CN104615977B - The winter wheat remote sensing recognition method of comprehensive crucial Aspection character and fuzzy classification technology - Google Patents
The winter wheat remote sensing recognition method of comprehensive crucial Aspection character and fuzzy classification technology Download PDFInfo
- Publication number
- CN104615977B CN104615977B CN201510037425.4A CN201510037425A CN104615977B CN 104615977 B CN104615977 B CN 104615977B CN 201510037425 A CN201510037425 A CN 201510037425A CN 104615977 B CN104615977 B CN 104615977B
- Authority
- CN
- China
- Prior art keywords
- winter wheat
- remote sensing
- resolution
- modis
- pixel
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 241000209140 Triticum Species 0.000 title claims abstract description 121
- 235000021307 Triticum Nutrition 0.000 title claims abstract description 121
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000005516 engineering process Methods 0.000 title claims abstract description 28
- 238000011160 research Methods 0.000 claims abstract description 36
- 238000002360 preparation method Methods 0.000 claims abstract description 4
- 230000001932 seasonal effect Effects 0.000 claims description 13
- 239000013598 vector Substances 0.000 claims description 13
- 238000001228 spectrum Methods 0.000 claims description 11
- 238000004364 calculation method Methods 0.000 claims description 10
- 238000012549 training Methods 0.000 claims description 6
- 238000000611 regression analysis Methods 0.000 claims description 4
- 238000012952 Resampling Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000012937 correction Methods 0.000 claims description 2
- 238000012544 monitoring process Methods 0.000 abstract description 14
- 230000033764 rhythmic process Effects 0.000 abstract description 7
- 230000003595 spectral effect Effects 0.000 abstract description 5
- 238000011156 evaluation Methods 0.000 abstract description 4
- 238000012795 verification Methods 0.000 description 6
- 238000011835 investigation Methods 0.000 description 4
- 238000005070 sampling Methods 0.000 description 4
- 230000007547 defect Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000009605 growth rhythm Effects 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
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
- G06V20/188—Vegetation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (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)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Image Processing (AREA)
Abstract
The invention belongs to remote sensing monitoring technical field, and in particular to a kind of winter wheat remote sensing recognition method for comprehensively utilizing crucial Aspection character and fuzzy classification technology.This method includes：Data prediction, the abundance figure under preparation research area coarse resolution, pixel is obtained under the middle high-resolution yardstick in research area for steps such as the degree of membership of winter wheat, Comprehensive Evaluations.The present invention combines the method based on the aspect rhythm and pace of moving things（Utilize the jump of low resolution remote sensing）And fuzzy classification technology（Utilize the spectral information of middle high-resolution remote sensing）Obtain the middle high-resolution recognition result with definite spatial distribution, the shortcomings that compensate for two methods each, both solved fuzzy classification technology pixel belong to all kinds of probability it is suitable when uncertain problem, the abundance figure for solving the problems, such as to obtain using Aspection character again can not show the definite spatial distribution of crop, and new monitoring, evaluation measures are provided for the remote sensing monitoring of winter wheat.
Description
Technical Field
The invention belongs to the technical field of remote sensing monitoring, and particularly relates to a winter wheat remote sensing identification method comprehensively utilizing key seasonal features and a fuzzy classification technology.
Background
In the prior art, methods for identifying crop types based on remote sensing technology mainly include methods based on spectral information and methods based on quaternary rhythms.
The method based on spectral information identification is mainly used for medium-high resolution remote sensing images, the principle is that the statistical characteristics of pixel values are utilized for classification and identification, but due to the existence of the phenomena of 'same object and different spectrum' and 'foreign object and same spectrum', the identification result is deviated.
The identification method based on the quaternary phase rhythm characteristics is mainly used for forming a low-spatial resolution remote sensing image of a time sequence, and the principle is to utilize the difference of growth rhythms of crops and other vegetation, but because the defect of low image resolution and low identification result precision exists, the identification method of sub-pixels, such as a linear spectrum separation technology, an area index method based on a key phenological period and the like, is commonly utilized. However, the results obtained by the methods are an abundance map, namely the proportion of the area of the crops in the image elements with coarse resolution, and the specific distribution of the crops cannot be determined exactly, so that great inconvenience still exists in specific use.
Because of the advantages and disadvantages of each of the two types of methods, a spectral information-based method is also often used in combination with a quaternary rhythm-based method. However, the combination of the two methods is still limited to the pure elements decomposed by the mixed pixels provided by the medium and high resolution images, and then the mixed pixels are decomposed by the spectral characteristics of the low resolution data. Although the combination has certain advantages, the advantages of the two methods are not fully exerted, and the identification result is only the planting area of the crops and does not extract the real spatial distribution of the crops, so the practical use is still limited.
The fuzzy classification technology is applied to crop type identification, a useful theoretical basis is provided for the uncertain problem of pixel attribution in the remote sensing image, and the principle is as follows: when a pixel can belong to different classes at the same time, the key is to determine the membership degree of the pixel to each class, and finally, the pixel is determined as the class with the maximum membership degree. However, the problem with this approach is that when there are two or more of the highest classes of membership that are substantially equivalent, there is significant uncertainty in attributing the picture element to the class with the highest degree of membership.
In the prior art, monitoring the planting area of winter wheat by using a remote sensing image is a relatively quick and intuitive technical means, and reading of the remote sensing image has various technical means, but due to inherent advantages and disadvantages of various technical means, one or more technologies are often used in combination and are mutually corrected at the same time to obtain accurate monitoring data, and in the prior art, relevant reports of remote sensing identification of winter wheat by integrating key season features and fuzzy classification technologies are not seen.
Disclosure of Invention
The invention aims to provide a winter wheat remote sensing identification method comprehensively utilizing key seasonal phase characteristics and a fuzzy classification technology. The main technical idea of the invention is as follows: obtaining an abundance map under a coarse resolution scale of a research area by using key quaternary features; obtaining the membership degree of the pixels under high resolution scales in the research area to winter wheat by using a fuzzy classification technology; and combining the two results, correcting the phenomenon of foreign body co-spectrum in fuzzy classification by utilizing quaternary phase characteristic information, determining the specific position of the neutron pixel in the abundance map by utilizing the membership degree of the neutron pixel in the fuzzy classification, and complementing the advantages of the two methods to obtain the exact spatial distribution of winter wheat in the research area under the medium-high resolution scale.
The technical scheme adopted by the invention is specifically introduced as follows.
The winter wheat remote sensing identification method integrating key seasonal features and fuzzy classification technology comprises the following steps:
(1) data pre-processing
Completely registering data of the TM high-resolution image and the MODIS coarse-resolution image on a projection and coordinate system;
the method specifically comprises the following steps: the TM data is resampled to 25 meters and is completely registered with MODIS 250 meter data on a projection and coordinate system, so that each MODIS pixel can correspond to 10 x 10 TM pixels in spatial position;
(2) preparation of abundance maps at coarse resolution of the study area
A. According to the winter wheat planting condition in a research area, under the condition of ensuring the precision, a typical sample is laid in the field to investigate the winter wheat planting area proportion in one pixel of MODIS coarse resolution scale, and a sample is provided for regression analysis; the planting condition of winter wheat can be interpreted by utilizing a high-resolution remote sensing image, and pixel percentage data of MODIS coarse resolution scale is made as a sample on the basis of the interpretation, so that a field layout sample is replaced;
B. extracting a time period reflecting the typical phenological period of the winter wheat from the MODIS NDVI time sequence, and establishing a slope image of the typical phenological period by using MODIS NDVI data of the time period; the typical phenological period of the winter wheat is preferably from the green returning period of the winter wheat to the booting period of the winter wheat;
C. establishing a regression model by using the typical phenological period slope image obtained in the step B and the sample data obtained in the step A, and expanding the regression model to the whole research area so as to obtain an abundance map of the winter wheat planting area of the research area;
(3) obtaining the membership degree of the pixels to winter wheat under the medium and high resolution scales of the research area
A. Obtaining the membership degree of each pixel belonging to winter wheat by using a fuzzy classification technology and a membership degree determination method based on a Bayesian rule according to TM data; the specific calculation method is as follows:
hypothesis classification of vectors in imagesWhereinX TIs a vectorXThe research area is divided intomClass (A)k i ，i=1,2，…，m) According to the Bayes formula, inXConditions of occurrence ascribed tok i The class attribution probability is:
，
vector quantityXIn a categoryk i The conditional probability density function of (a) is:
，
in the formula,P(k i ) Is a categoryk i A priori probability of (a); p: (X/k i ) Is a categoryk i In-appearance vectorXThe probability of (d); n isThe dimension of the feature space;mthe number of research area categories;μ i is a categoryk i The mean vector of the training samples;is a categoryk i Covariance matrix of training samplesDeterminant of (4);
B. data correction is carried out, and the phenomenon of 'same object and different spectrum' is eliminated; for the phenomenon of 'same-object different-spectrum' such as spectrum difference of winter wheat caused by different conditions such as climate, growth vigor and the like, the problem can be solved by increasing investigation samples and respectively calculating membership degrees, so that all pixels belonging to the winter wheat can be extracted;
(4) and (4) comprehensive judgment, namely comprehensive step (2) abundance graph and step (3) membership degree data are used for judging the planting condition of winter wheat in the remote sensing graph
A. In each MODIS pixel, the membership degree of the TM pixel corresponding to the MODIS pixel to winter wheat is ranked from high to low;
B. if the abundance of the winter wheat of one MODIS pixel is F%, determining the first F% pixels which correspond to 10 multiplied by 10 pixels in spatial position and are sorted according to the membership degree of the winter wheat in the step A under the condition of medium and high resolution as the winter wheat, and determining the F value according to the abundance map in the step (2);
C. the method is extended to the whole research area, and the planting area and distribution of winter wheat can be obtained.
The winter wheat remote sensing identification method integrating key quaternary phase characteristics and the fuzzy classification technology provided by the invention integrates a quaternary phase rhythm-based method (utilizing the time advantage of low-resolution remote sensing) and a fuzzy classification technology (utilizing the spectrum information of medium-high resolution remote sensing), obtains a medium-high resolution identification result with exact spatial distribution, makes up the respective defects of the two methods, solves the uncertain problem of the fuzzy classification technology when the probability of pixels belonging to various types is relative, solves the problem that an abundance map obtained by utilizing the quaternary phase characteristics cannot show the exact spatial distribution of crops, provides a new monitoring and evaluation means for the remote sensing monitoring of winter wheat, simultaneously provides a new reference for the remote sensing monitoring of other crop types, and the used remote sensing image data with medium-low resolution scale is easy to obtain and is easy to carry out practical monitoring application on a certain regional scale, therefore, the method has good popularization and application values.
Drawings
FIG. 1 is a general technical framework diagram of the present invention;
FIG. 2 is a diagram of a field sample layout in Luoyang;
FIG. 3 is a NDVI time series curve of different vegetation types in 2009-2010 winter wheat growing season in Luoyang, from which the season features of different vegetation can be reflected;
FIG. 4 is a schematic diagram of winter wheat identification incorporating key seasonal features and fuzzy classification techniques; wherein, fig. 4 (a) is a spatial correspondence relationship between one pixel of a coarse resolution image and a pixel of a medium resolution, the whole represents an MODIS 250 m pixel, each small square represents one pixel of 25 m resolution, and the number represents the membership degree of each pixel of 25 m resolution to winter wheat obtained by using a fuzzy technology; fig. 4 (B) represents the recognition result when the MODIS pel abundance is F% =58%, the specific location of the winter wheat component recognized for this MODIS pel is highlighted in grey;
FIG. 5 shows the result of MODIS 250-meter scale winter wheat abundance calculation;
FIG. 6 is a result of calculation of membership to winter wheat for pixels of TM25 m scale;
fig. 7 is a recognition result of growing season winter wheat of a remote sensing map of the Luoyang area in 2009-2010 by using the method.
Detailed Description
The present invention is further illustrated by the following examples.
Before describing the embodiments, the concept of satellite image resolution in the present application is briefly explained as follows: for satellite image resolution, the meanings of low (coarse) resolution and high resolution are not completely consistent for different satellites, and for the purposes of the present application, an image with resolution of hundreds of meters, such as MODIS, is generally considered as low resolution; TM is a ten-meter image of medium resolution, and the meter-level image of high resolution. Therefore, the present application is based on this as a standard unless otherwise specified.
Examples
In the embodiment, the region of luoyang is taken as a research area, and the related remote sensing images in 2009-2010 of the region are interpreted and verified and analyzed by the invention, and the following brief description is provided.
The main technical idea of the invention is shown in figure 1: firstly, resampling is carried out according to a TM remote sensing image, for example, the TM resolution is set to be 25 meters, TM25 meter image data is obtained, then the data and an MODIS coarse resolution image are subjected to spatial registration, and meanwhile, membership degree calculation is carried out on TM25 meter image data. Secondly, obtaining various vegetation time sequence curves according to the MODIS time sequence, selecting a key time period of the winter wheat, and obtaining a slope image of the key time period of the winter wheat; meanwhile, field sample distribution investigation is carried out according to the MODIS image or sampling is carried out according to the remote sensing image after high-resolution interpretation, and the winter wheat area proportion under the MODIS scale is obtained; and (3) performing regression analysis by combining the slope image of the winter wheat in the key time period and the winter wheat area proportion data obtained by sample survey to obtain an identification model of the winter wheat, and further obtaining an abundance map of the winter wheat. Thirdly, calculating membership degree according to TM25 m image data, and further attributing probability images. And finally, judging by integrating the membership result of TM25 m image data and the abundance map result under the MODIS scale to obtain the recognition result of winter wheat, wherein the judgment standard is as follows: 10 multiplied by 10 TM25 m pixels corresponding to each MODIS pixel are arranged from large to small according to the membership degree, and the first 10 multiplied by F percent of pixels are judged as winter wheat according to the abundance value F percent of the MODIS pixel.
The present invention is specifically described below.
The method for identifying winter wheat by remote sensing by integrating key seasonal features and fuzzy classification technology comprises the following steps:
(1) remote sensing data preprocessing
The TM data is resampled to 25 meters and fully registered with MODIS 250 meters data in the projection and coordinate system, so that each MODIS pel can correspond to 10 × 10 TM pels in spatial location.
(2) Preparation of abundance maps at coarse resolution of the study area
A. According to the winter wheat planting condition in a research area, under the condition of ensuring the precision, a typical sample is laid in the field to investigate the winter wheat planting area proportion in one pixel of the MODIS scale, and a sample is provided for regression analysis; in order to reduce field investigation, the winter wheat planting condition can be interpreted by using a high-resolution remote sensing image, and pixel percentage data of MODIS scale is made as a sample on the basis of the interpreted winter wheat planting condition, so that a sample prescription is laid instead of field work.
Specifically, in this embodiment, the inventors performed field investigation on the winter wheat planting situation in the luoyang area, and the distribution of the specific samples is shown in fig. 2.
55 samples are arranged, wherein the actual size of each sample is 250 meters multiplied by 250 meters, and simultaneously, the samples with different planting proportions (the area proportion of winter wheat in the samples) are distributed in the selection and arrangement processes of the samples. In the actual layout process of the sample, about 20 sample squares can be generally laid in one county to meet the requirement.
B. Extracting a time interval reflecting the typical phenological period of the winter wheat from the MODIS NDVI time sequence, and establishing a slope image of the typical phenological period by using MODIS NDVI data of the time interval.
Specifically, an NDVI time series curve of pure pixels of different vegetation types is extracted based on MODIS NDVI time series data, as shown in FIG. 3.
Because the curves of winter wheat are obviously different from other vegetation types in NDVI time series curves of different vegetation types in the same time period in a research area, A, B, C, D, E, F six key characteristic points can be obviously distinguished. Thus, if an MODIS pixel element is closer to a winter wheat pure pixel element, the more similar the NDVI curve thereof is to a standard curve (NDVI time series curve of the winter wheat pure pixel element); however, as the number of other vegetation types mixed in the pixel elements increases, the slope of the curve between key feature points changes, which is also the theoretical basis for identifying winter wheat by using the seasonal rhythm features.
It is emphasized and explained that the recognition of winter wheat in the prior art is mostly achieved by combining the AB and EF segments, namely: if other types of ground objects are mixed in the pixel elements, the slope of the AB section and the EF section can be changed, and finally, a regression model is constructed by the pixel elements and the measured data to identify the winter wheat. However, in the actual monitoring process, there are often serious errors, and the main reason is that the NDVI of other surface features in the AB segment and the EF segment is very different, so that when different surface features of the same proportion are mixed in the pixel, the pixel value is also very different, so that the slope changes of the AB segment and the EF segment are different, that is, when the proportion of other surface features mixed in the pixel is fixed, the slope changes of the AB segment and the EF segment should be changed correspondingly theoretically, and then the crop can be identified well, but actually, the slope changes of the AB segment and the EF segment are caused by different types of the mixed surface features, so that the identification result has errors. In short, identifying winter wheat in segments AB and EF is subject to large errors.
The inventor believes that the time point C in FIG. 3 roughly corresponds to the green returning period of winter wheat, the time point D roughly corresponds to the booting period of winter wheat, and the time point is the starting point of the growth of other vegetation types, so that the CD section is the period of rapid growth of winter wheat, and the NDVI values of other ground features are relatively stable and have not very obvious difference, so that the more the proportion of other vegetation types mixed in the pixel is, the smaller the slope of the CD section is, and the influence of the difference of other vegetation types is also reduced to the minimum. Therefore, the inventor thinks that the problem of large error in the process of judging winter wheat by using the AB section and the EF section commonly used in the prior art can be better solved by using the CD section to construct the identification model.
Through calculation, the regression model established by using the slope of the CD section and the actually measured pixel percentage of the winter wheat is as follows:
；
wherein,FWWis the area percentage of the winter wheat pixels;NDVI D-NDVI Cis the slope of the CD segment (in general, the time period of a CD is a unit time); a and b are coefficients to be found.
Furthermore, image data representing the slope of the CD section is constructed by using MODIS NDVI data of the research area, regression calculation is carried out on the image data and field actual measurement data, coefficients a and b to be solved are solved, and the model is expanded to the whole research area, so that a winter wheat planting area abundance map of the MODIS scale of the research area can be solved. Generally, if the area of the research area is large, different models can be constructed by utilizing ecological subareas for identification.
Specifically, in this embodiment, the regression model established by using the MODIS NDVI image reflecting the CD segment slope and the 55 sample data is as follows:
；
further, regression verification analysis was performed, a determination coefficient (square of correlation coefficient) was calculated,the degree of fit of the linear regression (which is intended to illustrate the degree of interpretation of the dependent variable variation by the independent variable, R) is determined2Greater degree of fit), R2And =0.8, which shows a good fitting effect.
The regression model was extended to the entire study area to obtain the abundance map of the winter wheat planting area in the study area, as shown in fig. 5.
(3) Membership degree of pixels to winter wheat under medium and high resolution scales of research area
A. And classifying the images with the resolution of 25 meters according to the TM resampling by using a fuzzy classification technology to obtain the membership degree of each pixel belonging to winter wheat.
Specifically, in this embodiment, a fuzzy classification technique and a membership determination method based on a bayesian criterion are adopted to finally obtain membership images of each category, where only the membership image belonging to "winter wheat" is used. The calculation method is as follows:
hypothesis classification of vectors in imagesDivide the research area intomClass (A)k i ，i=1,2，…， m) According to the Bayes formula, inXConditions of occurrence ascribed tok i Class attribution probability (XFor thek i Class membership) is:
，
vector quantityXIn a categoryk i The conditional probability density function of (a) is:
，
in the formula,P(k i ) Is a categoryk i A priori probability of (a); p: (X/k i ) Is a categoryk i In-appearance vectorXThe probability of (d); n is the dimension of the feature space;mthe number of research area categories;μ i is a categoryk i The mean vector of the training samples;is a categoryk i Covariance matrix of training samplesDeterminant (c).
By using the method, the attribution probability of each pixel attributing to winter wheat is calculated on the basis of the TM image with the resolution of 25 meters, and the result is shown in FIG. 6.
B. For the phenomenon of 'same-object different-spectrum' such as spectrum difference of winter wheat caused by different conditions such as climate, growth vigor and the like, the problem can be solved by selecting more samples, and all pixels belonging to the winter wheat can be extracted.
(4) And (3) integrating the abundance map in the step (2) and membership data in the step (3) to judge the planting condition of the winter wheat in the remote sensing map
A. In each MODIS pixel, the corresponding 10 × 10 TM pixels are sorted from high to low according to the membership degree of the TM pixels to winter wheat.
Specifically, in this embodiment, as shown in fig. 4 (a), a spatial correspondence relationship between a MODIS 250-meter pixel and a resampled 25-meter TM pixel is shown, and numbers in small squares are membership degrees of the TM pixel to winter wheat, which are obtained based on a fuzzy classification technique.
B. If the abundance of the winter wheat of the MODIS pixels is F%, the first 10 × 10 pixels corresponding to the MODIS pixels in spatial position and sorted according to the membership degree of the winter wheat in the step A are judged as the winter wheat, namely the first 100 × F% pixels of the TM scale corresponding to the MODIS pixels are judged as the winter wheat, and the method is expanded to the whole research area to obtain the planting area and distribution of the winter wheat.
Specifically, if the value of the MODIS pixel in the abundance map is 58%, it indicates that the area proportion of winter wheat in the MODIS pixel is 58%, and the number of TM25 m pixels corresponding to the pixel is 100, then the number of TM scale winter wheat pixels corresponding to the pixel should be 100 × 58% =58, the 100 TM pixels are arranged from large to small according to the membership degree, the first 58 pixels are determined as winter wheat pixels, and the result is shown in fig. 4 (B), where the gray highlighted part is the TM pixel identified as winter wheat.
All pixels in the whole research area are processed, and then the winter wheat in the research area can be identified. According to the method, the recognition result of the winter wheat in the research area is calculated and obtained as shown in figure 7.
(5) Authentication
In order to verify the accuracy of the method for identifying winter wheat by remote sensing by using the key seasonal aspect characteristics and the fuzzy classification technology, the inventor further measures the identification result from two aspects of position precision and area precision, and the method is briefly introduced as follows.
The position accuracy reflects the degree of accuracy in the position of the picture elements identified as winter wheat using the method. The verification method comprises the following steps: randomly generating a plurality of verification sampling points in a research area, taking the result of visual interpretation as a true value, and defining the position precision as the ratio of the number of the sampling points correctly identified in a space position to the total number of the sampling points, wherein the calculation formula is as follows:
；
in the formula,Apposition accuracy;NUM R the number of samples for correctly identifying the winter wheat;NUM T the total number of sample points.
Area accuracy reflects how close the method identifies an area relative to a quasi-true value. The verification method comprises the following steps: and evaluating the identification result of the method by taking the planting area of the research area in the statistical yearbook as a quasi-true value. The area precision is defined as the ratio of the difference between the recognition area and the actual area subtracted from the actual planting area of the winter wheat to the actual area, and the calculation formula is as follows:
；
in the formula,Aaarea accuracy;Athe total area of winter wheat in the research area extracted by the method;A O the data is the statistical data of the winter wheat total yearbook.
The results of the verification of the positional accuracy and the area accuracy are shown in the following table.
From the verification results, the method provided by the invention achieves high precision for identifying the winter wheat planting area and the specific distribution position, and can reflect the real winter wheat planting condition more accurately and truly.
In summary, the invention obtains the exact spatial distribution result of the winter wheat under the condition of medium and high resolution by comprehensively utilizing the low-resolution and medium and high-resolution remote sensing images based on the seasonal rhythm method and the fuzzy classification technology, and simultaneously makes up the respective defects of the two methods, so that the method has important and practical application values in the aspects of remote sensing monitoring, yield evaluation and the like of the winter wheat, and simultaneously provides a new reference for remote sensing monitoring of other crop types. The remote sensing image data with medium and low resolution scales used by the invention is easy to obtain, and is easy to carry out practical monitoring application on a certain regional scale, so that the method also has good popularization and application values.
Claims (4)
1. The winter wheat remote sensing identification method integrating key seasonal features and fuzzy classification technology is characterized by comprising the following steps:
(1) data pre-processing
Completely registering data of the TM high-resolution image and the MODIS coarse-resolution image on a projection and coordinate system;
(2) preparation of abundance maps at coarse resolution of the study area
A. According to the winter wheat planting condition in a research area, under the condition of ensuring the precision, a typical sample is laid in the field to investigate the winter wheat planting area proportion in one pixel of MODIS coarse resolution scale, and a sample is provided for regression analysis; or the planting condition of the winter wheat is interpreted by utilizing the high-resolution remote sensing image, and pixel percentage data of MODIS coarse resolution scale is made as a sample on the basis of the interpretation, so that a field layout sample is replaced;
B. extracting a time period reflecting the typical phenological period of the winter wheat from the MODIS NDVI time sequence, and establishing a slope image of the typical phenological period by using MODIS NDVI data of the time period;
C. establishing a regression model by using the typical phenological period slope image obtained in the step B and the sample data obtained in the step A, and expanding the regression model to the whole research area so as to obtain an abundance map of the winter wheat planting area of the research area;
(3) obtaining the membership degree of the pixels to winter wheat under the medium and high resolution scales of the research area
A. Obtaining the membership degree of each pixel belonging to winter wheat by using a fuzzy classification technology and a membership degree determination method based on a Bayesian rule according to TM data;
B. data correction is carried out, and the phenomenon of 'same object and different spectrum' is eliminated;
(4) and (4) comprehensive judgment, namely comprehensive step (2) abundance graph and step (3) membership degree data are used for judging the planting condition of winter wheat in the remote sensing graph
A. In each MODIS pixel, the membership degree of the TM pixel corresponding to the MODIS pixel to winter wheat is ranked from high to low;
B. if the abundance of the winter wheat of one MODIS pixel is F%, determining the first F% pixels which correspond to 10 multiplied by 10 pixels in spatial position and are sorted according to the membership degree of the winter wheat in the step A under the condition of medium and high resolution as the winter wheat, and determining the F value according to the abundance map in the step (2);
C. the method is extended to the whole research area, and the planting area and distribution of winter wheat can be obtained.
2. The remote sensing identification method for winter wheat by integrating key seasonal features and fuzzy classification technology as claimed in claim 1, wherein in the step (1), the TM resolution is 25 meters, and the MODIS resolution is 250 meters; the registration process is specifically to set the TM data resampling to 25 meters, and completely register with the MODIS 250 meter data in the projection and coordinate system, so that each MODIS pixel can correspond to 10 × 10 TM pixels in spatial position.
3. The remote sensing identification method for winter wheat by integrating key seasonal aspect characteristics and fuzzy classification technology as claimed in claim 1, wherein the typical phenological period of winter wheat in the step (2) is from the green-turning period of winter wheat to the booting period of winter wheat.
4. The remote sensing identification method for winter wheat by integrating key seasonal features and fuzzy classification technology as claimed in claim 1, wherein the fuzzy classification calculation method is as follows:
hypothesis classification of vectors in imagesDivide the research area intomClass i, ik i ，i=1,2，…，mAccording to the Bayes formula, inXConditions of occurrence ascribed tok i The class attribution probability is:，
vector quantityXIn a categoryk i The conditional probability density function of (a) is:
，
in the formula,P(k i ) Is a categoryk i A priori probability of (a); p: (X/k i ) Is a categoryk i In-appearance vectorXThe probability of (d);nis the dimension of the feature space;mthe number of research area categories;μ i is a categoryk i The mean vector of the training samples;is a categoryk i Covariance matrix of training samplesDeterminant (c).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510037425.4A CN104615977B (en) | 2015-01-26 | 2015-01-26 | The winter wheat remote sensing recognition method of comprehensive crucial Aspection character and fuzzy classification technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510037425.4A CN104615977B (en) | 2015-01-26 | 2015-01-26 | The winter wheat remote sensing recognition method of comprehensive crucial Aspection character and fuzzy classification technology |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104615977A CN104615977A (en) | 2015-05-13 |
| CN104615977B true CN104615977B (en) | 2018-02-06 |
Family
ID=53150414
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510037425.4A Active CN104615977B (en) | 2015-01-26 | 2015-01-26 | The winter wheat remote sensing recognition method of comprehensive crucial Aspection character and fuzzy classification technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104615977B (en) |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104915660A (en) * | 2015-07-01 | 2015-09-16 | 中国科学院遥感与数字地球研究所 | Winter wheat automatic recognition method based on GF-1/WFV NDVI time sequence |
| CN104951772B (en) * | 2015-07-10 | 2017-12-29 | 中国科学院遥感与数字地球研究所 | A kind of winter wheat extracting method based on NDVI time-serial positions integration |
| CN107330801A (en) * | 2017-06-07 | 2017-11-07 | 北京师范大学 | The computational methods and device of a kind of winter wheat planting proportion |
| CN109493247A (en) * | 2017-09-11 | 2019-03-19 | 苏州市农业科学院 | The method that confirmation field crop rotation is lain fallow |
| CN108805079B (en) * | 2018-06-12 | 2020-10-09 | 中国科学院地理科学与资源研究所 | Winter wheat identification method and device |
| CN109242875B (en) * | 2018-10-19 | 2020-08-18 | 北京工业职业技术学院 | Winter wheat planting area extraction method and system |
| CN108984803B (en) * | 2018-10-22 | 2020-09-22 | 北京师范大学 | Method and system for spatializing crop yield |
| CN109492568B (en) * | 2018-10-31 | 2021-10-08 | 武汉珈和科技有限公司 | Method and system for extracting planting distribution of global specific crop main producing area |
| CN111709379B (en) * | 2020-06-18 | 2023-04-18 | 广西壮族自治区农业科学院 | Remote sensing image-based hilly area citrus planting land plot monitoring method and system |
| CN112347992B (en) * | 2020-12-01 | 2023-07-18 | 中国林业科学研究院 | Remote sensing estimation method for time sequence AGB in desert area |
| CN114283335B (en) * | 2021-12-27 | 2022-11-22 | 河南大学 | Historical period remote sensing identification precision verification preparation method |
| CN114529097B (en) * | 2022-02-26 | 2023-01-24 | 黑龙江八一农垦大学 | Multi-scale crop phenological period remote sensing dimensionality reduction prediction method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2004001659A1 (en) * | 2001-12-17 | 2003-12-31 | Digitalglobe Inc. | System, method, and apparatus for satellite remote sensing |
| CN102346808A (en) * | 2011-06-08 | 2012-02-08 | 北京师范大学 | Method for inverting LAI (leaf area index) from HJ-1 satellite data |
| CN102609726A (en) * | 2012-02-24 | 2012-07-25 | 中国科学院东北地理与农业生态研究所 | Method for classifying remote sensing images blended with high-space high-temporal-resolution data by object oriented technology |
| CN103033150A (en) * | 2011-10-09 | 2013-04-10 | 黄青 | Method for quickly extracting main crop planting area through utilization of moderate resolution imaging spectroradiometer (MODIS) data |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103336956B (en) * | 2013-07-10 | 2016-08-10 | 福州大学 | A kind of winter wheat area evaluation method based on remote sensing time series data |
-
2015
- 2015-01-26 CN CN201510037425.4A patent/CN104615977B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2004001659A1 (en) * | 2001-12-17 | 2003-12-31 | Digitalglobe Inc. | System, method, and apparatus for satellite remote sensing |
| CN102346808A (en) * | 2011-06-08 | 2012-02-08 | 北京师范大学 | Method for inverting LAI (leaf area index) from HJ-1 satellite data |
| CN103033150A (en) * | 2011-10-09 | 2013-04-10 | 黄青 | Method for quickly extracting main crop planting area through utilization of moderate resolution imaging spectroradiometer (MODIS) data |
| CN102609726A (en) * | 2012-02-24 | 2012-07-25 | 中国科学院东北地理与农业生态研究所 | Method for classifying remote sensing images blended with high-space high-temporal-resolution data by object oriented technology |
Non-Patent Citations (2)
| Title |
|---|
| "地块支持下MODIS-NDVI时间序列冬小麦种植面积测量研究";李乐 等;《光谱学与光谱分析》;20110531;第31卷(第5期);第1379-1383页 * |
| "模糊分类技术在作物类型识别中的应用";尤淑撑 等;《国土资源遥感》;20000315(第1期，总第43期);第39-43页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104615977A (en) | 2015-05-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104615977B (en) | The winter wheat remote sensing recognition method of comprehensive crucial Aspection character and fuzzy classification technology | |
| Yang et al. | Automated registration of dense terrestrial laser-scanning point clouds using curves | |
| CN105139023B (en) | A kind of seed recognition methods based on multi-scale feature fusion and extreme learning machine | |
| CN109284786B (en) | SAR image terrain classification method for generating countermeasure network based on distribution and structure matching | |
| CN107239795A (en) | SAR image change detecting system and method based on sparse self-encoding encoder and convolutional neural networks | |
| CN105354841B (en) | A kind of rapid remote sensing image matching method and system | |
| Peng et al. | Object-based change detection from satellite imagery by segmentation optimization and multi-features fusion | |
| Davidson et al. | A comparison of three approaches for predicting C4 species cover of northern mixed grass prairie | |
| Herrault et al. | A comparative study of geometric transformation models for the historical" Map of France" registration | |
| Aplin et al. | Predicting missing field boundaries to increase per-field classification accuracy | |
| Ostrowski et al. | Analysis of 3D building models accuracy based on the airborne laser scanning point clouds | |
| CN115439654B (en) | Method and system for finely dividing weakly supervised farmland plots under dynamic constraint | |
| Stehman et al. | Estimation of fuzzy error matrix accuracy measures under stratified random sampling | |
| CN109726679B (en) | Remote sensing classification error spatial distribution mapping method | |
| CN112924967B (en) | Remote sensing monitoring method for crop lodging based on radar and optical data combination characteristics and application | |
| Engstrom et al. | Evaluating the Relationship between Contextual Features Derived from Very High Spatial Resolution Imagery and Urban Attributes: A Case Study in Sri Lanka | |
| Bunting et al. | An area based technique for image-to-image registration of multi-modal remote sensing data | |
| Hodgson et al. | A parameterization model for transportation feature extraction | |
| Priem et al. | Use of multispectral satellite imagery and hyperspectral endmember libraries for urban land cover mapping at the metropolitan scale | |
| CN118761909B (en) | Meteorological element super-resolution image evaluation method | |
| Rajeswari et al. | Classification based land use/land cover change detection through Landsat images | |
| CN117689658B (en) | Remote sensing image change detection method and device | |
| CN114283335B (en) | Historical period remote sensing identification precision verification preparation method | |
| Liu et al. | Speed detection of moving vehicles from one scene of QuickBird images | |
| Xing et al. | A Land Cover Change Detection Method Based On Change Difference Map Fusion |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |