CN110032963A - The dynamic monitoring method of Spartina alterniflora's new life patch - Google Patents
The dynamic monitoring method of Spartina alterniflora's new life patch Download PDFInfo
- Publication number
- CN110032963A CN110032963A CN201910270990.3A CN201910270990A CN110032963A CN 110032963 A CN110032963 A CN 110032963A CN 201910270990 A CN201910270990 A CN 201910270990A CN 110032963 A CN110032963 A CN 110032963A
- Authority
- CN
- China
- Prior art keywords
- image
- patch
- spartina alterniflora
- new life
- spot
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 241001149258 Sporobolus alterniflorus Species 0.000 title claims abstract description 88
- 238000000034 method Methods 0.000 title claims abstract description 71
- 238000012544 monitoring process Methods 0.000 title claims abstract description 38
- 238000005259 measurement Methods 0.000 claims abstract description 29
- 230000008569 process Effects 0.000 claims description 20
- 230000011218 segmentation Effects 0.000 claims description 20
- 230000006870 function Effects 0.000 claims description 17
- 238000012545 processing Methods 0.000 claims description 15
- 238000012549 training Methods 0.000 claims description 13
- 238000012937 correction Methods 0.000 claims description 12
- 238000001228 spectrum Methods 0.000 claims description 12
- 238000000605 extraction Methods 0.000 claims description 11
- 230000000007 visual effect Effects 0.000 claims description 11
- 230000005855 radiation Effects 0.000 claims description 10
- 238000013507 mapping Methods 0.000 claims description 5
- 238000004458 analytical method Methods 0.000 claims description 4
- 238000012946 outsourcing Methods 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 claims description 4
- 238000012952 Resampling Methods 0.000 claims description 3
- 230000004913 activation Effects 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 3
- 230000006835 compression Effects 0.000 claims description 3
- 238000007906 compression Methods 0.000 claims description 3
- 238000005520 cutting process Methods 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 claims description 3
- 230000003595 spectral effect Effects 0.000 claims description 3
- 230000035899 viability Effects 0.000 claims description 3
- 240000007594 Oryza sativa Species 0.000 claims description 2
- 235000007164 Oryza sativa Nutrition 0.000 claims description 2
- 235000009566 rice Nutrition 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 abstract description 9
- 230000008859 change Effects 0.000 abstract description 7
- 230000007613 environmental effect Effects 0.000 abstract description 3
- 238000001514 detection method Methods 0.000 abstract description 2
- 230000010354 integration Effects 0.000 abstract description 2
- 238000009792 diffusion process Methods 0.000 description 8
- 230000003321 amplification Effects 0.000 description 6
- 238000003199 nucleic acid amplification method Methods 0.000 description 6
- 230000002265 prevention Effects 0.000 description 6
- 241000746413 Spartina Species 0.000 description 5
- 238000013135 deep learning Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 4
- 239000000284 extract Substances 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 230000005200 bud stage Effects 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000002708 enhancing effect Effects 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 238000011835 investigation Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 241000894007 species Species 0.000 description 2
- 241000196324 Embryophyta Species 0.000 description 1
- PEDCQBHIVMGVHV-UHFFFAOYSA-N Glycerine Chemical compound OCC(O)CO PEDCQBHIVMGVHV-UHFFFAOYSA-N 0.000 description 1
- 240000007058 Halophila ovalis Species 0.000 description 1
- 240000004928 Paspalum scrobiculatum Species 0.000 description 1
- 235000003675 Paspalum scrobiculatum Nutrition 0.000 description 1
- 235000014676 Phragmites communis Nutrition 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- QTCANKDTWWSCMR-UHFFFAOYSA-N costic aldehyde Natural products C1CCC(=C)C2CC(C(=C)C=O)CCC21C QTCANKDTWWSCMR-UHFFFAOYSA-N 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000007850 degeneration Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000008029 eradication Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 235000012779 flatbread Nutrition 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 230000009545 invasion Effects 0.000 description 1
- ISTFUJWTQAMRGA-UHFFFAOYSA-N iso-beta-costal Natural products C1C(C(=C)C=O)CCC2(C)CCCC(C)=C21 ISTFUJWTQAMRGA-UHFFFAOYSA-N 0.000 description 1
- 238000003475 lamination Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4053—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- 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
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
-
- 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
- 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
- G06T2207/30188—Vegetation; Agriculture
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Geometry (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Image Processing (AREA)
Abstract
The invention discloses the dynamic monitoring methods of Spartina alterniflora's new life patch, belong to environmental monitoring technology field;It is complicated present invention is primarily directed to solve estuarine wetland environment, the field monitoring and measurement work of Spartina alterniflora's new life patch are difficult to carry out, and now application is more is limitation of the middle space definition satellite due to its spatial resolution, the detection requirement of landscape scale can only be met, can not identify newborn tiny patch;Using single platform high-resolution satellite since the weather of its revisiting period and river mouth cloud-prone and raining limits, it is unfavorable for the Monitoring on Dynamic Change of newborn patch, and multi-platform high-resolution satellite has different spatial resolutions, directly respectively using being unable to get accurately newborn patch identification and measurement, for this purpose, it is proposed that a kind of a wide range of, untouchable Spartina alterniflora's new life patch dynamics monitoring method based on multi-platform high resolution satellite remote sensing image integration technology.
Description
Technical field
The present invention relates to environmental monitoring technology field more particularly to the dynamic monitoring methods of Spartina alterniflora's new life patch.
Background technique
Ecotone of the seashore wetland between terrestrial ecosystems and marine ecosystems, bio-diversity are rich
Richness provides important Ecosystem Service;Simultaneously as the double influence of Nature and Man class, seashore wetland be also be sensitive
Ecologically fragile areas, Spartina alterniflora is the important foreign term species of China's seashore wetland, is put into former state in 2003
First invasive species list that environmental protection general bureau, family announces, currently, Spartina alterniflora is spread in China coast large area, it is fast
Speed diffusion significantly reduces the bio-diversity of seashore wetland, causes natural vegetation mortality, aggravates the degeneration of sea grass bed,
So as to cause serious problems of ecological security, it is huge that effective prevention and control for Spartina alterniflora are that current ecological environment department faces
Hang-up.
The diffusion of Spartina alterniflora is divided into diffusion over long distances and short distance diffusion, in general, is divided into three phases: mutually spending rice
Grass seed successfully settles down form new tiny patch first, non-conterminous between each patch to connect;Then, tiny patch is with " booth
Large flat bread " mode constantly expands outwardly to form big patch;Finally, patch connection slabbing continues to expand.Some researches show that mutual
The field planting diffusion initial stage of flower spartina takes it necessary weed eradication measure, can greatly improve the prevention and control effect of Spartina alterniflora's invasion
Rate.Therefore, EARLY RECOGNITION Spartina alterniflora new life patch and timely dynamic monitoring is carried out to its diffusion velocity, for Spartina alterniflora's
Efficient prevention and control play a significant role.
Monitoring and measurement for newborn patch, mostly at present is dimension by traditional field investigation, to small patch, face
Product, perimeter measure.The method of field investigation not only takes time and effort, and can only realize the measurement to research area's localised patches.
In addition, estuarine wetland environment is complicated, many areas are difficult to enter, and more increase difficulty to field measurement.
Satellite remote sensing technology is gradually become by researcher applied to landscape scale vegetation because of its multidate, large-scale feature
In the monitoring of change.It is space definition satellite in Landsat etc. that application is more at present, due to the limitation of its spatial resolution,
It can not identify newborn small patch, it could can only be identified after Spartina alterniflora is spread in flakes.High spatial resolution is distant
Feel data, especially sub-meter grade resolution remote sense data, the identification for newborn tiny patch provides possibility.Single satellite due to
The limitation of the weather of its revisiting period and river mouth cloud-prone and raining, is unfavorable for the Monitoring on Dynamic Change of newborn patch.Multi-source, multi-platform height
Definition satellite data are used in combination, then can help to solve the problems, such as this.High resolution satellite remote sensing image has different skies more
Between resolution ratio, directly will increase the error and uncertainty of patch identification and monitoring using these data, to solve above-mentioned deficiency,
The present invention will propose a kind of a wide range of, untouchable Spartina alterniflora based on multi-platform high resolution satellite remote sensing image integration technology
Newborn patch dynamics monitoring method.
Summary of the invention
The purpose of the present invention is to provide the dynamic monitoring methods of Spartina alterniflora's new life patch, to solve above-mentioned background technique
The problem of middle proposition:
(1) estuarine wetland environment is complicated, and many places are difficult to enter.The field monitoring and measurement of Spartina alterniflora's new life patch
It takes time and effort, and is difficult to understand the distribution situation of newborn patch on the whole;
(2) what application was more is limitation of the middle space definition satellite due to its spatial resolution, can only meet landscape ruler
The detection requirement of degree can not identify newborn tiny patch;
(3) single platform high-resolution satellite is unfavorable for due to the limitation of the weather of its revisiting period and river mouth cloud-prone and raining
The Monitoring on Dynamic Change of newborn patch;
(4) multi-platform high-resolution satellite has different spatial resolutions, directly accurate using being unable to get respectively
Newborn patch identification and measurement.
To achieve the goals above, present invention employs following technical solutions:
1. the dynamic monitoring method of Spartina alterniflora's new life patch, comprising the following steps:
S1, the potential field planting region recognition of newborn patch: intermediate-resolution remote sensing satellite image is utilized, according to the side of visual interpretation
Method, and fully consider the phase difference of Spartina alterniflora Yu other vegetation, identification Spartina alterniflora and are tied by the range boundary of Hai Xianglu
It closes tide level data and obtains intertidal zone range, thus using the intertidal zone region of unidentified Spartina alterniflora's range boundary as newborn patch
Potential field planting region;
S2, the acquisition of multi-source multidate high spatial resolution satellite image: many years in the above-mentioned potential field planting area of acquisition covering are same
One growth month in busy season GF-1, GF-2, SPOT-6 or WorldView-2 image, guarantees that the annual growth busy season, there are one
Scape image;
S3, the pretreatment of multi-source multidate high spatial resolution satellite image: to multidate multi-source high score satellite image used
Radiation calibration is carried out, atmospheric correction, image co-registration, is registrated, inlayed and cut pretreatment;
The super-resolution rebuilding of satellite image after S4, processing: by the high partial image of different resolution, FSRCNN model is used
Deep learning method carries out super-resolution rebuilding;
The identification of S5, Spartina alterniflora's new life patch: using the fractional spins of pixel gray value, an image is carried out
Pre-segmentation obtains preliminary cutting unit (since it is not last segmentation result, can be called sub-primitive);Then by spectrum,
Shape feature coupling carries out figure spot and merges the identifying purpose for completing overall segmentation to reach Spartina alterniflora's patch;
The extraction and measurement of S6, Spartina alterniflora's new life patch: the result figure after segmentation is added on the image of 0.5m,
Spartina alterniflora's new life patch in first step is extracted in interpretation by visual observation;Secondly at ARCGIS (the GIS-Geographic Information System groupware)
Neighbours' coordinate of the middle minimum outsourcing rectangle for calculating each figure spot;Again by attribute list summarize to obtain each figure spot north and south,
East and West direction distance;The area and perimeter of each figure spot are obtained finally by the computational geometry of attribute list;
S7, field measurement and calibration of the output results: identified patch is chosen from extracting in result figure, according to plaque area
It is descending fall into 5 types, every a kind of patch chosen 10 fields and be easier to reach, utilize GPS positioning and visual interpretation high definition
The method of image, goes to field, positions to each patch, and measure the area, perimeter, the four corners of the world of each patch to away from
From, shape, height, strain number, coverage, phenological period (trophophase, bud stage, florescence, productive phase, dormant period), viability
The attributes such as (strong, in, weak) and ambient enviroment.It is X with field measurement value, the patch measured value that satellite image extracts in result figure is
Y does linear fit, obtains fit equation:
Y=a*X+b
The satellite measurement of all patches is corrected with fitting formula again;
S8, Spartina alterniflora's new life patch expansion dynamic measure: being executed using the multi-platform high spatial resolution images of multidate
First five step obtains Spartina alterniflora's new life patch expansion dynamic and measures, and more periods that first five step is obtained are accurately mutual
Flower spartina new life patch measured value is by analysis methods such as statistics, to the diffusion mould of Spartina alterniflora's new life patch of EARLY RECOGNITION
Formula carries out timely dynamic monitoring.
Preferably, the pre-processing image data process mentioned in the S3 are as follows:
1), radiation calibration: the Apply Gain and provided according to ENVI (complete Remote Sensing Image Processing)
Offset tool carries out radiation calibration to multispectral data and full-colored data;
2), atmospheric correction: using the FLAASH atmospheric correction mould in ENVI (complete Remote Sensing Image Processing) software
Block carries out atmospheric correction to high score multispectral image;
3), ortho-rectification: image space distortion is corrected caused by being risen and fallen to image because of elevation using dem data,
Generate multicenter projection plane ortho-rectification image;
4), image co-registration: using Gram-Schmidt method respectively to multi-source multidate high spatial resolution satellite image into
Row fusion, obtains fused multidate multi-source high score satellite image;
5), image registration: by the existing corrected remote sensing images in region as reference picture, 20-25 ground is chosen
Face control point is corrected using quadratic polynomial resampling technique, and overall error controls within 0.5 pixel;
6), image cropping: by the potential field planting region vector data of existing newborn patch, oneself corrected image is carried out
It cuts, image after being pre-processed.
Preferably, the FSRCNN model mentioned in the S4 is divided into five layers, respectively feature extraction layer, shrinkage layer, mapping
Layer, extension layer and reconstruction of layer, first four layers are process of convolution, for extracting low-dimensional data from high dimensional data;And last part
Reconstruction of layer is deconvolution processing, for low-dimensional data to be mapped to higher-dimension output, indicates convolution with Conv (fi, ni, ci) respectively
Processing, DeConv (fi, ni, ci) indicate deconvolution processing, and wherein variable fi, ni, ci respectively represent filter size, filtering
Device number and number of channels.
Preferably, use PRe LU function as the activation primitive after each convolutional layer, PRe in FSRCNN network model
LU function is shown below:
The penalty values between reconstruction image and target image are reduced using back-propagation algorithm, it can after the completion of training
The weighting parameter of network is obtained, loss function is used as using mean square deviation (MSE);It is quantitative by calculating Y-PSNR (PSNR)
Image compares the quality of quality with original image after assessment evaluation piece image is rebuild, and Y-PSNR is higher, and distortion is got over after compression
It is small.Y-PSNR is often defined simply by mean square error (MSE), two m × n monochrome image I and K, if one
It is approximate for the noise of another, then their mean square error is defined as:
Y-PSNR mathematical formulae is as follows:
Wherein, MAXIIt is the greatest measure for indicating picture point color.
Preferably, the identification and extraction and measurement for the Spartina alterniflora's new life patch being previously mentioned in the S5 and S6, this hair
The bright features such as the reflected color of high-definition picture, shape that have also combined merge, so that final segmentation result is obtained,
Two above step passes through watershed algorithm respectively and quick figure spot folding is realized, particular technique step are as follows:
I, the acquisition of sub-primitive: reading color image first converts it into gray level image, then passes through watershed segmentation
Algorithm carries out primary just segmentation to image, obtains sub-primitive;
II, the cost criterion that figure spot merges: using cost criterion function is merged, the function is heterogeneous by the spectrum of merging figure spot
Property parameter and shape heterogeneity parameter two parts constitute:
F=w × hcolor+(1-w)×hshape
Wherein w is the weight distributed for spectrum, shape heterogeneity, and section is [0,1], hcolorTone weight and hshapeShape
Shape weight, section are (0,1);
Spectrum heterogeneity is father's figure spot standard deviation and the difference that merges the sum of preceding two subgraphs spot standard deviation after merging, and by area
It is weighted:
Wherein c is wave band sum, wcIt is then the weight (default is 1.0) that each wave band is distributed by user, is calculated with this
The SPECTRAL DIVERSITY that patch merges in multi-band image.
Shape is heterogeneous again heterogeneous by compact degree and the weighting of smoothness heterogeneity two parts is constituted:
hshape=wcmpct×hcmpct+(1-wcmpct)×hsmooth
Compact degree difference is then calculated by following formula:
Smoothness difference is calculated by following formula:
In above formula, l is object perimeter, and n is object pixel number, and b is the perimeter of the boundary rectangle of object.Tightly
Weight shared by cause degree, smoothness is set by the user adjustment.
The obvious inefficiency of standard deviation of father's figure spot is computed repeatedly in merging process, thus using sample after following merging
The quick calculation of this standard difference:
M in above formula1、m2The respectively mean value of two patches;
III, the quick merging of figure spot uses on the basis of initial small figure spot has been determined, has merged the conditions such as cost function
" scale parameter " controls entire merging process, and the essence of the parameter is to merge cost threshold value.
Compared with prior art, the present invention provides the dynamic monitoring method of Spartina alterniflora's new life patch, having following has
Beneficial effect:
(1) compared with traditional field sampling method, the invention saves time and labor, can mutually spend rice on a large scale to seashore wetland
The carry out overview of careless new life patch;Have the advantages that untouchable, avoids due to environment complexity, people cannot be introduced into and have no way of
The case where solution Spartina alterniflora's initial stage spreads;By multi-source multidate satellite data to the precise measurement of Spartina alterniflora's new life patch,
The case where Spartina alterniflora spreads initial stage can be accurately captured in short time, improve the prevention and control efficiency to Spartina alterniflora;
(2) compared with medium resolution satellite remote sensing technology, the present invention uses high resolution ratio satellite remote-sensing technology, spatial discrimination
Rate and revisiting period greatly improve, and the extraction for tiny patch provides possibility;
(3) present invention employs deep learning super-resolution rebuildings merges multi-platform, different resolution data, very
The error and uncertainty of the result as caused by the resolution ratio difference of data are reduced in big degree, precision is higher, Neng Gouzhun
The slight change of the true tiny patch of reflection, is more suitable for the early monitoring of Spartina alterniflora quickly spread.
Detailed description of the invention
Fig. 1 is the overall flow figure of the dynamic monitoring method of Spartina alterniflora new life patch proposed by the present invention;
Fig. 2 is the multidate multi-source high score satellite mapping of the dynamic monitoring method of Spartina alterniflora new life patch proposed by the present invention
As pretreatment process figure;
Fig. 3 is the FSRCNN model structure of the dynamic monitoring method of Spartina alterniflora new life patch proposed by the present invention;
Fig. 4 is more rulers based on watershed segmentation of the dynamic monitoring method of Spartina alterniflora new life patch proposed by the present invention
Spend dividing method flow chart;
Fig. 5 is field measurement and the calibration of the output results of the dynamic monitoring method of Spartina alterniflora new life patch proposed by the present invention
Flow chart;
Fig. 6 is Spartina alterniflora's new life patch expansion of the dynamic monitoring method of Spartina alterniflora new life patch proposed by the present invention
The structural schematic diagram of mode.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.
In the description of the present invention, it is to be understood that, term " on ", "lower", "front", "rear", "left", "right", "top",
The orientation or positional relationship of the instructions such as "bottom", "inner", "outside" is to be based on the orientation or positional relationship shown in the drawings, merely to just
In description the present invention and simplify description, rather than the device or element of indication or suggestion meaning must have a particular orientation, with
Specific orientation construction and operation, therefore be not considered as limiting the invention.
Embodiment 1:
Please refer to Fig. 1;
The dynamic monitoring method of Spartina alterniflora's new life patch, comprising the following steps:
S1, the potential field planting region recognition of newborn patch: intermediate-resolution remote sensing satellite image is utilized, according to the side of visual interpretation
Method, and fully consider the phase difference of Spartina alterniflora Yu other vegetation, identification Spartina alterniflora and are tied by the range boundary of Hai Xianglu
It closes tide level data and obtains intertidal zone range, thus using the intertidal zone region of unidentified Spartina alterniflora's range boundary as newborn patch
Potential field planting region;
S2, the acquisition of multi-source multidate high spatial resolution satellite image: many years in the above-mentioned potential field planting area of acquisition covering are same
One growth month in busy season GF-1, GF-2, SPOT-6 or WorldView-2 image, guarantees that the annual growth busy season, there are one
Scape image;
S3, the pretreatment of multi-source multidate high spatial resolution satellite image: to multidate multi-source high score satellite image used
Radiation calibration is carried out, atmospheric correction, image co-registration, is registrated, inlayed and cut pretreatment;
The super-resolution rebuilding of satellite image after S4, processing: by the high partial image of different resolution, FSRCNN model is used
Deep learning method carries out super-resolution rebuilding;
The identification of S5, Spartina alterniflora's new life patch: using the fractional spins of pixel gray value, an image is carried out
Pre-segmentation obtains preliminary cutting unit (since it is not last segmentation result, can be called sub-primitive);Then by spectrum,
Shape feature coupling carries out figure spot and merges the identifying purpose for completing overall segmentation to reach Spartina alterniflora's patch;
The extraction and measurement of S6, Spartina alterniflora's new life patch: the result figure after segmentation is added on the image of 0.5m,
Spartina alterniflora's new life patch in first step is extracted in interpretation by visual observation;Secondly at ARCGIS (the GIS-Geographic Information System groupware)
Neighbours' coordinate of the middle minimum outsourcing rectangle for calculating each figure spot;Again by attribute list summarize to obtain each figure spot north and south,
East and West direction distance;The area and perimeter of each figure spot are obtained finally by the computational geometry of attribute list;
S7, field measurement and calibration of the output results: identified patch is chosen from extracting in result figure, according to plaque area
It is descending fall into 5 types, every a kind of patch chosen 10 fields and be easier to reach, utilize GPS positioning and visual interpretation high definition
The method of image, goes to field, positions to each patch, and measure the area, perimeter, the four corners of the world of each patch to away from
From, shape, height, strain number, coverage, phenological period (trophophase, bud stage, florescence, productive phase, dormant period), viability
The attributes such as (strong, in, weak) and ambient enviroment.It is X with field measurement value, the patch measured value that satellite image extracts in result figure is
Y does linear fit, obtains fit equation:
Y=a*X+b
The satellite measurement of all patches is corrected with fitting formula again;
S8, Spartina alterniflora's new life patch expansion dynamic measure: being executed using the multi-platform high spatial resolution images of multidate
First five step obtains Spartina alterniflora's new life patch expansion dynamic and measures, and more periods that first five step is obtained are accurately mutual
Flower spartina new life patch measured value is by analysis methods such as statistics, to the diffusion mould of Spartina alterniflora's new life patch of EARLY RECOGNITION
Formula carries out timely dynamic monitoring.
Compared with traditional field sampling method, the invention saves time and labor, can be to a wide range of Spartina alterniflora of seashore wetland
The carry out overview of newborn patch;Have the advantages that untouchable, avoids due to environment complexity, people cannot be introduced into and have no way of understanding
The case where Spartina alterniflora's initial stage spreads;By multi-source multidate satellite data to the precise measurement of Spartina alterniflora's new life patch, energy
The case where Spartina alterniflora spreads initial stage is accurately captured in enough short time, improves the prevention and control efficiency to Spartina alterniflora;With middle resolution
Rate satellite remote sensing technology is compared, and the present invention uses high resolution ratio satellite remote-sensing technology, and spatial resolution and revisiting period mention significantly
Height, the extraction for tiny patch provide possibility;In addition, present invention employs deep learning super-resolution rebuilding to it is multi-platform,
Different resolution data are merged, largely reduce as the resolution ratio of data it is different caused by result error and
Uncertainty, precision is higher, can accurately reflect the slight change of tiny patch, be more suitable for the Spartina alterniflora quickly spread
Early monitoring.
Embodiment 2:
Referring to Fig. 2, based on embodiment 1, different place is again;
The pre-processing image data process mentioned in S3 are as follows:
1), radiation calibration: the Apply Gain and provided according to ENVI (complete Remote Sensing Image Processing)
Offset tool carries out radiation calibration to multispectral data and full-colored data;
2), atmospheric correction: using the FLAASH atmospheric correction mould in ENVI (complete Remote Sensing Image Processing) software
Block carries out atmospheric correction to high score multispectral image;
3), ortho-rectification: image space distortion is corrected caused by being risen and fallen to image because of elevation using dem data,
Generate multicenter projection plane ortho-rectification image;
4), image co-registration: using Gram-Schmidt method respectively to multi-source multidate high spatial resolution satellite image into
Row fusion, obtains fused multidate multi-source high score satellite image;
5), image registration: by the existing corrected remote sensing images in region as reference picture, 20-25 ground is chosen
Face control point is corrected using quadratic polynomial resampling technique, and overall error controls within 0.5 pixel;
6), image cropping: by the potential field planting region vector data of existing newborn patch, oneself corrected image is carried out
It cuts, image after being pre-processed.
Using intermediate-resolution remote sensing satellite image (such as-No. 2 Landsat, sentry satellites), according to the method for visual interpretation,
And it is (main to choose 10 below the moon to fully consider that Spartina alterniflora combines with the phase difference of other local vegetation and multi-temporal image
Ten days image, local vegetation reed has entered the phase of decaying at this time, and Spartina alterniflora is in the maturity period), identify Spartina alterniflora in 2013
Intertidal zone range is obtained by the range boundary of Hai Xianglu, and in conjunction with tide level data, thus by unidentified Spartina alterniflora's range boundary
Intertidal zone region as the newborn potential field planting region of patch;It is new to choose costal wetland in Yellow River Delta one typical Spartina alterniflora
The raw potential field planting region of patch, many years (5 years such as nearly) the same growth month in busy season in the above-mentioned potential field planting area of acquisition covering is (such as
The 7-9 month) GF-1 (2m resolution ratio), GF-2 (1m resolution ratio), SPOT-6 (1.5m resolution ratio) image, guarantee that annual growth is prosperous
Season, there are a scape images.
Data are as shown in table 1.
Radiation calibration subsequently is carried out to multidate multi-source high score satellite image used, atmospheric correction, image co-registration, is matched
Standard inlays and cuts pretreatment, with reference to above-mentioned pre-processing image data process, image after being pre-processed, as shown in Figure 2
Embodiment 3:
Referring to Fig. 3, based on embodiment 1 or 2 again different place be;
The FSRCNN model mentioned in S4 is divided into five layers, respectively feature extraction layer, shrinkage layer, mapping layer, extension layer and
Reconstruction of layer, first four layers are process of convolution, for extracting low-dimensional data from high dimensional data;And last part reconstruction of layer is warp
Product processing indicates process of convolution, DeConv with Conv (fi, ni, ci) respectively for low-dimensional data to be mapped to higher-dimension output
(fi, ni, ci) indicates deconvolution processing, and wherein variable fi, ni, ci respectively represent filter size, number of filter and channel
Quantity.
This step is by the high partial image (table 1) of different resolution, unified to 0.5m points by the method for super-resolution rebuilding
Under resolution.The present invention carries out super-resolution rebuilding using FSRCNN deep learning method.Training obtains 2 times, 3 times and 4 times first
FSRCNN model.Secondly it carries out FSRCNN test: selecting 5 image data set of Set as test data, we have selected model
After parameter to 5 image of Set under the conditions of 2 times, 3 times and 4 times amplifications, carry out being respectively adopted bicubic interpolation respectively, SRCNN,
Tri- kinds of algorithms of FSRCNN are rebuild, and are provided and are objectively evaluated index, while by partial enlargement more intuitively to compare reconstruction matter
Amount, finally provides image after the image enhancement for meeting human eye vision.From objectively evaluating standard and from the point of view of subjective vision, FSRCNN
Model all has significant advantage on reconstruction quality.The GF-1 of 2m is finally realized into 4 times of super-resolution rebuildings, image resolution
Rate amplifies 4 times;The GF-2 image resolution ratio of same 1m amplifies 2 times;The SPOT-6 image resolution ratio of 1.5m amplifies 3 times.All receipts
Under the different resolution high score image unification to 0.5m resolution ratio collected.
Embodiment 4:
Referring to Fig. 3, based on embodiment 1-3 wherein any one again different place be;
Use PRe LU function as the activation primitive after each convolutional layer, PRe LU function in FSRCNN network model
It is shown below:
The penalty values between reconstruction image and target image are reduced using back-propagation algorithm, it can after the completion of training
The weighting parameter of network is obtained, loss function is used as using mean square deviation (MSE);It is quantitative by calculating Y-PSNR (PSNR)
Image compares the quality of quality with original image after assessment evaluation piece image is rebuild, and Y-PSNR is higher, and distortion is got over after compression
It is small.Y-PSNR is often defined simply by mean square error (MSE), two m × n monochrome image I and K, if one
It is approximate for the noise of another, then their mean square error is defined as:
Y-PSNR mathematical formulae is as follows:
Wherein, MAXIIt is the greatest measure for indicating picture point color.
Carry out FRCNN model training first: the present invention uses 91-image and General-100 image data set
As training image, select 5 image data set of Set as test image.These images are subjected to data enhancing and are used as high-resolution
Rate image, and carry out 2,3 times of size or 4 times of scaling processings respectively to high resolution gray image, these scalings are schemed herein
As being used as low resolution image.The method of training convolutional neural networks of the present invention is to be instructed by Caffe to convolutional neural networks
Practice, needs for 2 times, 3 times and 4 times amplification conditions respectively by low-resolution/high-resolution image according to 102/192, 72/192With
62/212Carry out piecemeal.Layer Initialize installation that deconvolute is Gaussian type and 0.001 standard deviation value.Starting training convolutional mind
It when through network, is trained using only 91-image data set, the initial learning rate of convolutional layer is set as 0.001 at this time, goes
The learning rate of convolutional layer is 0.0001.In network close to after saturation, made using 91-image and General-100 enhancing image
For training data, need for the learning rate of convolutional layer to be changed to 0.0005 at this time, warp lamination is set as 0.00005.In order to examine
The property of FSRCNN structure is tested, we devise one group of control experiment, in the case where fixed amplification factor is 3, respectively to low
Image in different resolution intrinsic dimensionality d, shrinkage filter number s, mapping these three sensitive variables of depth m carry out different values, have
Body choose d=48,56;S=12,16 and m=2,3,4, therefore we have carried out the experiment of 12 various combinations in total.Finally
It was found that performance and parameter reach optimum balance in FSRCNN (56,12,4), PSNR obtains one of highest result.We select
FSRCNN (56,12,4) is used as default network.Above-mentioned training process is carried out in the case where amplification factor is 3, i.e. 3 times of moulds
The training result of type oneself through obtaining, as long as now fine tuning deconvolution network x times of amplification factor is trained, obtain x times of model
Training result.
Secondly it carries out FSRCNN model measurement: selecting 5 image data set of Set as test data, we have selected model
After parameter to 5 image of Set under the conditions of x times of amplification, carry out being respectively adopted bicubic interpolation respectively, tri- kinds of SRCNN, FSRCNN
Algorithm is rebuild, and is provided and is objectively evaluated index, while from vision intuitively comparing reconstruction quality, finally being provided and being met human eye vision
Image enhancement after image.From objectively evaluating standard and from the point of view of subjective vision, FSRCNN model all has on reconstruction quality
Significant advantage.
Embodiment 5:
Fig. 4-6 is please referred to, based on any one in embodiment 1-4 and different place is;
The identification and extraction of the Spartina alterniflora's new life patch being previously mentioned in S5 and S6 and measurement, the present invention have also combined height
The features such as the reflected color of image in different resolution, shape merge, so that final segmentation result is obtained, two above step
It is realized respectively by watershed algorithm and quick figure spot folding, particular technique step are as follows:
I, the acquisition of sub-primitive: reading color image first converts it into gray level image, then passes through watershed segmentation
Algorithm carries out primary just segmentation to image, obtains sub-primitive;
II, the cost criterion that figure spot merges: using cost criterion function is merged, the function is heterogeneous by the spectrum of merging figure spot
Property parameter and shape heterogeneity parameter two parts constitute:
F=w × hcolor+(1-w)×hshape
Wherein w is the weight distributed for spectrum, shape heterogeneity, and section is [0,1], hcolorTone weight and hshapeShape
Shape weight, section are (0,1);
Spectrum heterogeneity is father's figure spot standard deviation and the difference that merges the sum of preceding two subgraphs spot standard deviation after merging, and by area
It is weighted:
Wherein c is wave band sum, wcIt is then the weight (default is 1.0) that each wave band is distributed by user, is calculated with this
The SPECTRAL DIVERSITY that patch merges in multi-band image.
Shape is heterogeneous again heterogeneous by compact degree and the weighting of smoothness heterogeneity two parts is constituted:
hshape=wcmpct×hcmpct+(1-wcmpct)×hsmooth
Compact degree difference is then calculated by following formula:
Smoothness difference is calculated by following formula:
In above formula, l is object perimeter, and n is object pixel number, and b is the perimeter of the boundary rectangle of object.Tightly
Weight shared by cause degree, smoothness is set by the user adjustment.
The obvious inefficiency of standard deviation of father's figure spot is computed repeatedly in merging process, thus using sample after following merging
The quick calculation of this standard difference:
M in above formula1、m2The respectively mean value of two patches;
III, the quick merging of figure spot uses on the basis of initial small figure spot has been determined, has merged the conditions such as cost function
" scale parameter " controls entire merging process, and the essence of the parameter is to merge cost threshold value.
If all figure spots and the merging cost of its neighbour to be combined are all larger than the scale parameter in certain merging process
Square, then entire merging process terminates, complete image segmentation;Otherwise, after the merging of all figure spots, need to recalculate new figure
The various features and topological relation of spot, and merged again (see Fig. 3), until meeting condition, algorithm can be terminated, and be completed
Segmentation.
It extracts and the process of measurement is completed in ArcMap10.2.The result figure after segmentation is added to the figure of 0.5m first
As upper, Spartina alterniflora's new life patch in first step is extracted in interpretation by visual observation;Secondly each figure is calculated in ARCGIS10.2
Neighbours' coordinate (i.e. xmax, xmin, ymin, ymax) of the minimum outsourcing rectangle of spot;Summarize to obtain again by attribute list each
The north and south of figure spot, East and West direction distance (north and south distance=ymax-ymin, thing distance=xmax-xmin);Finally by attribute list
Computational geometry obtain the area and perimeter of each figure spot.
Embodiment 6:
Based on any one in embodiment 1-5 and different place is:
The present invention executes first five step using the multi-platform high spatial resolution images of multidate, obtains Spartina alterniflora new life
Patch expansion dynamic measures.The first step is extracted Spartina alterniflora's distribution map in research area using Landsat, is obtained in conjunction with tidal data
To the potential field planting region of newborn patch.Second step, (5 years such as nearly) Spartina alterniflora grows the more of busy season for many years in collection research area
Platform high spatial resolution images (at least annual there are a scape image), and being merged, the pretreatment such as registration obtain being used for the
Three steps use pretreated image.Third step obtains x times of model by the training of FSRCNN prototype network, can be pre- by second step
Treated more period images are by the method for super-resolution rebuilding, using under suitable x times of Unified Model to 0.5m resolution ratio.
4th step embodies the essential characteristics such as color, shape by fractional spins combination high score image, third step is obtained
More period images of 0.5m are split, and obtain Spartina alterniflora's patch, and tone weight 0.7 therein, shape 0.3 is proper,
Shape weight is excessive often to will cause segmentation result and visual segments effect difference is larger, weight one shared by compact degree, smoothness
As take 0.5 appropriate, scale parameter is generally set to 50.Remaining work will be completed in ArcGis, mutual to what is identified
Flower spartina patch result, which mentions, extracts while obtaining relevant attribute value, including area, perimeter, north-south, East and West direction distance
Deng.The patch measured value Y that 5th step, field measurement value X and preceding four step are extracted does linear fit, obtains fitting formula Y=a*X+
B corrects the satellite measurement of all time limits using the fitting formula.Final step has obtained more periods by first five step
Accurate Spartina alterniflora's new life patch measured value, by these measured values by analysis methods such as statistics, to the mutual of EARLY RECOGNITION
The dispersal pattern of flower spartina new life patch carries out timely dynamic monitoring, plays a significant role for the efficient prevention and control of Spartina alterniflora
The foregoing is only a preferred embodiment of the present invention, but scope of protection of the present invention is not limited thereto,
Anyone skilled in the art in the technical scope disclosed by the present invention, according to the technique and scheme of the present invention and its
Inventive concept is subject to equivalent substitution or change, should be covered by the protection scope of the present invention.
Claims (5)
1. the dynamic monitoring method of Spartina alterniflora's new life patch, which comprises the following steps:
S1, the newborn potential field planting region recognition of patch: utilizing intermediate-resolution remote sensing satellite image, according to the method for visual interpretation,
And fully considering the phase difference of Spartina alterniflora Yu other vegetation, identification Spartina alterniflora and is combined by the range boundary of Hai Xianglu
Tide level data obtains intertidal zone range, to dive the intertidal zone region of unidentified Spartina alterniflora's range boundary as newborn patch
In field planting region;
S2, the acquisition of multi-source multidate high spatial resolution satellite image: many years in the above-mentioned potential field planting area of acquisition covering are the same as all one's life
Long month in busy season GF-1, GF-2, SPOT-6 or WorldView-2 image guarantees that the annual growth busy season, there are a scape figures
Picture;
S3, the pretreatment of multi-source multidate high spatial resolution satellite image: multidate multi-source high score satellite image used is carried out
Radiation calibration atmospheric correction, image co-registration, is registrated, inlays and cut pretreatment;
The super-resolution rebuilding of satellite image after S4, processing: by the high partial image of different resolution, FSRCNN model depth is used
Learning method carries out super-resolution rebuilding;
The identification of S5, Spartina alterniflora's new life patch: it using the fractional spins of pixel gray value, carries out an image and divides in advance
It cuts, obtains preliminary cutting unit;Then spectrum, shape feature are coupled, figure spot is carried out and merges completion overall segmentation to reach
The identifying purpose of Spartina alterniflora's patch;
The extraction and measurement of S6, Spartina alterniflora's new life patch: the result figure after segmentation is added on the image of 0.5m, mesh is passed through
Spartina alterniflora's new life patch in first step is extracted depending on interpretation;Secondly the minimum outsourcing rectangle of each figure spot is calculated in ARCGIS
Neighbours' coordinate;Summarize the north and south, the East and West direction distance that obtain each figure spot again by attribute list;Finally by the meter of attribute list
It calculates geometry and obtains the area and perimeter of each figure spot;
S7, field measurement and calibration of the output results: identified patch is chosen from extracting in result figure, according to plaque area by big
To it is small fall into 5 types, every a kind of patch chosen 10 fields and be easier to reach, utilize GPS positioning and visual interpretation high-definition image
Method, go to field, each patch positioned, and measure the area, perimeter, the four corners of the world of each patch to distance,
The attributes such as shape, height, strain number, coverage, phenological period, viability and ambient enviroment.It is X, satellite image with field measurement value
The patch measured value extracted in result figure is Y, does linear fit, obtains fit equation:
Y=a*X+b
The satellite measurement of all patches is corrected with fitting formula again;
S8, Spartina alterniflora's new life patch expansion dynamic measure: executing first five using the multi-platform high spatial resolution images of multidate
A step obtains Spartina alterniflora's new life patch expansion dynamic and measures, and more periods that first five step is obtained accurately mutually spend rice
Careless new life patch measured value by analysis methods such as statistics, to the dispersal pattern of Spartina alterniflora's new life patch of EARLY RECOGNITION into
The timely dynamic monitoring of row.
2. the dynamic monitoring method of Spartina alterniflora new life patch according to claim 1, it is characterised in that: mentioned in the S3
The pre-processing image data process arrived are as follows:
1), radiation calibration: the Apply Gain and Offset tool provided according to ENVI is to multispectral data and full-colored data
Carry out radiation calibration;
2) atmosphere school, atmospheric correction: is carried out to high score multispectral image using the FLAASH atmospheric correction module in ENVI software
Just;
3), ortho-rectification: image space distortion is corrected caused by being risen and fallen to image because of elevation using dem data, is generated
Multicenter projection plane ortho-rectification image;
4), image co-registration: multi-source multidate high spatial resolution satellite image is melted respectively using Gram-Schmidt method
It closes, obtains fused multidate multi-source high score satellite image;
5), image registration: by the existing corrected remote sensing images in region as reference picture, 20-25 ground control is chosen
It is processed, it is corrected using quadratic polynomial resampling technique, overall error controls within 0.5 pixel;
6), image cropping: by the potential field planting region vector data of existing newborn patch, oneself corrected image is cut out
It cuts, image after being pre-processed.
3. the dynamic monitoring method of Spartina alterniflora new life patch according to claim 1, it is characterised in that: mentioned in the S4
To FSRCNN model be divided into five layers, respectively feature extraction layer, shrinkage layer, mapping layer, extension layer and reconstruction of layer, first four layers are
Process of convolution, for extracting low-dimensional data from high dimensional data;And last part reconstruction of layer is deconvolution processing, being used for will be low
Dimension data is mapped to higher-dimension output, indicates process of convolution with Conv (filter size, number of filter, number of channels) respectively,
DeConv (filter size, number of filter, number of channels) indicates deconvolution processing.
4. the dynamic monitoring method of Spartina alterniflora new life patch according to claim 1 or 3, it is characterised in that: FSRCNN
Use PRe LU function as the activation primitive after each convolutional layer in network model, PRe LU function is shown below:
The penalty values between reconstruction image and target image are reduced using back-propagation algorithm, it is available after the completion of training
The weighting parameter of network, using mean square deviation as loss function;By calculating Y-PSNR, evaluation piece image is quantitatively evaluated
Image compares the quality of quality with original image after reconstruction, and Y-PSNR is higher, is distorted after compression smaller.Y-PSNR Chang Jian
It singly is defined by mean square error, two m × n monochrome image I and K, if a noise for another is approximate, that
Their mean square error is defined as:
Y-PSNR mathematical formulae is as follows:
Wherein, MAXIIt is the greatest measure for indicating picture point color.
5. the dynamic monitoring method of Spartina alterniflora new life patch according to claim 1, it is characterised in that: the S5 and S6
In the identification of Spartina alterniflora's new life patch that is previously mentioned and extraction and measurement, the present invention have also combined high-definition picture embodiment
The features such as color, shape out merge, to obtain final segmentation result, two above step passes through watershed respectively
Algorithm and the realization of quick figure spot folding, particular technique step are as follows:
I, the acquisition of sub-primitive: reading color image first converts it into gray level image, then passes through fractional spins
Primary just segmentation is carried out to image, obtains sub-primitive;
II, the cost criterion that figure spot merges: using cost criterion function is merged, which is joined by the spectrum heterogeneity of merging figure spot
Amount and shape heterogeneity parameter two parts are constituted:
F=w × hcolor+(1-w)×hshape
Wherein w is the weight distributed for spectrum, shape heterogeneity, and section is [0,1], hcolorTone weight and hshapeShape power
Weight, section is (0,1);
Spectrum heterogeneity is father's figure spot standard deviation and the difference that merges the sum of preceding two subgraphs spot standard deviation after merging, and presses area progress
Weighting:
Wherein c is wave band sum, wcIt is then the weight that each wave band is distributed by user, patch in multi-band image is calculated with this and is closed
And SPECTRAL DIVERSITY.
Shape is heterogeneous again heterogeneous by compact degree and the weighting of smoothness heterogeneity two parts is constituted:
hshape=wcmpct×hcmpct+(1-wcmpct)×hsmooth
Compact degree difference is then calculated by following formula:
Smoothness difference is calculated by following formula:
In above formula, l is object perimeter, and n is object pixel number, and b is the perimeter of the boundary rectangle of object.It is compact
Weight shared by degree, smoothness is set by the user adjustment.
The obvious inefficiency of standard deviation of father's figure spot is computed repeatedly in merging process, thus using sample mark after following merging
The quasi- quick calculation of difference:
M in above formula1、m2The respectively mean value of two patches;
III, the quick merging of figure spot, on the basis of initial small figure spot has been determined, has merged the conditions such as cost function, using " scale
Parameter " controls entire merging process, and the essence of the parameter is to merge cost threshold value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910270990.3A CN110032963B (en) | 2019-04-04 | 2019-04-04 | Dynamic monitoring method for spartina alterniflora new-born plaques |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910270990.3A CN110032963B (en) | 2019-04-04 | 2019-04-04 | Dynamic monitoring method for spartina alterniflora new-born plaques |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110032963A true CN110032963A (en) | 2019-07-19 |
CN110032963B CN110032963B (en) | 2020-10-30 |
Family
ID=67237546
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910270990.3A Expired - Fee Related CN110032963B (en) | 2019-04-04 | 2019-04-04 | Dynamic monitoring method for spartina alterniflora new-born plaques |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110032963B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113128525A (en) * | 2021-05-19 | 2021-07-16 | 内蒙古农业大学 | Control device and method for desert grassland population patch identification |
CN113408460A (en) * | 2021-06-30 | 2021-09-17 | 中国科学院东北地理与农业生态研究所 | Method for detecting spartina alterniflora distribution based on remote sensing big data and cloud platform |
CN113436071A (en) * | 2021-06-21 | 2021-09-24 | 武汉光谷信息技术股份有限公司 | Multi-source satellite remote sensing image mosaic method, system and storage medium |
CN113487493A (en) * | 2021-06-02 | 2021-10-08 | 厦门大学 | SAR image automatic colorization method based on GANILA |
CN114220002A (en) * | 2021-11-26 | 2022-03-22 | 通辽市气象台(通辽市气候生态环境监测中心) | Method and system for monitoring invasion of foreign plants based on convolutional neural network |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100067799A1 (en) * | 2008-09-17 | 2010-03-18 | Microsoft Corporation | Globally invariant radon feature transforms for texture classification |
CN103528570A (en) * | 2013-10-23 | 2014-01-22 | 环境保护部卫星环境应用中心 | Spatial distribution acquiring method and system for spartina alterniflora in northern sea area of Guangxi |
US20180373932A1 (en) * | 2016-12-30 | 2018-12-27 | International Business Machines Corporation | Method and system for crop recognition and boundary delineation |
-
2019
- 2019-04-04 CN CN201910270990.3A patent/CN110032963B/en not_active Expired - Fee Related
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100067799A1 (en) * | 2008-09-17 | 2010-03-18 | Microsoft Corporation | Globally invariant radon feature transforms for texture classification |
CN103528570A (en) * | 2013-10-23 | 2014-01-22 | 环境保护部卫星环境应用中心 | Spatial distribution acquiring method and system for spartina alterniflora in northern sea area of Guangxi |
US20180373932A1 (en) * | 2016-12-30 | 2018-12-27 | International Business Machines Corporation | Method and system for crop recognition and boundary delineation |
Non-Patent Citations (2)
Title |
---|
ZHENMING GE 等: "《A process-based grid model for the simulation of range expansion of Spartina alterniflora on the coastal saltmarshes in the Yangtze Estuary》", 《ECOLOGICAL ENGINEERING》 * |
刘会玉 等: "《基于个体的空间显性模型和遥感技术模拟入侵植物扩张机制》", 《生态学报》 * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113128525A (en) * | 2021-05-19 | 2021-07-16 | 内蒙古农业大学 | Control device and method for desert grassland population patch identification |
CN113487493A (en) * | 2021-06-02 | 2021-10-08 | 厦门大学 | SAR image automatic colorization method based on GANILA |
CN113487493B (en) * | 2021-06-02 | 2023-08-18 | 厦门大学 | GANilla-based SAR image automatic colorization method |
CN113436071A (en) * | 2021-06-21 | 2021-09-24 | 武汉光谷信息技术股份有限公司 | Multi-source satellite remote sensing image mosaic method, system and storage medium |
CN113436071B (en) * | 2021-06-21 | 2022-04-29 | 武汉光谷信息技术股份有限公司 | Multi-source satellite remote sensing image mosaic method, system and storage medium |
CN113408460A (en) * | 2021-06-30 | 2021-09-17 | 中国科学院东北地理与农业生态研究所 | Method for detecting spartina alterniflora distribution based on remote sensing big data and cloud platform |
CN113408460B (en) * | 2021-06-30 | 2022-03-11 | 中国科学院东北地理与农业生态研究所 | Method for detecting spartina alterniflora distribution based on remote sensing big data and cloud platform |
CN114220002A (en) * | 2021-11-26 | 2022-03-22 | 通辽市气象台(通辽市气候生态环境监测中心) | Method and system for monitoring invasion of foreign plants based on convolutional neural network |
CN114220002B (en) * | 2021-11-26 | 2022-11-15 | 通辽市气象台(通辽市气候生态环境监测中心) | Method and system for monitoring invasion of foreign plants based on convolutional neural network |
Also Published As
Publication number | Publication date |
---|---|
CN110032963B (en) | 2020-10-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110032963A (en) | The dynamic monitoring method of Spartina alterniflora's new life patch | |
CN104915674B (en) | The method that Landsat8 and MODIS constructs high-spatial and temporal resolution data identification autumn grain crop | |
CN105004320B (en) | A kind of high score satellite data land table vegetation coverage inversion method and system | |
CN102565778B (en) | Relative radiometric correction method for automatically extracting pseudo-invariant features for remote sensing image | |
KR100678395B1 (en) | System and method for real time position correction of geometric information data using satellite and aerospace image | |
CN111832518B (en) | Space-time fusion-based TSA remote sensing image land utilization method | |
CN108932521B (en) | Deep learning-based crop classification method and system | |
CN113128401B (en) | Regional actual irrigation area monitoring method based on optical and radar remote sensing data | |
CN114926748A (en) | Soybean remote sensing identification method combining Sentinel-1/2 microwave and optical multispectral images | |
CN109977991A (en) | Forest resourceies acquisition method based on high definition satellite remote sensing | |
CN112733596A (en) | Forest resource change monitoring method based on medium and high spatial resolution remote sensing image fusion and application | |
CN107688777A (en) | A kind of urban green space extracting method for cooperateing with multi-source Remote Sensing Images | |
CN114821261A (en) | Image fusion algorithm | |
CN107688776A (en) | A kind of urban water-body extracting method | |
Stroppiana et al. | Estimating crop density from multi-spectral uav imagery in maize crop | |
CN117197668A (en) | Crop lodging level prediction method and system based on deep learning | |
CN112330672B (en) | Crop leaf area index inversion method based on PROSAIL model and canopy coverage optimization | |
CN117035066A (en) | Ground surface temperature downscaling method coupling geographic weighting and random forest | |
CN112733746B (en) | Collaborative classification method for fusing InSAR coherence and multispectral remote sensing | |
Nelson et al. | Combining moderate-resolution time-series RS data from SAR and optical sources for rice crop characterisation: examples from Bangladesh | |
Zhang et al. | Estimation of leaf area index over heterogeneous regions using the vegetation type information and PROSAIL model | |
Atzberger et al. | Geostatistical regularization of inverse models for the retrieval of vegetation biophysical variables | |
de Carvalho Jr et al. | Normalization of multi-temporal images using a new change detection method based on the spectral classifier | |
Kumar et al. | Accuracy assessment of land cover classification and change detection using remote sensing and GIS | |
Mercier et al. | Estimation and monitoring of bare soil/vegetation ratio with SPOT VEGETATION and HRVIR |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20201030 |