811400783.7A CN109543612B (en) | 2018-11-22 | 2018-11-22 | Multi-granularity-based remote sensing data processing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811400783.7A CN109543612B (en) | 2018-11-22 | 2018-11-22 | Multi-granularity-based remote sensing data processing method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109543612A true CN109543612A (en) | 2019-03-29 |
| CN109543612B CN109543612B (en) | 2023-04-21 |
Family
ID=65850231
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811400783.7A Active CN109543612B (en) | 2018-11-22 | 2018-11-22 | Multi-granularity-based remote sensing data processing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109543612B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104217254A (en) * | 2014-08-29 | 2014-12-17 | 中国水产科学研究院东海水产研究所 | Construction method of quick forecasting operation system of fishery fishing condition |
| CN106295613A (en) * | 2016-08-23 | 2017-01-04 | 哈尔滨理工大学 | A kind of unmanned plane target localization method and system |
| CN108614992A (en) * | 2018-03-19 | 2018-10-02 | 中国地质大学（武汉） | A kind of sorting technique of high-spectrum remote sensing, equipment and storage device |
-
2018
- 2018-11-22 CN CN201811400783.7A patent/CN109543612B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104217254A (en) * | 2014-08-29 | 2014-12-17 | 中国水产科学研究院东海水产研究所 | Construction method of quick forecasting operation system of fishery fishing condition |
| CN106295613A (en) * | 2016-08-23 | 2017-01-04 | 哈尔滨理工大学 | A kind of unmanned plane target localization method and system |
| CN108614992A (en) * | 2018-03-19 | 2018-10-02 | 中国地质大学（武汉） | A kind of sorting technique of high-spectrum remote sensing, equipment and storage device |
Non-Patent Citations (5)
| Title |
|---|
| S. K. MEHER ET AL.: "Ensemble of Adaptive Rule-Based Granular Neural Network Classifiers for Multispectral Remote Sensing Images", 《IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING》 * |
| 吴伟志: "多粒度粗糙集数据分析研究的回顾与展望", 《西北大学学报(自然科学版)》 * |
| 杨玉芳: "空间遥感数据的多粒度标记分类", 《中国优秀硕士学位论文全文数据库（信息科技辑）》 * |
| 苗夺谦等: "从人类智能到机器实现模型———粒计算理论与方法", 《智能系统学报》 * |
| 陈杰等: "结合支持向量机与粒度计算的高分辨率遥感影像面向对象分类", 《测绘学报》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109543612B (en) | 2023-04-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108388927B (en) | Small sample polarization SAR terrain classification method based on deep convolution twin network | |
| Spiesman et al. | Assessing the potential for deep learning and computer vision to identify bumble bee species from images | |
| Guevara et al. | diverse: an R Package to Analyze Diversity in Complex Systems. | |
| Su et al. | Rock classification in petrographic thin section images based on concatenated convolutional neural networks | |
| CN109218223B (en) | Robust network traffic classification method and system based on active learning | |
| Weaver et al. | LeafMachine: Using machine learning to automate leaf trait extraction from digitized herbarium specimens | |
| CN106055573B (en) | Shoe print image retrieval method and system under multi-instance learning framework | |
| CN111401426B (en) | Small sample hyperspectral image classification method based on pseudo label learning | |
| CN106528874A (en) | Spark memory computing big data platform-based CLR multi-label data classification method | |
| CN108985360A (en) | Hyperspectral classification method based on expanding morphology and Active Learning | |
| CN102324038A (en) | A kind of floristics recognition methods based on digital picture | |
| Deng et al. | Citrus disease recognition based on weighted scalable vocabulary tree | |
| Yu et al. | Automated diatom searching in the digital scanning electron microscopy images of drowning cases using the deep neural networks | |
| CN106203508A (en) | A kind of image classification method based on Hadoop platform | |
| De Lutio et al. | Digital taxonomist: Identifying plant species in community scientists’ photographs | |
| Nezamabadi-pour et al. | Concept learning by fuzzy k-NN classification and relevance feedback for efficient image retrieval | |
| CN111898418A (en) | Human body abnormal behavior detection method based on T-TINY-YOLO network | |
| Pastore et al. | An anomaly detection approach for plankton species discovery | |
| CN106570514A (en) | Automobile wheel hub classification method based on word bag model and support vector machine | |
| CN117633627A (en) | Deep learning unknown network traffic classification method and system based on evidence uncertainty evaluation | |
| CN106775694A (en) | A kind of hierarchy classification method of software merit rating code product | |
| CN109543612A (en) | A kind of Remote Sensing Data Processing method based on more granularities | |
| CN115063692B (en) | Remote sensing image scene classification method based on active learning | |
| CN113505863B (en) | Image multistage classification method and system based on cascade mean vector comprehensive scoring | |
| CN115757769A (en) | Artificial intelligence method and system applying big data analysis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |