CN113160359A - Coastal wetland remote sensing mapping method based on harmonic fitting parameters - Google Patents
Coastal wetland remote sensing mapping method based on harmonic fitting parameters Download PDFInfo
- Publication number
- CN113160359A CN113160359A CN202110560269.5A CN202110560269A CN113160359A CN 113160359 A CN113160359 A CN 113160359A CN 202110560269 A CN202110560269 A CN 202110560269A CN 113160359 A CN113160359 A CN 113160359A
- Authority
- CN
- China
- Prior art keywords
- remote sensing
- harmonic
- image
- fitting
- coastal wetland
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000013507 mapping Methods 0.000 title claims abstract description 26
- 238000003066 decision tree Methods 0.000 claims abstract description 19
- 238000007637 random forest analysis Methods 0.000 claims abstract description 15
- 238000001914 filtration Methods 0.000 claims abstract description 13
- 230000008859 change Effects 0.000 claims abstract description 8
- 238000007781 pre-processing Methods 0.000 claims abstract description 7
- 230000006870 function Effects 0.000 claims description 54
- 238000002310 reflectometry Methods 0.000 claims description 19
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 13
- 238000011160 research Methods 0.000 claims description 12
- 238000012937 correction Methods 0.000 claims description 11
- 238000013441 quality evaluation Methods 0.000 claims description 10
- 238000003384 imaging method Methods 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims description 4
- 230000001419 dependent effect Effects 0.000 claims description 4
- 230000004660 morphological change Effects 0.000 claims description 4
- 239000000126 substance Substances 0.000 claims description 4
- 238000011156 evaluation Methods 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 8
- 238000005070 sampling Methods 0.000 description 7
- 238000002474 experimental method Methods 0.000 description 5
- 235000014676 Phragmites communis Nutrition 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 3
- 150000003839 salts Chemical class 0.000 description 3
- 108010076504 Protein Sorting Signals Proteins 0.000 description 2
- 241000202758 Scirpus Species 0.000 description 2
- 241001149258 Sporobolus alterniflorus Species 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000003595 spectral effect Effects 0.000 description 2
- 241000196324 Embryophyta Species 0.000 description 1
- 241001415959 Grus japonensis Species 0.000 description 1
- 241000918237 Suaeda glauca Species 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000008033 biological extinction Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000009545 invasion Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 239000004576 sand Substances 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 239000013535 sea water Substances 0.000 description 1
- 239000010802 sludge Substances 0.000 description 1
- 241000894007 species Species 0.000 description 1
- 239000004577 thatch Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T11/00—2D [Two Dimensional] image generation
- G06T11/60—Editing figures and text; Combining figures or text
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/80—Geometric correction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30181—Earth observation
-
- 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
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A10/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE at coastal zones; at river basins
- Y02A10/40—Controlling or monitoring, e.g. of flood or hurricane; Forecasting, e.g. risk assessment or mapping
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a coastal wetland remote sensing mapping method based on harmonic fitting parameters, which is characterized by comprising the following steps: s1, preprocessing the remote sensing image; s2, constructing time sequence filtering; s3, collecting sample pixels of the classified sample; s4, fitting a vegetation index change trend by a harmonic function; s5, classifying the random forest decision tree, and drawing a coastal wetland drawing; the method has the advantages that pixel-by-pixel time sequence filtering is constructed, and the effective observation quantity of remote sensing time sequence products on a coastal zone is remarkably improved; aiming at sparse and non-equal-time-distance remote sensing time sequences, a harmonic function is constructed to fully represent the variation trend of the remote sensing time sequences, fitting parameters are extracted to unify classification characteristics, and the method realizes the fine classification mapping of coastal wetland types.
Description
Technical Field
The invention relates to the technical field of remote sensing mapping, in particular to a coastal wetland remote sensing mapping method based on harmonic fitting parameters.
Background
The coastal wetland is a salt water or light salt water sludge beach covered with herbaceous plant communities and periodically or intermittently influenced by ocean tides, and is one of the most valuable ecological systems. However, the change of natural environment and the influence of human activities cause the coastal wetland ecosystem to undergo continuous replacement and extinction; meanwhile, the invasion of foreign species also enables the wetland type landscape pattern to be constantly changed. The method can accurately obtain the spatial distribution of the coastal wetland type, provide scientific basis for protecting the coastal zone ecosystem, and serve the stable and continuous development of coastal economy and society.
Remote sensing has macroscopicity and repeatability, and therefore becomes an important means for coastal zone monitoring. With the continuous development of remote sensing technology, in recent years, quality evaluation bands are distributed in the standard data of the divided images, and the coverage conditions of image clouds (shadows) and snow at the imaging time are recorded pixel by pixel. The quality evaluation wave band is beneficial to screening useful information of the image pixel by pixel, the effective observation of a time sequence product is obviously promoted, and the possibility of distinguishing wetland types by using a remote sensing time sequence method is provided. However, due to the randomness of cloud layer spatial distribution at the imaging moment, effective observation of each pixel time sequence is often sparsely distributed, and time intervals of adjacent observations are not equal to each other, so that the existing classification mode using each observation as a feature is not applicable any more. How to extract stable characteristics on a remote sensing time sequence with sparse and unequal time intervals and carry out large-range and fine coastal wetland mapping is a main problem at present.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a coastal wetland remote sensing mapping method which extracts steady characteristics on a continuous time sequence constructed by harmonic function fitting and performs large-range and fine mapping, and has the advantages of wide application range, high classification precision and strong robustness.
The technical scheme adopted by the invention for solving the technical problems is as follows: a coastal wetland remote sensing mapping method based on harmonic fitting parameters comprises the following steps:
s1, preprocessing the remote sensing image;
s2, constructing time sequence filtering;
s3, collecting sample pixels of the classified sample;
s4, fitting a vegetation index change trend by a harmonic function;
and S5, classifying the random forest decision tree, and drawing the coastal wetland drawing.
Preferably, the remote sensing image preprocessing in step S1 specifically includes the following steps:
s11, obtaining a remote sensing image of a research area at a specified time, performing geometric fine correction on the image, and converting an image pixel value into a surface reflectivity through radiometric calibration and atmospheric correction to obtain a surface reflectivity image;
s12, automatically generating a roughly classified product of each scene image as a quality evaluation waveband, and performing geometric fine correction on the quality evaluation waveband;
and S13, cutting the converted earth surface reflectivity image and the roughly classified product to obtain image data covering the research area.
Preferably, the time-series filtering construction in step S2 specifically includes the following steps:
s21, converting the earth surface reflectivity image of each scene into a vegetation index image;
s22, eliminating noise pixels in the vegetation index image;
and S23, arranging the vegetation index images according to the imaging time sequence to construct a time sequence.
Preferably, the specific steps of removing the noise pixel in the vegetation index image in step S22 are as follows:
s221, removing pixels covered by clouds, cloud shadows and snow by using a roughly classified product corresponding to each vegetation index image so as to remove cloud and snow noise;
s222, eliminating pixels of the tidal level interference in the vegetation index image by using the improved normalized difference water body index.
Preferably, the method for collecting the sample pixels of the classified samples in step S3 is field sample collection or image interpretation collection, so as to obtain the corresponding sample pixels on the vegetation index image.
Preferably, the step of fitting the harmonic function to the vegetation index change trend in the step S4 specifically includes the following steps: and for each sample pixel, taking the effective observed value of the vegetation index as a dependent variable, taking the julian day corresponding to the date obtained by the remote sensing image as an independent variable, and fitting the relationship between the effective observed value of the vegetation index and the julian day by adopting a harmonic function.
Preferably, the fitting formula of the harmonic function in step S4 is:
wherein the content of the first and second substances,in the case of the julian day,for the purpose of the frequency of the function,、、、、respectively are undetermined coefficients of harmonic functions, are obtained by least square fitting, the undetermined coefficients corresponding to different types of vegetation of the coastal wetland are different,the overall height of the fitting function is determined,、、、together determine the morphological change of the fitting function.
Preferably, in step S4, a threshold is set for the number of effective observation values in the sample pixel, and when the number of effective observation values is smaller than the threshold, the fitting of the harmonic function fails, and a harmonic fitting parameter value is set manually.
Preferably, the specific steps of the random forest decision tree classification in step S5 are as follows:
s51, constructing a random forest classifier by taking the fitting parameters of the harmonic function as initial classification features and taking the fitting parameter data of the harmonic function as input variables;
and S52, drawing the coastal wetland drawing according to the category attribution of each pixel.
Preferably, the specific drawing step of the coastal wetland drawing in the step S52 is as follows:
s521, selecting partial sample pixels, determining the category attribution of each pixel on the image by using a random forest classifier, and drawing a primary classification map of the coastal wetland according to the category attribution of each pixel;
s522, taking the rest sample pixels as decision tree precision evaluation;
s523, the sparse and small-area pattern spots distributed in the primary classification pattern are removed and merged by utilizing filtering processing and clustering processing, so that the classification result is closer to reality.
Compared with the prior art, the method has the advantages that pixel-by-pixel time sequence filtering is constructed, the effective observation quantity of remote sensing time sequence products on a coastal zone is remarkably improved, and conditions are created for fine classification mapping by utilizing the remote sensing time sequence; aiming at sparse and non-equal-time-distance remote sensing time sequences, a harmonic function is constructed to fully represent the variation trend of the remote sensing time sequences, fitting parameters are extracted to unify classification characteristics, and the method realizes the fine classification mapping of coastal wetland types.
Drawings
FIG. 1 is a flow chart of the present invention;
fig. 2(a) -2 (d) are graphs of an original vegetation index of a reed pixel, a julian day, a vegetation index after cloud (shadow) noise is removed, a vegetation index after tide level interference is removed, and a variation trend of a harmonic function fitting vegetation index according to an embodiment of the present invention;
FIG. 3(a) is a sample original classification feature distribution diagram according to a first embodiment of the present invention;
FIG. 3(b) is a classification feature spatial distribution diagram of the fitting parameters after the harmonic function fitting according to the first embodiment of the present invention;
fig. 4 is a distribution diagram of a sampling method according to a first embodiment of the present invention;
FIG. 5 is a diagram of a coastal wetland remote sensing map according to a first embodiment of the invention;
fig. 6(a) -6 (d) are graphs of the original vegetation index of scirpus marigoldenrod, the julian day, the vegetation index after cloud (shadow) noise removal, the vegetation index after tide level interference removal, and the variation trend of the vegetation index fitted by a harmonic function, respectively, according to the second embodiment of the present invention;
FIG. 7(a) is a sample original classification feature distribution diagram according to a second embodiment of the present invention;
FIG. 7(b) is a classification feature spatial distribution diagram of the fitting parameters after the harmonic function fitting according to the second embodiment of the present invention;
FIG. 8 is a distribution diagram of the collected samples according to the second embodiment of the present invention;
fig. 9 is a coastal wetland remote sensing map according to a second embodiment of the invention.
Detailed Description
The invention is described in further detail below with reference to the accompanying examples.
The first embodiment is as follows: as shown in fig. 1-5, a coastal wetland remote sensing mapping method based on harmonic fitting parameters, which takes coastal wetland in the middle of Jiangsu as a research area, comprises the following steps:
s1, remote sensing image preprocessing:
downloading and acquiring a remote sensing image of a Sentinel-2 MSI L1C product which is imaged in a research area in 2018 and is geometrically corrected corresponding to a 68 scene with the line code number of T51STT through a website of the European Bureau, geometrically and finely correcting the image, carrying out radiometric calibration and atmospheric correction through Sen2Cor software provided in an European space, and converting an image pixel value (DN) into an earth surface reflectivity ((DN) ((earth surface reflectivity)) Obtaining an earth surface reflectivity image, automatically generating a coarse Classification product (SCL) of each scene image as a quality evaluation waveband, performing geometric fine correction processing on the quality evaluation waveband, and cutting the converted earth surface reflectivity image and the coarse Classification product on ArcGIS software to obtain image data covering a research area;
s2, time series filtering construction:
converting the reflectivity image of each scene of earth surface into a vegetation index image, then eliminating the pixels covered by clouds, cloud shadows and snow by using the rough classification product (SCL) corresponding to the vegetation index image, namely eliminating the pixel with the value of 3 (corresponding to the cloud shadows) on the rough classification product (SCL),8-10 (corresponding to clouds with different probabilities) and 11 (corresponding to snow), and then eliminating pixels with tidal level interference in the vegetation index image by using the improved normalized difference water body index, wherein the calculation formula of the improved normalized water body index is as follows:, andrespectively setting the earth surface spectral reflectivities of a green light wave band and a short wave infrared wave band, namely setting the threshold value of the improved normalized difference water body index to be 0, removing image pixels of which the threshold value of the improved normalized difference water body index is more than 0 at each period, and then arranging 68 scene vegetation index images according to an imaging time sequence to construct a time sequence;
s3, collecting sample pixels of the classified sample prescription:
the classification sample prescription of the research area mainly adopts a field sample prescription collection mode, the collection work is carried out between 2017 and 2018, the classification sample prescription is mainly positioned in a natural protection area of a red-crowned crane in a salt city, the sample prescription selects an area with single coastal wetland type and coverage degree exceeding 20 multiplied by 20 m (corresponding to 4 pixels of a Sentinel-2 MSI optical wave band), the collection type comprises spartina alterniflora, suaeda glauca, reed and thatch, 526 blocks of field survey sample prescriptions are obtained in total, each corner coordinate of the sample prescription is recorded by a GPS handset to determine the position of the sample prescription, the sample prescription is projected to WGS84 UTM 51N, in addition, in order to ensure the classification result scientific investigation of the research area, 69 samples of mudflats and water bodies are collected by an image interpretation mode, and 595 samples are used as sample pixels in total;
s4, fitting vegetation index change trend by using a harmonic function:
for each sample pixel, the vegetation index effective observed value is used as a dependent variable, the julian day corresponding to the date of obtaining the remote sensing image is used as an independent variable, a harmonic function is adopted to fit the relation between the vegetation index effective observed value and the julian day, the harmonic function is defined by using the self-defined function of a Matlab software Curve fixing module, and the Fitting formula of the harmonic function is as follows:
wherein the content of the first and second substances,in the case of the julian day,for the purpose of the frequency of the function,、、、、respectively are undetermined coefficients of harmonic functions, are obtained by least square fitting, the undetermined coefficients corresponding to different types of vegetation of the coastal wetland are different,the overall height of the fitting function is determined,、、、a morphological change which jointly determines the fit function, its initial value [ [ alpha ] ],,,,,]Is set to [0, 0, 0, 0, 0, 0.085]The upper and lower bounds of the fit are set to [ Inf, Inf, Inf, Inf, Inf, 0.01, respectively]、[-Inf,-Inf,-Inf,-Inf,-Inf,0.07]In the experiment, the threshold value of the number of the effective observation values is set to be 6, the number of the effective observation values in the pixel is less than 6, so that the fitting is invalid, and at the moment, the fitting parameter values need to be set to be [0, 0, 0, 0 ] manually]In an experiment, only 0.92 per thousand of effective observed values of pixels are less than 6, so that the fitting is invalid, the pixels mostly correspond to tidal flat pixels frequently interacting with seawater, at the moment, fitting parameter values are manually set, the constraint force of a harmonic fitting function on the parameters is small, the fitting success rate is high, the time cost is low, the method is not only suitable for the field of remote sensing mapping, but also can provide beneficial reference for sparse signal sequence characteristic expression in the information field, on the basis, various coastal wetland sampling pixels are combined, a Jeffries-Matusita distance is calculated by using an ROI Separability tool of ENVI 5.1 software, and the Separability of various coastal wetlands is found to be greater than 1.8 (good Separability), so that 6 fitting parameters of the harmonic function can be directly used as final classification characteristics;
s5, random forest decision tree classification:
the method comprises the steps of taking fitting parameter data of a harmonic function as input variables, taking fitting parameters of the harmonic function as initial classification features, constructing a random forest decision tree by using a randomForest extension packet of an R language, selecting about 50% (298) of sampling pixels of the coastal wetland for constructing the decision tree, and using the remaining 50% (297) of the sampling pixels for evaluating the precision of the decision tree. The number of feature choices of each decision tree is set to be 3, the minimum sample number of leaf nodes is set to be 15, 100 decision trees are established to comprehensively judge the category attribution of each pixel, and a primary classification chart is obtained. Using a Sieve tool of ENVI 5.1 software to carry out primary classification map filtering treatment, setting 8 neighborhood analysis methods to judge and remove spot pixels in the classification map; and (3) clustering the classification diagram by using a column tool, setting a 3 multiplied by 3 sliding window, clustering and combining the types of adjacent pixels by using a mathematical morphology operator to obtain the final coastal wetland drawing, wherein the total precision reaches 84.4%.
Example two: as shown in fig. 1, 6-9, a coastal wetland remote sensing mapping method based on harmonic fitting parameters takes nine sections of sand wetland at the Yangtze estuary as a research area, and comprises the following steps:
s1, remote sensing image preprocessing:
downloading and acquiring a remote sensing image of a research area imaged in 2018 corresponding to a 72 scene with a line code number T51RUQ from a Sentinel-2 MSI L1C product through an European space website, performing geometric fine correction on the image, performing radiation calibration and atmospheric correction through Sen2Cor software provided in an European space, converting an image pixel value (DN) into an earth surface reflectivity (rho) to obtain an earth surface reflectivity image, automatically generating a coarse Classification product (SCL) of each scene image as a quality evaluation waveband, performing geometric fine correction on the quality evaluation waveband, and cutting the converted earth surface reflectivity and the coarse Classification product on ArcGIS software to obtain image data covering the research area;
s2, time series filtering construction:
converting each scene of the earth surface reflectivity image into a vegetation index image, and then eliminating pixels covered by clouds, cloud shadows and snow by using a rough classification product (SCL) corresponding to each scene of the vegetation index image, namely eliminating the numerical value of 3 on the rough classification product (SCL)Corresponding to cloud shadow), 8-10 (corresponding to clouds with different probabilities), and 11 (corresponding to snow), and then eliminating pixels of tidal level interference in vegetation index images by using the improved normalized difference water body index, wherein the calculation formula of the improved normalized water body index is as follows: , ,andrespectively the earth surface spectral reflectivity of a green light wave band and a short wave infrared wave band, modifying the MNDWI threshold value to-0.1 due to larger fluctuation range of the tidal level of the Yangtze river mouth, removing image pixels with the threshold value of the improved normalized difference water body index larger than-0.1 at each period, and then arranging 72-scene vegetation index images according to the imaging time sequence to construct a time sequence;
s3, collecting sample pixels of the classified sample prescription:
the experiment adopts a Google Earth high-definition image interpretation mode to collect coastal wetland samples, the corresponding time of the high-definition image is 2017-2018 years, the collection types comprise spartina alterniflora, scirpus maritima, reed, mudflat and water, and total 566 sample pixels are used for subsequent classification;
s4, fitting vegetation index change trend by using a harmonic function:
for each sample pixel, the vegetation index effective observed value is used as a dependent variable, the julian day corresponding to the date of obtaining the remote sensing image is used as an independent variable, a harmonic function is adopted to fit the relation between the vegetation index effective observed value and the julian day, the harmonic function is defined by using the self-defined function of a Matlab software Curve fixing module, and the Fitting formula of the harmonic function is as follows:
wherein the content of the first and second substances,in the case of the julian day,for the purpose of the frequency of the function,、、、、respectively are undetermined coefficients of harmonic functions, are obtained by least square fitting, the undetermined coefficients corresponding to different types of vegetation of the coastal wetland are different,the overall height of the fitting function is determined,、、、a morphological change which jointly determines the fit function, its initial value [ [ alpha ] ],,,,,]Is set to [0, 0, 0, 0, 0, 0.085]The upper and lower bounds of the fit are set to [ Inf, Inf, Inf, Inf, Inf, 0.01, respectively]、[-Inf,-Inf,-Inf,-Inf,-Inf,0.07]In the experiment, the threshold value of the number of the effective observation values is set to be 6, the fitting failure can be caused when the number of the effective observation values in the pixel is less than 6, and at the moment, the fitting parameter values need to be set to be [0, 0, 0, 0 ] artificially]In an experiment, only 1.31 per thousand of effective observed values of pixels are less than 6, so that the fitting is invalid, at the moment, fitting parameter values are manually set, the harmonic fitting function has less constraint on the parameters, the fitting success rate is higher, the time cost is lower, the method is not only suitable for the field of remote sensing mapping, but also can provide beneficial reference for sparse signal sequence characteristic expression in the information field, on the basis, the Jeffries-Matusita distance is calculated by combining various coastal wetland sampling pixels and utilizing an ROI Separability tool of ENVI 5.1 software, and the Separability of various coastal wetlands is greater than 1.8 (good Separability) so that 6 fitting parameters of the harmonic function can be directly used as final classification characteristics;
s5, random forest decision tree classification:
taking the fitting parameter data of the harmonic function as an input variable, taking the fitting parameter of the harmonic function as an initial classification characteristic, constructing a random forest decision tree by using a randomForest extension packet of an R language, selecting about 50 percent (283) of sampling pixels of the coastal wetland for constructing the decision tree, and using the remaining 50 percent (283) of sampling pixels for evaluating the precision of the decision tree. The number of feature choices of each decision tree is set to be 3, the minimum sample number of leaf nodes is set to be 15, 100 decision trees are established to comprehensively judge the category attribution of each pixel, and a primary classification chart is obtained. Using a Sieve tool of ENVI 5.1 software to carry out primary classification map filtering treatment, setting 8 neighborhood analysis methods to judge and remove spot pixels in the classification map; and (3) clustering the classification map by using a column tool, setting a 3 multiplied by 3 sliding window, clustering and combining the types of adjacent pixels by using a mathematical morphology operator to obtain the final coastal wetland drawing, wherein the total precision reaches 89.8%.
The scope of the present invention includes, but is not limited to, the above embodiments, and the present invention is defined by the appended claims, and any alterations, modifications, and improvements that may occur to those skilled in the art are all within the scope of the present invention.
Claims (10)
1. A coastal wetland remote sensing mapping method based on harmonic fitting parameters is characterized by comprising the following steps:
s1, preprocessing the remote sensing image;
s2, constructing time sequence filtering;
s3, collecting sample pixels of the classified sample;
s4, fitting a vegetation index change trend by a harmonic function;
and S5, classifying the random forest decision tree, and drawing the coastal wetland drawing.
2. The coastal wetland remote sensing mapping method based on harmonic fitting parameters of claim 1, characterized in that: the remote sensing image preprocessing in the step S1 specifically includes the following steps:
s11, obtaining a remote sensing image of a research area at a specified time, performing geometric fine correction on the image, and converting an image pixel value into a surface reflectivity through radiometric calibration and atmospheric correction to obtain a surface reflectivity image;
s12, automatically generating a roughly classified product of each scene image as a quality evaluation waveband, and performing geometric fine correction on the quality evaluation waveband;
and S13, cutting the converted earth surface reflectivity image and the roughly classified product to obtain image data covering the research area.
3. The coastal wetland remote sensing mapping method based on harmonic fitting parameters of claim 1, characterized in that: the time series filtering construction in the step S2 specifically includes the following steps:
s21, converting the earth surface reflectivity image of each scene into a vegetation index image;
s22, eliminating noise pixels in the vegetation index image;
and S23, arranging the vegetation index images according to the imaging time sequence to construct a time sequence.
4. The coastal wetland remote sensing mapping method based on harmonic fitting parameters of claim 3, characterized in that: the specific steps of eliminating the noise pixel in the vegetation index image in the step S22 are as follows:
s221, removing pixels covered by clouds, cloud shadows and snow by using a roughly classified product corresponding to each vegetation index image so as to remove cloud and snow noise;
s222, eliminating pixels of the tidal level interference in the vegetation index image by using the improved normalized difference water body index.
5. The coastal wetland remote sensing mapping method based on harmonic fitting parameters of claim 1, characterized in that: the method for collecting the sample pixels of the classified samples in step S3 is field sample collection or image interpretation collection, so as to obtain the corresponding sample pixels on the vegetation index image.
6. The coastal wetland remote sensing mapping method based on harmonic fitting parameters of claim 1, characterized in that: the step S4 of fitting the harmonic function to the vegetation index change trend specifically includes the following steps: and for each sample pixel, taking the effective observed value of the vegetation index as a dependent variable, taking the julian day corresponding to the date obtained by the remote sensing image as an independent variable, and fitting the relationship between the effective observed value of the vegetation index and the julian day by adopting a harmonic function.
7. The coastal wetland remote sensing mapping method based on harmonic fitting parameters of claim 6, characterized in that: the fitting formula of the harmonic function in step S4 is:
wherein the content of the first and second substances,in the case of the julian day,for the purpose of the frequency of the function,、、、、respectively are undetermined coefficients of harmonic functions, are obtained by least square fitting, the undetermined coefficients corresponding to different types of vegetation of the coastal wetland are different,the overall height of the fitting function is determined,、、、together determine the morphological change of the fitting function.
8. The coastal wetland remote sensing mapping method based on harmonic fitting parameters of claim 6, characterized in that: in step S4, a threshold is set for the number of effective observation values in the sample pixel, and when the number of effective observation values is smaller than the threshold, the fitting of the harmonic function fails, and a harmonic fitting parameter value is set manually.
9. The coastal wetland remote sensing mapping method based on harmonic fitting parameters of claim 1, characterized in that: the specific steps of the random forest decision tree classification in the step S5 are as follows:
s51, constructing a random forest classifier by taking the fitting parameters of the harmonic function as initial classification features and taking the fitting parameter data of the harmonic function as input variables;
and S52, drawing the coastal wetland drawing according to the category attribution of each pixel.
10. The coastal wetland remote sensing mapping method based on harmonic fitting parameters of claim 9, characterized in that: the specific drawing steps of the coastal wetland drawing in the step S52 are as follows:
s521, selecting partial sample pixels, determining the category attribution of each pixel on the image by using a random forest classifier, and drawing a primary classification map of the coastal wetland according to the category attribution of each pixel;
s522, taking the rest sample pixels as decision tree precision evaluation;
s523, the sparse and small-area pattern spots distributed in the primary classification pattern are removed and merged by utilizing filtering processing and clustering processing, so that the classification result is closer to reality.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110560269.5A CN113160359A (en) | 2021-05-21 | 2021-05-21 | Coastal wetland remote sensing mapping method based on harmonic fitting parameters |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110560269.5A CN113160359A (en) | 2021-05-21 | 2021-05-21 | Coastal wetland remote sensing mapping method based on harmonic fitting parameters |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113160359A true CN113160359A (en) | 2021-07-23 |
Family
ID=76877648
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110560269.5A Pending CN113160359A (en) | 2021-05-21 | 2021-05-21 | Coastal wetland remote sensing mapping method based on harmonic fitting parameters |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113160359A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114511790A (en) * | 2022-02-11 | 2022-05-17 | 广东海启星海洋科技有限公司 | Method and device for realizing marine oil spill event monitoring based on optical satellite data |
CN114612793A (en) * | 2022-02-22 | 2022-06-10 | 中国自然资源航空物探遥感中心 | Multi-temporal remote sensing coastline and tidal flat detection method based on high-frequency observation of water sideline |
CN115795255A (en) * | 2022-09-21 | 2023-03-14 | 深圳大学 | Method, device, medium and terminal for detecting time series change of wetland |
CN115810155A (en) * | 2023-01-18 | 2023-03-17 | 中关村睿宸卫星创新应用研究院 | Tidal wetland classification method |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2020015326A1 (en) * | 2018-07-19 | 2020-01-23 | 山东科技大学 | Remote sensing image cloud shadow detection method supported by earth surface type data |
CN111402169A (en) * | 2020-03-23 | 2020-07-10 | 宁波大学 | Method for repairing remote sensing vegetation index time sequence under influence of coastal tide |
-
2021
- 2021-05-21 CN CN202110560269.5A patent/CN113160359A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2020015326A1 (en) * | 2018-07-19 | 2020-01-23 | 山东科技大学 | Remote sensing image cloud shadow detection method supported by earth surface type data |
CN111402169A (en) * | 2020-03-23 | 2020-07-10 | 宁波大学 | Method for repairing remote sensing vegetation index time sequence under influence of coastal tide |
Non-Patent Citations (1)
Title |
---|
CHAO SUN等: "Plant species classification in salt marshes using phenological parameters derived from Sentinel-2 pixel-differential time-series", 《ELSEVIER》 * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114511790A (en) * | 2022-02-11 | 2022-05-17 | 广东海启星海洋科技有限公司 | Method and device for realizing marine oil spill event monitoring based on optical satellite data |
CN114511790B (en) * | 2022-02-11 | 2022-09-30 | 广东海启星海洋科技有限公司 | Method and device for realizing marine oil spill event monitoring based on optical satellite data |
CN114612793A (en) * | 2022-02-22 | 2022-06-10 | 中国自然资源航空物探遥感中心 | Multi-temporal remote sensing coastline and tidal flat detection method based on high-frequency observation of water sideline |
CN115795255A (en) * | 2022-09-21 | 2023-03-14 | 深圳大学 | Method, device, medium and terminal for detecting time series change of wetland |
CN115795255B (en) * | 2022-09-21 | 2024-03-26 | 深圳大学 | Method, device, medium and terminal for detecting time sequence change of wetland |
CN115810155A (en) * | 2023-01-18 | 2023-03-17 | 中关村睿宸卫星创新应用研究院 | Tidal wetland classification method |
CN115810155B (en) * | 2023-01-18 | 2023-04-18 | 中关村睿宸卫星创新应用研究院 | Tidal wetland classification method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113160359A (en) | Coastal wetland remote sensing mapping method based on harmonic fitting parameters | |
Ramachandra et al. | Geographic Resources Decision Support System for land use, land cover dynamics analysis | |
CN113033670B (en) | Rice planting area extraction method based on Sentinel-2A/B data | |
CN111832518B (en) | Space-time fusion-based TSA remote sensing image land utilization method | |
US20210318253A1 (en) | Method for retrieving atmospheric aerosol based on statistical segmentation | |
Chang et al. | Multisensor satellite image fusion and networking for all-weather environmental monitoring | |
Huang et al. | Using remote sensing of land cover change in coastal watersheds to predict downstream water quality | |
CN111310681A (en) | Mangrove forest distribution remote sensing extraction method integrated with geoscience knowledge | |
Pimple et al. | A history of the rehabilitation of mangroves and an assessment of their diversity and structure using Landsat annual composites (1987–2019) and transect plot inventories | |
Tu et al. | A 30 m annual cropland dataset of China from 1986 to 2021 | |
Wang et al. | Continuous change detection of flood extents with multisource heterogeneous satellite image time series | |
Yang et al. | Monitoring water color anomaly of lakes based on an integrated method using Landsat-8 OLI images | |
Chen et al. | Fractional monitoring of desert vegetation degradation, recovery, and greening using optimized multi-endmembers spectral mixture analysis in a dryland basin of Northwest China | |
Abbas et al. | An assessment of status and distribution of mangrove forest cover in Pakistan | |
CN116702065B (en) | Method and system for monitoring ecological treatment pollution of black and odorous water based on image data | |
Nelson et al. | Spatial statistical techniques for aggregating point objects extracted from high spatial resolution remotely sensed imagery | |
Fisette et al. | Methodology for a Canadian agricultural land cover classification | |
CN116341932A (en) | Tidal flat change monitoring method and system based on double remote sensing time sequence indexes | |
Huang et al. | Comparing the effects of temporal features derived from synthetic time-series NDVI on fine land cover classification | |
Yang et al. | A comparative analysis of machine learning methods for algal Bloom detection using remote sensing images | |
Verma et al. | Comparative evaluation of change detection techniques based on multispectral images for measuring land cover dynamics of mango (Mangifera indica L.) in Malihabad Lucknow (UP) | |
Patil et al. | Detection and delineation of critical areas using echelons and spatial scan statistics with synoptic cellular data | |
Dong et al. | Optimization of post-classification processing of high-resolution satellite image: A case study | |
Raxana et al. | Spatio temporal change detection analysis of land use/land cover (LULC) using geospatial tools in the Kuzhithuraiyar sub-basin, Kodayar River, Kanniyakumari district, Tamilnadu, India | |
Pardeshi et al. | Water Quality Analysis from Satellite Images |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20210723 |