CN102222238A - Automatic simulation method of natural-color products of high-space-resolution remote sensing images - Google Patents
Automatic simulation method of natural-color products of high-space-resolution remote sensing images Download PDFInfo
- Publication number
- CN102222238A CN102222238A CN2011101273074A CN201110127307A CN102222238A CN 102222238 A CN102222238 A CN 102222238A CN 2011101273074 A CN2011101273074 A CN 2011101273074A CN 201110127307 A CN201110127307 A CN 201110127307A CN 102222238 A CN102222238 A CN 102222238A
- Authority
- CN
- China
- Prior art keywords
- spectrum
- remote sensing
- wave band
- blue
- simulation
- 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
Images
Landscapes
- Spectrometry And Color Measurement (AREA)
Abstract
The invention provides a simulation method of natural colors of high-resolution remote sensing images based on field spectroscopic data; especially for high-resolution remote sensing images without blue-light waveband (such as SPOT, IRS and the like), the difficulty of natural color synthesis of the images can be solved by simulation of the blue-light wave band. The simulation method comprises the steps of: firstly, preprocessing the field object wave spectrum data according to wavelength bandwidth setting of images to be simulated and a spectrum response function; then selecting control points of spectrum samples automatically according to the cluster result of ISODATA (iterative self-organizing data) algorithm, and selecting spectrum candidate samples by a spectral matching algorithm; next, learning and training by using a support vector machine to construct a nonlinear relation model among the blue-light wavebands to be simulated and known wavebands; and finally, realizing calculation of the blue-light wavebands according to the nonlinear relation model (SVM). The simulated natural-color image product is natural in color tone and real in color, and can be used in multiple fields, the automatic simulation of the missing blue-light wavebands of the high-space-resolution remote sensing images and the making of natural-color images are realized and the workload of manual image adjustment is greatly reduced.
Description
Technical field
The present invention relates to remote sensing image treatment technology, remote sensing image simulation and Enhancement Method.Specifically, relate to data pre-service, digital simulation and spectrum relation excavation, the present invention produces applicable to simulation of middle high-resolution remote sensing image data and true color image data product.
Background technology
Because blue wave band factor such as easier characteristic that is scattered and sensor design technology in radiation delivery, the blue band signal that sensor receives relatively a little less than, even the multispectral sensor of some high spatial resolutions (for example is not provided with blue channel; SPOT, IRS etc.), user's synthetic true color image in application process has just run into difficulty like this.For this reason, lot of domestic and foreign scholar and research institution simulate true color image and have carried out a series of research, and its essence is how to utilize the correlativity between the remote sensing image wave band to simulate blue wave band.For example, simulation SPOT Natural color audio and video products commonly used mainly contain following three kinds of schemes: (1) mean value method: this is the method that SPOTIMAGE company (2008) provides on its website, and blue wave band adopts green light band XS1 to replace; Green light band adopts the mean value of green, red and near-infrared band, i.e. (XS1+XS2+XS3)/3; Red wave band still adopts red wave band XS2 to represent; (2) weighted method: this is the method that image processing software ERDAS IMAGE provides for treatment S POT multispectral image specially, blue wave band still adopts green wave band XS1 to replace, green wave band then adopts (3XS1+XS3)/4 synthetic by green wave band and near-infrared band, and red wave band still adopts red wave band XS2 to represent; (3) uncertain parametric method: this method panchromatic wave-band (P) when the blue wave band of simulation also participates in computing, is blue wave band with 2P * XS1/ (XS1+XS2) wave band algorithm, is green wave band with 2P * XS2/ (XS1+XS2) wave band algorithm; Red wave band is represented with (aP+ (1-a) XS3).Huang Chi brave (1999) uses reversing neural network (BP-ANN) to the match that concerns between TM image NIR, R, G-band and the B wave band in addition, then this non-linear wave band relational model is applied to the blue wave band that simulates the SPOT image in the SPOT image, and then simulates SPOT Natural color image.Thomas Knudsen people such as (2005) adopts least-squares algorithm, simulates the linear relationship between NIR, R, G, NDVI and the B wave band, realizes the blue wave band simulation to the infrared aerophotograph of coloured silk, reaches the near nature color simulation of color infrared boat sheet.
In present wave band analogy method, application be empirical model, perhaps simulate relation between blue wave band and its all band from other remote sensing image.Model lacks image analysing computer, and the distortion of the true color color of simulation is serious, and tone is nature, and the aftertreatment workload of manually changing the line map is big.
Summary of the invention
At the problems referred to above, the purpose of this invention is to provide a kind of Natural color automatic simulation method, particularly the situation of the blue wave band disappearance of satellite such as SPOT, IRS multispectral sensor at the high-resolution remote sensing image.By machine learning, the training of wave spectrum storehouse wave spectrum sample, set up the nonlinear relationship model between the wave band, realize the efficient accurately simulation of blue wave band.
Thinking of the present invention is: at first treat the colored remote sensing image of simulates real according to input, select sensor type and bandwidth setting, ground-object spectrum data pre-service to the ground-object spectrum storehouse, by wave spectrum response function and spectrum integral algorithm, the ground-object spectrum data-switching is arrived and the same spectrum yardstick of remote sensing image data; By the ISODATA algorithm, to the remote sensing image initial clustering, according to the initial clustering result, on remote sensing image, choose 200 spectrum reference mark, adopt spectrum angle matching algorithm (SAM), pretreated ground-object spectrum storehouse spectroscopic data is carried out screening sample, select suitable this regional optimal candidate spectrum samples; Make up regression vector machine (SVR) nonlinear regression model (NLRM), to the candidate samples classification, be divided into sample training collection (training sample) and test sample collection (test sample book), the adjustment by cross validation and continuous training and parameter obtains the best model parameter; According to the predictions and simulations of non-linear machine learning model realization to blue wave band.
Technical scheme of the present invention provides the high-resolution remote sensing image Natural color to simulate algorithmic method automatically, it is characterized in that comprising following implementation step:
1) to the satellite remote-sensing image pre-service, transfer pixel DN value to radiance or earth surface reflection rate, recover the remote sensing image physical attribute, and make up wave band compositional variable ratio;
2) selectedly wait to simulate the remote sensing image sensor type,, unify the spectrum yardstick by spectrum integral to the remotely-sensed data pre-service of ground-object spectrum storehouse;
3),, on remote sensing image, choose 200 representative spectrum reference mark according to the requirement at spectrum reference mark to the remote sensing image initial clustering;
4) adopt spectrum similarity judgment models, choose 200 spectrum training samples in the ground-object spectrum storehouse according to image spectrum reference mark;
5) make up nonlinear model,, obtain the optimization model parameter, set up the nonlinear relationship model of blue wave band and known wavelength range by the training checking;
6) application of model calculates blue wave band.
Above-mentioned implementation step is characterised in that:
Step 1) has been carried out radiant correction and spectral reflectivity conversion to remote sensing image, has recovered the remote sensing physical attribute, and wave band concerns explicit physical meaning;
Step 2) the wave band bandwidth of consideration different sensors is provided with and spectral response functions;
Step 3), step 4) are according to the face of land view characteristic distributions of different regions remote sensing image, and the chosen spectrum reference mark distributes, and select the sample spectral data collection of machine learning again from global ground-object spectrum storehouse automatically according to the spectrum reference mark;
Step 5) model training and precision test, a wave spectrum sample set part is used for model training, and another part is used for the model accuracy check; Training sample set and test sample book set pair are changed, cross validation reaches best training effect again.The form of wave band ratio has been adopted in the input of model, promptly by the ratio calculation of each wave band with synthetic wave band, has eliminated the influence of atmosphere radiation transmission on the one hand, makes the input value of model satisfy the data input requirement of SVR on the other hand in [0~1] scope.
Ground-object spectrum storehouse source of reference data is except the global ground-object spectrum storehouse (speclib06) of USGS and the built-in wave spectrum storehouse of ENVI in addition, can obtain by multiple means, as: the typical feature wave spectrum storehouse of China, from the image pickup area end member wave spectrum and open-air measured spectra.
The present invention compared with prior art has following characteristics: from remote sensing wave spectrum physical characteristics, fully used the profuse global object spectrum library information of spectral information.Utilize image spectrum reference mark and Spectral matching algorithm, the automatic selection of implementation model training sample has taken into full account regional view characteristics and object spectrum characteristics, by machine learning and cross validation, obtains the best model parameter.Its theoretical foundation is abundant, explicit physical meaning; True, the nature of the true color image product colour that simulates, spectrum distortion is little, and the spectral information amount is abundant.Saved the artificial toning time in the last handling process greatly.
Description of drawings
Fig. 1. techniqueflow synoptic diagram of the present invention
Fig. 2. ground-object spectrum data base management system (DBMS) master interface
Fig. 3. spectrum yardstick conversion synoptic diagram
Fig. 4. distribution plan is selected at image spectrum reference mark
The blue wave band analog result contrast of Fig. 5 .ETM+ image
Fig. 6. blue band spectrum statistic histogram is relatively
Blue wave band of Fig. 7 .SPOT image simulation and Natural color audio and video products
Embodiment
Below in conjunction with accompanying drawing and the experiment case to the detailed description of the invention.
Fig. 1 is the overall technology flow process of remote sensing image true color simulation, wherein ground-object spectrum database data global ground-object spectrum storehouse sixth version this (speclib06), the built-in ground-object spectrum database data of ENVI and the open-air measured spectra data of some typical features of USGS that have been integrated.For convenience the unified management and the standardization of existing spectral data are used, we rely on the development environment of VC++ according to existing ground-object spectrum data, have set up a cover wave spectrum base management system.Fig. 2 has shown wave spectrum base management system master interface, and what transverse axis was represented is wavelength, and what the longitudinal axis showed is reflectivity.This system can realize management and the visual inquiry and the demonstration of ground-object spectrum data, and leaves interface and make things convenient for other software transfer.
Original remote sensing image, general record be the DN value, be not the true record of clutter reflections feature, do not have strict physical significance.According to wave band analog computation requirement, the original DN Value Data of remote sensing image is converted to radiance value or reflectivity.Computing formula is as follows:
L
λ=Gain×DN+Bias (1)
In the formula (1), DN is the pixel gray-scale value of image, L
λBe radiance, Gain and Bias represent gain and biasing respectively.In the formula (2), ρ represents atmosphere top layer reflectivity, and d is the solar distance parameter, and value is 1, ESUN
λExpression solar spectrum radiant quantity, θ represents solar zenith angle.
What every spectrum character curve in the wave spectrum storehouse write down is the reflectance value that a certain atural object disperses from the ultraviolet to the far infrared band, and its wave spectrum sampling interval is 1nm only.So before Spectral matching, need the wave spectrum record preprocessing,,, realize the conversion of spectrum yardstick by spectrum integral according to the bandwidth setting for the treatment of analog image.Shown the wave spectrum integration that carries out according to the wave band bandwidth of SPOT as Fig. 3.The bandwidth of blue wave band then is provided with reference to the blue wave band bandwidth of some sensors, and the wave spectrum limit of integration is 0.45-0.52 μ m.The spectroscopic data of a series of narrow band of ground light spectrometer test is fitted to the wave band data of a certain remote sensor, and its spectrum integral computing formula is:
In the formula, f (λ) is a reflectivity function, and Γ (λ) is the wave band response function, and λ is a wavelength, λ
1, λ
2Be respectively the wavelength of corresponding wave band start-stop position.
At present, the metric algorithm of spectrum similarity have a lot as, spectrum angle coupling (SAM), spectrum similarity (SSV), spectrum identification (SIM), spectral correlation (SCM), spectral signature filtering (SSF) or the like.Consider in the spectrum vector space, each vector all has particular length and direction, " color " of material determined by the direction of pixel vector, the present invention has adopted spectrum angle matching algorithm, by calculating vector angle between target optical spectrum and reference spectra to determine their similarity degree.If n is wave band number (dimension),
Be pixel wave spectrum vector,
Be the reference spectra vector, then broad sense spectrum angle θ between the two is expressed as:
Formula (3) can be expressed as:
Fig. 4 has shown the spectrum reference mark of choosing automatically on ETM+ remote sensing image to be simulated, initial clustering by image, selection by classification control spectrum reference mark, the spectrum that guarantees each classification can both be chosen, in addition, control the spectrum reference mark number of each class according to the area (landscape pattern) of classification.The purpose of above algorithm is to make the spectrum reference mark of choosing automatically have comprehensive and representative.
Compare with other machine learning algorithm, the SVM method is calculated quick and precisely, and has good generalization ability under the small sample restriction, is applicable under the situation at a small amount of spectrum reference mark, ground the simulation of blue wave band data.The present invention is integrated libsvm2.9 algorithm bag (Chang﹠amp; Lin 2001), homing method is chosen as epsilon-SVR, and kernel function is gaussian kernel (RBF), and each training pattern all needs to have chosen optimum SVR model parameter g and C through cross validation, to obtain reliable and stable blue wave band appraising model.
In order to reduce the error of calculation and the input requirement of satisfying the SVR model, the form of wave band ratio has all been adopted in the input and output of model.Specific algorithm is as follows:
Secondly, model training parameter;
According to synthetic variable, to the input variable (X of each training
I, i) ask and calculate the variable ratio
The input parameter of SVR model is like this:
Be input variable,
Be reference variable.
The 3rd, model input, output;
According to compositional variable than principle, the input of model, the same form that adopts band ratio of output, being input as of SVR forecast model:
Model is output as:
B wherein
sFor simulating blue wave band.
At last, blue wave band reflectivity transforms;
Because the direct input of SVR model is the form of synthetic wave band ratio, at this moment we need transform this, transfer the wave band ratio to the wave band reflectivity, and computing formula is:
Fig. 5 has shown the blue wave band simulate effect of ETM+ image, and Fig. 5 (a) is the blue wave band of original ETM+, and Fig. 5 (b) is blue wave band for ETM+ simulates.Because the ETM+ sensor has blue wave band, by blue wave band simulation to it, validity that can testing model.Differentiate experimental result as seen from visual angle, two width of cloth image difference are very little, and the present invention can simulate original blue wave band more exactly.Fig. 6 is that the histogram to the blue wave band of original blue wave band and simulation compares, on the shape and distribution of histogram curve, the histogram of two blue wave bands is closely similar, and the similarity of simulating blue wave band and original blue wave band has been described equally, has proved validity of the present invention.For further verifying this conclusion, table 1 has provided the basic statistics information of simulating blue wave band and original blue wave band.As can be seen from the table, statistical value is more approaching, and related coefficient has reached 0.944, and the blue wave band of the ETM+ of simulation can reduce original blue band spectrum information more exactly, and analog result is with a high credibility.
Table 1 blue wave band of simulation and original blue wave band Statistics table
Fig. 7 has shown the true color simulation of the present invention to the SPOT-5 image.Fig. 7 (a) is the blue wave band of the SPOT-5 of simulation; Fig. 7 (b) is the SPOT-5 Natural color image that simulation obtains.From the SPOT-5 Natural color design sketch of Fig. 7 (b) simulation, simulating nature look audio and video products, true, the nature of color does not have tangible spectrum distortion phenomenon; On visual effect, can satisfy the requirement of high-definition remote sensing Natural color audio and video products drawing.The present invention has obtained using widely in the project SPOT Natural color audio and video products drawing of Xinjiang forestry remote sensing investigation.
Claims (3)
1. self-colored analogy method of the high-resolution remote sensing image based on the object spectrum data is characterized in that following step:
(1) chooses sensor type, band setting and the spectral response functions of waiting to simulate remote sensing image;
(2),,, realize the conversion of wave spectrum yardstick by spectrum integral to the ground-object spectrum data pre-service in ground-object spectrum storehouse according to setting of wave band bandwidth and the spectral response functions for the treatment of analog image;
(3) the characteristic spectrum sample is selected: automatic chosen spectrum feature reference mark on remote sensing image to be simulated, (SAM:spectral mapping angle) filters out the object spectrum data in the ground-object spectrum storehouse by spectrum angle matching algorithm, as candidate's training sample;
(4) non-linear simulation: make up the non-linear machine learning model of support vector machine (SVM), by the wave spectrum sample data, to the learning training and the precision test of regression vector machine (SVR), set up nonlinear relationship implicit between blue wave band to be simulated and the known remote sensing image wave band;
(5) calculate (SVR) by the nonlinear model between the wave band, realize blue wave band simulation the high-resolution remote sensing image, and then synthetic Natural color image data product.
2. according to the system of selection at automatic spectrum feature reference mark on the described remote sensing image of claim 1, it is characterized in that by introducing ISODATA spectrum clustering algorithm, original remote sensing image has been carried out initial clustering, then according to the ratio of each class, on each class, choose some spectrum reference mark randomly, guarantee that the characteristic spectrum reference mark of choosing is representative and comprehensive; Again according to the SAM algorithm, preferred ground-object spectrum training sample data in the ground-object spectrum storehouse.
3. according to the structure of the described non-linear simulation method of claim 1----supporting vector machine model (libsvm), it is characterized in that libsvm2.9 algorithm bag (Chang﹠amp integrated; Lin 2001), the training sample data part in the ground-object spectrum storehouse is used for training, and another part is used for accuracy test; Optimum SVM parameter g and C have been chosen through cross validation, to obtain reliable and stable blue wave band appraising model.In addition, improve simulation precision in order to reduce error, the input of model, output parameter all adopt the form of band ratio.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2011101273074A CN102222238A (en) | 2011-05-17 | 2011-05-17 | Automatic simulation method of natural-color products of high-space-resolution remote sensing images |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2011101273074A CN102222238A (en) | 2011-05-17 | 2011-05-17 | Automatic simulation method of natural-color products of high-space-resolution remote sensing images |
Publications (1)
Publication Number | Publication Date |
---|---|
CN102222238A true CN102222238A (en) | 2011-10-19 |
Family
ID=44778785
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN2011101273074A Pending CN102222238A (en) | 2011-05-17 | 2011-05-17 | Automatic simulation method of natural-color products of high-space-resolution remote sensing images |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102222238A (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102445712A (en) * | 2011-11-22 | 2012-05-09 | 成都理工大学 | Character window weighting related spectrum matching method facing rocks and minerals |
CN109460755A (en) * | 2018-10-29 | 2019-03-12 | 北京航天宏图信息技术股份有限公司 | A kind of green light estimation model training method and device, image synthesis method and device |
CN109508397A (en) * | 2018-11-16 | 2019-03-22 | 中国铁路设计集团有限公司 | A kind of remote sensing image processing method based on master drawing library |
CN110070513A (en) * | 2019-04-30 | 2019-07-30 | 上海同繁勘测工程科技有限公司 | The radiation correction method and system of remote sensing image |
CN111598961A (en) * | 2020-05-22 | 2020-08-28 | 中国科学院西北生态环境资源研究院 | Surface feature information extraction method and device, electronic equipment and storage medium |
US20210272321A1 (en) * | 2020-02-13 | 2021-09-02 | The Government Of The United States Of America, As Represented By The Secretary Of The Navy | Chromatic Domain Mapping |
CN113807219A (en) * | 2021-09-06 | 2021-12-17 | 苏州中科蓝迪软件技术有限公司 | Method for identifying types of grain and oil crops in planting land by steps |
CN114445717A (en) * | 2022-04-11 | 2022-05-06 | 山东沃能安全技术服务有限公司 | Remote sensing image processing method and system for land resource management |
CN115082582A (en) * | 2022-06-09 | 2022-09-20 | 珠江水利委员会珠江水利科学研究院 | True color simulation method, system, equipment and medium for satellite remote sensing data |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101276420A (en) * | 2008-04-17 | 2008-10-01 | 中国科学院地理科学与资源研究所 | Classification method for syncretizing optical spectrum information and multi-point simulation space information |
CN101604017A (en) * | 2009-07-16 | 2009-12-16 | 北京航空航天大学 | A kind of method that realizes remote sensing image simulation under the given elements of exterior orientation |
-
2011
- 2011-05-17 CN CN2011101273074A patent/CN102222238A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101276420A (en) * | 2008-04-17 | 2008-10-01 | 中国科学院地理科学与资源研究所 | Classification method for syncretizing optical spectrum information and multi-point simulation space information |
CN101604017A (en) * | 2009-07-16 | 2009-12-16 | 北京航空航天大学 | A kind of method that realizes remote sensing image simulation under the given elements of exterior orientation |
Non-Patent Citations (2)
Title |
---|
朱长明等: "地物波谱数据辅助的SPOT影像模拟真彩色方法研究", 《测绘学报》, vol. 39, no. 2, 30 April 2010 (2010-04-30), pages 169 - 174 * |
程熙等: "利用地物波谱学习的遥感影像波段模拟方法", 《红外与毫米波学报》, vol. 29, no. 1, 28 February 2010 (2010-02-28), pages 45 - 48 * |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102445712A (en) * | 2011-11-22 | 2012-05-09 | 成都理工大学 | Character window weighting related spectrum matching method facing rocks and minerals |
CN109460755A (en) * | 2018-10-29 | 2019-03-12 | 北京航天宏图信息技术股份有限公司 | A kind of green light estimation model training method and device, image synthesis method and device |
CN109508397A (en) * | 2018-11-16 | 2019-03-22 | 中国铁路设计集团有限公司 | A kind of remote sensing image processing method based on master drawing library |
CN110070513A (en) * | 2019-04-30 | 2019-07-30 | 上海同繁勘测工程科技有限公司 | The radiation correction method and system of remote sensing image |
CN110070513B (en) * | 2019-04-30 | 2021-10-01 | 上海同繁勘测工程科技有限公司 | Radiation correction method and system for remote sensing image |
US20210272321A1 (en) * | 2020-02-13 | 2021-09-02 | The Government Of The United States Of America, As Represented By The Secretary Of The Navy | Chromatic Domain Mapping |
US11816867B2 (en) * | 2020-02-13 | 2023-11-14 | The Government Of The United States Of America, As Represented By The Secretary Of The Navy | Chromatic domain mapping |
CN111598961A (en) * | 2020-05-22 | 2020-08-28 | 中国科学院西北生态环境资源研究院 | Surface feature information extraction method and device, electronic equipment and storage medium |
CN113807219A (en) * | 2021-09-06 | 2021-12-17 | 苏州中科蓝迪软件技术有限公司 | Method for identifying types of grain and oil crops in planting land by steps |
CN114445717A (en) * | 2022-04-11 | 2022-05-06 | 山东沃能安全技术服务有限公司 | Remote sensing image processing method and system for land resource management |
CN114445717B (en) * | 2022-04-11 | 2022-06-28 | 山东沃能安全技术服务有限公司 | Remote sensing image processing method and system for land resource management |
CN115082582A (en) * | 2022-06-09 | 2022-09-20 | 珠江水利委员会珠江水利科学研究院 | True color simulation method, system, equipment and medium for satellite remote sensing data |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102222238A (en) | Automatic simulation method of natural-color products of high-space-resolution remote sensing images | |
Guo et al. | FSDAF 2.0: Improving the performance of retrieving land cover changes and preserving spatial details | |
Qiao et al. | UAV-based chlorophyll content estimation by evaluating vegetation index responses under different crop coverages | |
CN107392925B (en) | Remote sensing image ground object classification method based on super-pixel coding and convolutional neural network | |
Shi et al. | Remote chlorophyll-a estimates for inland waters based on a cluster-based classification | |
CN109359623B (en) | Hyperspectral image migration classification method based on depth joint distribution adaptive network | |
CN102829739B (en) | Object-oriented remote sensing inversion method of leaf area index of crop | |
CN103679675B (en) | Remote sensing image fusion method oriented to water quality quantitative remote sensing application | |
CN103971123B (en) | Hyperspectral image classification method based on linear regression Fisher discrimination dictionary learning (LRFDDL) | |
CN102110227B (en) | Compound method for classifying multiresolution remote sensing images based on context | |
CN103971115A (en) | Automatic extraction method for newly-increased construction land image spots in high-resolution remote sensing images based on NDVI and PanTex index | |
CN110390255A (en) | High-speed rail environmental change monitoring method based on various dimensions feature extraction | |
Xu et al. | Effects of image pansharpening on soil total nitrogen prediction models in South India | |
Feng et al. | Mapping of plastic greenhouses and mulching films from very high resolution remote sensing imagery based on a dilated and non-local convolutional neural network | |
CN112733596A (en) | Forest resource change monitoring method based on medium and high spatial resolution remote sensing image fusion and application | |
Jadhav et al. | Automatic semantic segmentation and classification of remote sensing data for agriculture | |
CN104134069B (en) | A kind of shale microsection automatic identification method | |
CN110705449A (en) | Land utilization change remote sensing monitoring analysis method | |
CN103426158A (en) | Method for detecting two-time-phase remote sensing image change | |
Qiao et al. | Estimating maize LAI by exploring deep features of vegetation index map from UAV multispectral images | |
CN102096826B (en) | Compound classification method for multi-resolution remote sensing image on basis of real likelihood characteristic | |
Yang et al. | Effect of spring irrigation on soil salinity monitoring with UAV-borne multispectral sensor | |
Liang et al. | Maximum likelihood classification of soil remote sensing image based on deep learning | |
Puzachenko et al. | Assessing the thermodynamic variables of landscapes in the southwest part of East European plain in Russia using the MODIS multispectral band measurements | |
CN115187865B (en) | Unmanned aerial vehicle image near-infrared band simulation method and medium based on ground object spectrum |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20111019 |