CN109543729A - Time series data land cover classification method based on characteristic parameter cluster - Google Patents

Time series data land cover classification method based on characteristic parameter cluster Download PDF

Info

Publication number
CN109543729A
CN109543729A CN201811322302.5A CN201811322302A CN109543729A CN 109543729 A CN109543729 A CN 109543729A CN 201811322302 A CN201811322302 A CN 201811322302A CN 109543729 A CN109543729 A CN 109543729A
Authority
CN
China
Prior art keywords
time series
characteristic parameter
series data
cluster
windy
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.)
Pending
Application number
CN201811322302.5A
Other languages
Chinese (zh)
Inventor
董超
赵庚星
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.)
Shandong Agricultural University
Original Assignee
Shandong Agricultural University
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 Shandong Agricultural University filed Critical Shandong Agricultural University
Priority to CN201811322302.5A priority Critical patent/CN109543729A/en
Publication of CN109543729A publication Critical patent/CN109543729A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
    • G06F18/24133—Distances to prototypes
    • G06F18/24137—Distances to cluster centroïds
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Astronomy & Astrophysics (AREA)
  • Remote Sensing (AREA)
  • Multimedia (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The purpose of the present invention is to provide the time series data land cover classification methods clustered based on characteristic parameter, read the MODIS timing image of windy and sandy soil first；Read cluster initial value；Research area's windy and sandy soil situation is classified, the value for extracting these characteristic points is drawn indicatrix and inputted as the initial value of subsequent classification；Calculate characteristic parameter；It calculates subordinated-degree matrix and calculates cluster centre；Distance is less than precision, carries out in next step, divides time series data classification according to subordinated-degree matrix, finally exports result.The beneficial effects of the invention are as follows algorithms preferably to provide important parameter for the acquisition of windy and sandy soil information, for the monitoring of large scale windy and sandy soil and the classification foundation of Monitoring of Resource and Environment.

Description

Time series data land cover classification method based on characteristic parameter cluster
Technical field
The invention belongs to remote sensing technology field, be related to it is a kind of based on characteristic parameter cluster to MODIS time series data Carry out land cover classification method.
Background technique
Land use change survey is always in constantly driving land cover pattern variation, and climate change is studied in these variations, ecological environment The work such as monitoring have great significance.Windy and sandy soil information is the premise of Changeement, but due to nature and social factor It influences, is constantly among dynamic change, it is difficult to quickly, timely obtain latest development.Remote sensing satellite data have macroscopic view, Dynamically, fast and accurately advantage has irreplaceable status in the detection of a wide range of surface cover.For monitoring on a large scale, High spatial resolution satellite generally has the lower return visit period, influences vulnerable to weather conditions, can not obtain effective image in time Data, it is difficult to effective monitoring in real time is carried out to big region, and middle low resolution image has greater advantage in these areas.It grinds Study carefully and shows more accurately portray windy and sandy soil type using high time resolution image.Vegetation index enhances vegetation letter Breath, weakens non-vegetation information, provides an important basis for Land Information cover.It is imaged using MODIS intermediate-resolution Spectrometer (moderate-resolution imaging spectroradiometer), the high time resolution timing of production Data carry out timing curve signature analysis in conjunction with land use data, extract timing curve characteristic parameter, based on fuzzy poly- Class algorithm fast implements the automatic classification of windy and sandy soil in regional scope, and is analyzed its classification results and precision, with Phase provides technical method for the quick and precisely monitoring that Land in Regional Land is covered and supports.
Summary of the invention
The purpose of the present invention is to provide the time series data land cover classification methods clustered based on characteristic parameter, originally Advantageous effect of the invention is that algorithm preferably can provide important parameter for the acquisition of windy and sandy soil information, is covered for large scale soil Monitored and Monitoring of Resource and Environment classification foundation.
The technical scheme adopted by the invention is that following the steps below:
Step 1: reading MODIS time sequence image data collection；
Step 2: reading cluster initial value；
Research area's windy and sandy soil situation is classified, the value for extracting these characteristic points draws indicatrix as subsequent classification Initial value input；
Merge ground class be five classes: one season crop, two season crop, Forest and sod, construction land, waters.
Wherein, classification water irrigate the fields as one season crop；Nonirrigated farmland as two season crop；Orchard, forest land are as Forest and sod；It is waste Meadow, city, organic town are as construction land；The reservoir water surface, the river water surface are as waters.
Step 3: calculating characteristic parameter；
Lance distance is selected to define the distance of two curves, it, which overcomes Euclidean distance, the shortcomings that dimension, to big Singular value is insensitive, is defined as:
Wherein m is time series Xi、XjLength, the instant number of phases.Lance distance, which can overcome, the shortcomings that dimension, but not Similar curves can be distinguished well, therefore the statistical nature of introducing curve and related coefficient are corrected.Basic statistics feature It is to describe some statistics of time series global structure including extreme value, mean value and variance, it is not the distinctive spy of time series Sign, but can be used for describing the feature of any group of data.Related coefficient can be very good reflection curve correlation and correlation Direction.Using the statistical nature of timing curve, lance distance and related coefficient common weight composition characteristic parameter, characteristic parameter is used Distance defines the similarity degrees of two curves.
d(Xi, Xj)=ω1Sij+ω2Lij+ω3RijI, j=1,2 ..., n (2)
Step 4: calculating subordinated-degree matrix and calculate cluster centre；
Fuzzyc-means (FCM) be it is a kind of permission data belong to two classes or more than clustering method.The present invention uses should Characteristic information in clustering time series data, based on the objective function minimized below:
Wherein m is greater than 1 any real number, uijIt is xiDegree of membership in j class, xiIt is the measured value of i-th of data, cj It is the cluster centre of j class.| | * | | it is the similarity benchmark between any one measured value and center.Fuzzy is by changing above In generation, arrives optimal objective function, during which constantly updates degree of membership uijWith cluster centre cj。
Step 6: distance is less than precision, then carries out in next step, otherwise jumping to step 3；
Step 7: dividing time series data classification according to subordinated-degree matrix；
Step 8: output result.
Detailed description of the invention
Fig. 1 status is the timing curve of water body.
Specific embodiment
The present invention is described in detail With reference to embodiment.
It studies area and selects Taian Shandong city, be located at middle Shandong Province, between 116 ° 20 ' -117 ° 59 ' of east longitude, north latitude 35-degree Between 38 ' -36 ° 28 ', entire topography is southwester tilted from northeast, geomorphic type multiplicity.Belong to the continental monsoon semi-moist in temperate zone Climatic province makes a clear distinction between the four seasons, and light temperature is synchronous, rain heat same season, 12.9 DEG C of annual gas low temperature, 697 millimeters of average annual precipitation.MAIN SOILS Type is brown earth, cinnamon soil, Shajiang black soil and moisture soil, and overall soil fertility is higher.By 2016, Tai'an territory total area 776141.48 hectares, wherein arable land 364721.23 hectares, 40559.69 hectares of field, 111280.47 hectares of forest land, construction are used 125649.68 hectares of ground, 62695.08 hectares of unused land.
Select vegetation product MODIS13Q1 (the MODIS/Terra Vegetation Indices 16- of MODIS/Terra Day L3Global 250m SIN Grid) data.The time range of data is 16 days to 2016 October in 2015 of September 29 Day, every 16 days time points, the data file at totally 23 time points.The spatial resolution of data is 250m.Research on utilization area Present landuse maps in 2009 and online high-resolution satellite image are as nicety of grading reference foundation.With present status of land utilization Figure is reference, in conjunction with day map high definition image, generates test samples, amounts to 2065 pixels.Using accuracy test sample to point Class result carries out accuracy test, the results are shown in Table 1.
1 classification results confusion matrix of table
Classification results Kappa coefficient is 0.8432.Wherein, nicety of grading of all categories is respectively construction land 82.71%, and one Season crop 88.87%, Forest and sod 93.95%, two season crop 87.59%, water body 58.82%.Wherein water body and construction land are missed Score highest, this is also consistent with indicatrix variation in cluster result.
It is water body lake on status Fig. 1, chooses middle of a lake unit and successively draw curve outward, it can be seen that EVI value exists always Rise, since curvilinear characteristic itself is similar, is just accidentally divided into construction land being raised to the L2 stage.
The present invention is based on the EVI product data of MODIS 250m resolution ratio to have made 2015-2016 time series data collection, chooses Statistical nature, lance distance and the related coefficient common weight composition characteristic parameter of timing curve, with the distance of characteristic parameter come Define the similarity degree of two curves.Data category is divided by the method for fuzzy clustering, is extracted research area's windy and sandy soil feelings Condition, and classification results precision is analyzed.Main Conclusions is as follows:
(1) timing curve is smoothed using Hants method, different vegetative coverage timing curve features can be made more Add obvious, and smoothing processing enchancement factor influences the exceptional value generated, provides the foundation for further sort research.
(2) by analyzing characteristic parameter, overture when statistical nature, lance distance and related coefficient respectively define is obtained The feature of line.Statistical nature reflects that the basic numerical attribute of curve, lance distance reflect the feature of timing curve, and related coefficient is anti- The similarity degree of curvilinear motion is reflected, characteristic parameter complements each other, and can obtain preferable classification knot by adjusting characteristic parameter Fruit.
(3) by the cluster to curvilinear characteristic parameter, pass through the comparative analysis to cluster front and back ground class curve, display cluster Algorithm can further mining data built-in attribute, the cluster curve closer to reality is provided.It can be avoided because threshold value is selected It selects.
(4) pass through fuzzy clustering operation, windy and sandy soil be divided into five classes: construction land, Forest and sod, one season Crops Land, Two season Crops Land and water body.As a result it is 0.8432 that overall classification accuracy, which reaches 89.37%, kappa coefficient,.
The above is only not to make limit in any form to the present invention to better embodiment of the invention System, any simple modification that embodiment of above is made according to the technical essence of the invention, equivalent variations and modification, Belong in the range of technical solution of the present invention.

Claims (4)

1. based on characteristic parameter cluster time series data land cover classification method, it is characterised in that according to the following steps into Row:
Step 1: reading the MODIS timing image of windy and sandy soil；
Step 2: reading cluster initial value；
Research area's windy and sandy soil situation is classified, the value for extracting these characteristic points draws indicatrix as the initial of subsequent classification Value input；
Step 3: calculating characteristic parameter；
Step 4: calculating subordinated-degree matrix and calculate cluster centre；
Step 5: distance is less than precision, then carries out in next step, otherwise jumping to step 3；
Step 6: dividing time series data classification according to subordinated-degree matrix；
Step 7: output result.
2. according to the time series data land cover classification method based on characteristic parameter cluster described in claim 1, feature Be: in the step 2 class be five classes: one season crop, two season crop, Forest and sod, construction land, waters；
Wherein, classification water irrigate the fields as one season crop；Nonirrigated farmland as two season crop；Orchard, forest land are as Forest and sod；Weeds Ground, city, organic town are as construction land；The reservoir water surface, the river water surface are as waters.
3. according to the time series data land cover classification method based on characteristic parameter cluster described in claim 1, feature It is: calculation of characteristic parameters method in the step 3: selects lance distance to define the distance of two curves；
Wherein m is time series Xi、XjLength, added jointly using the statistical nature, lance distance and related coefficient of timing curve Composition characteristic parameter is weighed, the similarity degree of two curves is defined with characteristic parameter distance:
d(Xi, Xj)=ω1Sij+ω2Lij+ω3RijI, j=1,2 ..., n. (2)
4. according to the time series data land cover classification method based on characteristic parameter cluster described in claim 1, feature It is: calculates subordinated-degree matrix in the step 4 and calculate cluster centre；
Using the characteristic information in Fuzzyc-means clustering method cluster time series data, based on the target letter minimized below Number:
Wherein m is greater than 1 any real number, uijIt is xiDegree of membership in j class, xiIt is the measured value of i-th of data, cjIt is j class Cluster centre, | | * | | be the similarity benchmark between any one measured value and center；Fuzzy is by iterating to above most During which excellent objective function constantly updates degree of membership uijWith cluster centre cj；
CN201811322302.5A 2018-11-08 2018-11-08 Time series data land cover classification method based on characteristic parameter cluster Pending CN109543729A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811322302.5A CN109543729A (en) 2018-11-08 2018-11-08 Time series data land cover classification method based on characteristic parameter cluster

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811322302.5A CN109543729A (en) 2018-11-08 2018-11-08 Time series data land cover classification method based on characteristic parameter cluster

Publications (1)

Publication Number Publication Date
CN109543729A true CN109543729A (en) 2019-03-29

Family

ID=65844787

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811322302.5A Pending CN109543729A (en) 2018-11-08 2018-11-08 Time series data land cover classification method based on characteristic parameter cluster

Country Status (1)

Country Link
CN (1) CN109543729A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110516552A (en) * 2019-07-29 2019-11-29 南京航空航天大学 A kind of multipolarization radar image classification method and system based on timing curve
CN111984702A (en) * 2020-08-17 2020-11-24 北京大学深圳研究生院 Method, device, equipment and storage medium for analyzing spatial evolution of village and town settlement
CN113989668A (en) * 2021-10-27 2022-01-28 辽宁工程技术大学 Remote sensing crop automatic classification method based on time series characteristics
CN113989668B (en) * 2021-10-27 2024-10-15 辽宁工程技术大学 Remote sensing crop automatic classification method based on time sequence characteristics

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104318270A (en) * 2014-11-21 2015-01-28 东北林业大学 Land cover classification method based on MODIS time series data
CN105005784A (en) * 2015-05-21 2015-10-28 中国科学院遥感与数字地球研究所 Time sequence remote sensing image land cover classification method based on CD-DTW distance
CN107273820A (en) * 2017-05-26 2017-10-20 中国科学院遥感与数字地球研究所 A kind of Land Cover Classification method and system
CN107345860A (en) * 2017-07-11 2017-11-14 南京康尼机电股份有限公司 Rail vehicle door sub-health state recognition methods based on Time Series Data Mining
CN108615054A (en) * 2018-04-18 2018-10-02 清华大学 The overall target construction method that similitude is weighed between drainage pipeline networks node

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104318270A (en) * 2014-11-21 2015-01-28 东北林业大学 Land cover classification method based on MODIS time series data
CN105005784A (en) * 2015-05-21 2015-10-28 中国科学院遥感与数字地球研究所 Time sequence remote sensing image land cover classification method based on CD-DTW distance
CN107273820A (en) * 2017-05-26 2017-10-20 中国科学院遥感与数字地球研究所 A kind of Land Cover Classification method and system
CN107345860A (en) * 2017-07-11 2017-11-14 南京康尼机电股份有限公司 Rail vehicle door sub-health state recognition methods based on Time Series Data Mining
CN108615054A (en) * 2018-04-18 2018-10-02 清华大学 The overall target construction method that similitude is weighed between drainage pipeline networks node

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
李深洛: "基于特征的时间序列聚类", 《中国优秀硕士学位论文全文数据库信息科技辑》 *
王伶俐 等: "植被指数时序数据距离测度方法评价", 《遥感学报》 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110516552A (en) * 2019-07-29 2019-11-29 南京航空航天大学 A kind of multipolarization radar image classification method and system based on timing curve
CN110516552B (en) * 2019-07-29 2021-02-05 南京航空航天大学 Multi-polarization radar image classification method and system based on time sequence curve
CN111984702A (en) * 2020-08-17 2020-11-24 北京大学深圳研究生院 Method, device, equipment and storage medium for analyzing spatial evolution of village and town settlement
CN113989668A (en) * 2021-10-27 2022-01-28 辽宁工程技术大学 Remote sensing crop automatic classification method based on time series characteristics
CN113989668B (en) * 2021-10-27 2024-10-15 辽宁工程技术大学 Remote sensing crop automatic classification method based on time sequence characteristics

Similar Documents

Publication Publication Date Title
Zhang et al. An automated early-season method to map winter wheat using time-series Sentinel-2 data: A case study of Shandong, China
CN113128401B (en) Regional actual irrigation area monitoring method based on optical and radar remote sensing data
Herbei et al. Processing and Use of Satellite Images in Order to Extract Useful Information in Precision Agriculture.
Duan et al. Mapping the soil types combining multi-temporal remote sensing data with texture features
Kang et al. Support vector machine classification of crop lands using sentinel-2 imagery
CN117611993A (en) Method for estimating vegetation classification based on remote sensing actual evapotranspiration
CN109543729A (en) Time series data land cover classification method based on characteristic parameter cluster
Dong et al. Fine mapping of key soil nutrient content using high resolution remote sensing image to support precision agriculture in Northwest China
Bao et al. A fine digital soil mapping by integrating remote sensing-based process model and deep learning method in Northeast China
Guizani et al. Enhancing water balance assessment in urban areas through high-resolution land cover mapping: Case study of Debrecen, Hungary
Fan et al. Large-scale Rice mapping based on Google earth engine and multi-source remote sensing images
CN116051993A (en) Artificial grassland identification method
You et al. Crop Mapping of Complex Agricultural Landscapes Based on Discriminant Space
Compaoré The impact of savannah vegetation on the spatial and temporal variation of the actual evapotranspiration in the Volta Basin, Navrongo, Upper East Ghana
Chen et al. Water requirement for irrigation of complicated agricultural land by using classified airborne digital sensor images
Yourek et al. Development and application of the soil moisture routing (SMR) model to identify subfield-scale hydrologic classes in dryland cropping systems using the Budyko framework
Yedage et al. Remote sensing and GIS base crop acreage estimation of the sugarcane for Solapur district, Maharashtra
Zhang et al. Precise classification of forest species based on multi-source remote-sensing images.
CN110135328A (en) Pakistani land cover pattern information extracting method based on multi-source Spatial Data
Yang et al. Integrating multidimensional feature indices and phenological windows for mapping cropping patterns in complex agricultural landscape regions
Paiboonvorachat Using remote sensing and GIS techniques to assess land use/land cover changes in the Nan Watershed, Thailand
Zhang et al. Detection of paddy rice with time series Sentinel 1/2 images and transfer learning algorithms
Wei et al. A novel two-step framework for mapping fraction of mulched film based on very high resolution satellite observation and deep learning
Zhang et al. Mapping wetlands in Northeast China by using knowledge-based algorithms and microwave (PALSAR-2, Sentinel-1), optical (Sentinel-2, Landsat), and thermal (MODIS) images
Chen et al. Extraction Methods for Small-Scale Features on a Large Scale: Investigating Object-Oriented Cart Decision Tree for Gravel Information Extraction

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
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20190329