CN113902717A - Satellite-borne hyperspectral farmland bare soil target identification method based on spectrum library - Google Patents
Satellite-borne hyperspectral farmland bare soil target identification method based on spectrum library Download PDFInfo
- Publication number
- CN113902717A CN113902717A CN202111193687.1A CN202111193687A CN113902717A CN 113902717 A CN113902717 A CN 113902717A CN 202111193687 A CN202111193687 A CN 202111193687A CN 113902717 A CN113902717 A CN 113902717A
- Authority
- CN
- China
- Prior art keywords
- pixel
- farmland
- image
- spectral
- reflectivity
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10052—Images from lightfield camera
-
- 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
Abstract
The invention discloses a satellite-borne hyperspectral farmland bare soil target identification method based on a spectrum library, which comprises the steps of preprocessing satellite-borne hyperspectral images of a research area to obtain images of the surface reflectivity of the farmland of the research area; resampling the spectrum according to the surface reflectivity image, and constructing a farmland typical feature spectrum library; calculating the spectrum similarity between the farmland surface reflectivity and the spectra of all the surface features in the spectrum library pixel by pixel, and selecting an image pixel most similar to the spectra of all the surface features as an initial sample of the surface feature type; performing spectral dimensional filtering and spatial dimensional filtering on the reflectivity image to obtain a filtered reflectivity image; carrying out feature extraction and feature normalization processing on the filtered reflectivity image to construct an image spectral feature set; calculating the spectral angular distance and Euclidean distance between pixels in all initial sample neighborhood windows and each type of initial samples, and amplifying the training samples; and carrying out multi-classifier comprehensive classification by using the training samples to obtain a farmland bare soil target space distribution map of the research area.
Description
Technical Field
The invention relates to the technical field of satellite-borne hyperspectral data farmland bare soil identification, in particular to a satellite-borne hyperspectral farmland bare soil target identification method based on a spectrum library.
Background
With the rapid development of satellite remote sensing, satellite-borne remote sensing data has gradually become an important data source for farmland soil quality investigation monitoring and evaluation. Compared with the traditional field investigation method, the satellite remote sensing technology has the advantages of rapidness, economy, environmental protection, no damage, repeatability and the like, and provides a new means for farmland soil quality investigation and monitoring with large range, high frequency, high density and continuous space. Accurate bare soil target identification is the key for realizing high-precision inversion of soil quality information.
Since the bare soil spectrum is controlled by various factors such as soil type, soil moisture, organic matter content and ferric oxide content, bare soil extracted by using traditional multispectral data is often interfered by straws, water, shadows and buildings. The hyperspectral remote sensing data has hundreds of spectral channels and nearly continuous spectral curves, and can be used for more finely depicting the soil spectral characteristics and realizing high-precision identification of bare soil targets.
With the launching of hyperspectral satellites such as the high-score fifth satellite, the first Zhuhai satellite, the first resource 02D, the high-score fifth 02 star and the like, the area coverage capacity and the coverage frequency of the satellite-borne hyperspectral data in China are greatly improved. Except for the first Zhuhai, each satellite-borne hyperspectral sensor covers the spectral range from visible light to near infrared (VNIR) to Short Wave Infrared (SWIR), can cover the absorption characteristics of main soil components such as organic matters, mineral matters, water and the like, and provides a stable detection means and rich data sources for regional and national scale soil quality large-scale and normalized investigation.
At present, the bare soil target identification research specially aiming at satellite-borne hyperspectrum at home and abroad is relatively less, and the method is mostly realized by manually selecting training samples to perform supervision classification or target identification. The method needs a large amount of manpower and time cost, results are greatly influenced by manual sample selection, and large-scale and batch processing matched with satellite data volume is difficult to perform. Therefore, the invention researches and provides a satellite-borne hyperspectral farmland bare soil target identification technology based on the spectrum library aiming at the characteristics of domestic satellite-borne hyperspectral remote sensing data, can realize the automatic identification of the satellite-borne hyperspectral farmland bare soil target after the construction of the spectrum library and the setting of necessary parameters are completed in advance, has better application prospects in natural resource investigation and monitoring, farmland soil quality investigation and evaluation and the like, and has important significance for promoting the business application of China hyperspectral satellites in soil quality monitoring.
Disclosure of Invention
In order to solve the technical problems, the invention aims to provide a satellite-borne hyperspectral farmland bare soil target identification method based on a spectrum library. The technology has the characteristics of simplicity in operation, high automation degree and the like, does not need to manually select training samples, and is suitable for large-scale and business processing of satellite-borne hyperspectral data.
The purpose of the invention is realized by the following technical scheme:
a satellite-borne hyperspectral farmland bare soil target identification method based on a spectrum library comprises the following steps:
step A, preprocessing a satellite-borne hyperspectral image of a research area to obtain a farmland surface reflectivity image of the research area;
b, resampling the spectrum according to the obtained surface reflectivity image, and constructing a farmland typical surface feature spectrum library;
step C, calculating the spectrum similarity between the reflectivity of the farmland surface and the spectrums of the ground objects in the spectrum library pixel by pixel, and selecting an image pixel most similar to the spectrums of the ground objects as an initial sample of the ground object type;
step D, performing spectral dimension filtering and spatial dimension filtering processing on the reflectivity image to obtain a filtered reflectivity image;
e, performing feature extraction and feature normalization processing on the filtered reflectivity image to construct an image spectral feature set;
f, calculating the spectral angular distance and Euclidean distance between pixels in all initial sample neighborhood windows and each type of initial samples, and amplifying the training samples;
and G, carrying out multi-classifier comprehensive classification by using the training samples, carrying out class combination on each group of classification results, combining different soil types into 'bare soil', marking the pixel as 'bare soil' if two or more groups of bare soil in three groups of bare soil identification are 'bare soil', otherwise marking the pixel as 'non-bare soil', and finally obtaining a farmland bare soil target space distribution map in the research area.
One or more embodiments of the present invention may have the following advantages over the prior art:
the invention provides a satellite-borne hyperspectral farmland bare soil target identification method based on a spectrum library, and provides a specific implementation process and a specific implementation method of satellite-borne hyperspectral image preprocessing, farmland typical ground object spectrum library construction, determination of initial samples of all ground object types, reflectance image filtering processing, spectral feature extraction and feature normalization, training sample amplification and multi-classifier comprehensive classification. The method has the advantages of large detectable range, high speed, high identification precision, real-time monitoring and the like, can be widely applied to hyperspectral satellite data such as the emitted hyperspectral fifth, the Zhuhai first, the resource first 02D, the hyperspectral fifth 02 star and the like in China, realizes the high-precision identification of bare soil targets in farmland areas in batches, in large scale and automatically, and provides important technical support for the satellite-borne hyperspectral soil quality monitoring analysis.
Drawings
FIG. 1 is a flow chart of a satellite-borne hyperspectral farmland bare soil target identification method based on a spectrum library;
FIGS. 2a and 2b are Savitzky-Golay filtered front and rear reflectance spectra, respectively, and FIGS. 2c and 2d are PCA-SAD filtered front and rear reflectance image maps, respectively;
fig. 3a, 3b, 3c, and 3d are graphs illustrating the bare soil target recognition result of the first area, the bare soil actual distribution situation of the first area, the bare soil target recognition result of the second area, and the bare soil actual distribution situation of the second area, respectively.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the following embodiments and accompanying drawings.
As shown in FIG. 1, the method for identifying the bare soil target of the satellite-borne hyperspectral farmland based on the spectral library, provided by the invention, comprises the following steps:
The step 10 is to collect and obtain a ZY1-02D satellite L1A-level spectral image according to the research area range, correct the radiation coefficient based on ENVI software, select a FLAASH model to correct the atmosphere, visually check the problems of image strip noise, CCD inter-slice inconsistency, bad lines and the like band by band, improve the quality of the band with problems by a moment matching method, correct the image orthographic projection, correct the terrain and the solar luminosity and inlay the image by utilizing the ENVI software, and respectively cut the image and mask the farmland area by utilizing vector files distributed in the research area range and the research area farmland to obtain the farmland surface reflectivity image of the research area.
In the step 20, typical ground object types possibly contained in the farmland of the research area are determined according to the pre-collected soil type distribution data, agricultural planting data, land utilization type maps and the like of the research area and the image shooting time, specifically including black soil, black calcium soil, meadow soil, marsh soil, saline-alkali soil, aeolian sandy soil, rice straws, corn straws, soybean straws, greenhouses, mulching films, water bodies, artificial buildings and the like, the type spectrums of the various ground objects are collected, resampling is performed according to the spectral response function of the ZY1-02D satellite hyperspectral images, and a typical ground object spectrum library of the farmland is constructed.
In the above step 30, the spectral angular distance SAD is used as an evaluation index in the spectral similarity calculation, and the smaller SAD, the more similar the two spectra are considered. The SAD between the two spectra is calculated as follows:
wherein, I and R respectively represent two spectral vectors to be compared;
when the initial sample of each ground feature type is specifically determined, firstly, for each ground feature type, taking the type A as an example, a pixel with the minimum SAD with the spectrum of the type A is searched, then the SAD of the pixel and other types is calculated, and if the SAD value is smaller than that of the type A, the image is considered to not contain the type A.
The foregoing step 40 sequentially performs Savitzky-Golay filtering in the spectral dimension and PCA-SAD filtering in the spatial dimension on the preprocessed reflectance image to obtain a filtered reflectance image, and the reflectance curve and the image before and after filtering are shown in fig. 2a, 2b, 2c, and 2 d. The PCA-SAD filtering method is realized by writing codes through IDL, and the specific realization process is as follows:
1) the iteration number, the size of a neighborhood window and an SAD threshold value MinSAD are set, the size of the neighborhood window can be set to be 9 multiplied by 9 or 11 multiplied by 11, and the like, the MinSAD can be set to be 0.015-0.020, and the iteration number can be set to be 1-3;
2) carrying out PCA (principal component analysis) conversion on the satellite-borne hyperspectral reflectivity data, reserving 5 principal components, and normalizing the principal components;
3) calculating the SAD value of each pixel in the central pixel and the neighborhood window pixel by pixel based on the main component normalized in the previous step, and determining the pixel in the window, the SAD value of which is less than MinSAD with the central pixel, as a neighborhood similar pixel;
4) based on the reflectivity data, calculating the average value of the similar pixels of the neighborhood as a central point value;
5) repeating 1) to 4) until the iteration number meets the requirement.
Spectral feature extraction and feature normalization in step 50 above: performing feature extraction and feature normalization on the reflectivity image after the filtering in the fourth step, firstly calculating 22 spectral features such as SOC, NDVI _800_680, SRI _800_680, EVI _800_680_450, redvi _750_705, MRESRI _750_705_445, MRENDVI _750_705_445, SGI _500_600, VREI _740_720, MSI _1599_819, REF _533, REF _791, REF _1005, REF _1576, REF _2249, SAI _1543_1644_1728, SAI _670_833_945, SAI _567_756, RefSum _1543_1728, RefSum _2098_2283, RefSum _567_756, RefSlope _499_ feature set and the like according to the formula in table 1, then performing feature-by-feature normalization, and constructing a spectrum interval [0, pixel ratio ] of all images according to the maximum value and minimum value of a farmland statistical area, and a farmland.
TABLE 1
In the step 60, training samples are amplified by calculating SAD and Euclidian Distances (ED) between pixels in all the initial sample neighborhood windows and each type of initial sample, and the number of the training samples of each type after amplification is greater than the feature number of the spectral feature set. And for each pixel in the 11 multiplied by 11 neighborhood window of each type of initial sample in the third step, calculating SAD and ED of the pixel and each type of initial sample pixel, and if the initial sample type corresponding to the minimum value of the SAD and ED is the same, adding the pixel into the training sample. If the number of the training samples of a certain category after amplification is smaller than the feature number of the spectral feature set, gradually expanding the window by taking 2 as a unit, and continuing to amplify the samples of the category until the number of the training samples of all the categories meets the requirement.
The step 70 comprises respectively using three classifiers, namely spectral angle mapping, support vector machine and random forest to supervise and classify based on the obtained training samples of each class, combining the classification results of each group, combining different soil types into 'bare soil', combining the rest types into 'non-bare soil' to obtain three groups of bare soil identification results, then carrying out pixel-by-pixel statistics on the three groups of results, for each pixel, when two or more groups of results in the three groups of results are identified as 'bare soil', the pixel is marked as 'bare soil', otherwise, the pixel is marked as 'non-bare soil', and finally a farmland bare soil target space distribution diagram in a research area is obtained, FIGS. 3a, 3b, 3c and 3d are the comparison between the identification result of the bare soil target in the research area and the verification data, that is, fig. 3a shows the bare soil object recognition result of the area one, and fig. 3b shows the actual bare soil distribution of the area one; fig. 3c shows the bare soil object recognition result in the second area, and fig. 3d shows the actual bare soil distribution in the second area.
Although the embodiments of the present invention have been described above, the above descriptions are only for the convenience of understanding the present invention, and are not intended to limit the present invention. It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (8)
1. A satellite-borne hyperspectral farmland bare soil target identification method based on a spectrum library is characterized by comprising the following steps:
step A, preprocessing a satellite-borne hyperspectral image of a research area to obtain a farmland surface reflectivity image of the research area;
b, resampling the spectrum according to the obtained surface reflectivity image, and constructing a farmland typical surface feature spectrum library;
step C, calculating the spectrum similarity between the reflectivity of the farmland surface and the spectrums of the ground objects in the spectrum library pixel by pixel, and selecting an image pixel most similar to the spectrums of the ground objects as an initial sample of the ground object type;
step D, performing spectral dimension filtering and spatial dimension filtering processing on the reflectivity image to obtain a filtered reflectivity image;
e, performing feature extraction and feature normalization processing on the filtered reflectivity image to construct an image spectral feature set;
f, calculating the spectral angular distance and Euclidean distance between pixels in all initial sample neighborhood windows and each type of initial samples, and amplifying the training samples;
and G, carrying out multi-classifier comprehensive classification by using the training samples, carrying out class combination on each group of classification results, combining different soil types into 'bare soil', marking the pixel as 'bare soil' if two or more groups of bare soil in three groups of bare soil identification are 'bare soil', otherwise marking the pixel as 'non-bare soil', and finally obtaining a farmland bare soil target space distribution map in the research area.
2. The method for identifying the bare soil target in the spaceborne hyperspectral farmland based on the spectral library as claimed in claim 1, wherein the image of the surface reflectivity of the farmland in the step A is obtained by preprocessing the image of the surface reflectivity of the farmland according to the product grade, such as radiance correction, atmospheric correction, image radiance detection and quality improvement, orthorectification, terrain and solar luminosity correction, image mosaic and cutting and farmland area masking.
3. The method for identifying the bare soil target of the satellite-borne hyperspectral farmland based on the spectral library as claimed in claim 1, wherein the image radiation problem is detected through a visual or automatic algorithm in the step A, the image strip noise, the inconsistency between CCD (charge coupled device) chips and bad lines are mainly detected, the quality of the wave band with the problem is improved by using a moment matching method, and the wave band with the serious problem after the quality is improved is abandoned.
4. The method for identifying the bare soil target of the satellite-borne hyperspectral farmland based on the spectrum library as claimed in claim 1, wherein the step 2 specifically comprises the steps of determining the type of typical ground objects contained in the farmland in a research area according to pre-collected data and image shooting time, collecting or actually measuring the spectrum of each typical ground object, and performing spectrum resampling according to the reflectivity image to construct the farmland typical ground object spectrum library; the spectrum library comprises various soil, crops, straws, water bodies, greenhouses, mulching films, clouds, snow, shadows and building ground substance spectrums in a research area.
5. The method for identifying the bare soil target of the satellite-borne hyperspectral farmland based on the spectral library as claimed in claim 1, wherein the spectral similarity calculation in the step C takes spectral angular distance SAD as an evaluation index, and the smaller the SAD is, the more similar the spectrums of the two are considered; the SAD between the two spectra is calculated as follows:
wherein, I and R respectively represent two spectral vectors to be compared;
when determining the initial samples of the various surface feature types, two conditions need to be satisfied simultaneously: 1) the SAD of the pixel and the surface feature type is smaller than the SAD of any other pixel and the type; 2) the SAD of the pixel and the ground object type is smaller than the SAD of the pixel and any other type; and if the image element which meets the two conditions simultaneously does not exist in a certain surface feature type, the image is considered not to contain the surface feature type.
6. The method for identifying the bare soil target of the satellite-borne hyperspectral farmland based on the spectral library as claimed in claim 1, wherein the image spectral dimensional filtering in the step D adopts a Savitzky-Golay filtering method, and the spatial dimensional filtering adopts a neighborhood SAD filtering method based on Principal Component Analysis (PCA); the PCA-SAD filtering method is specifically realized as follows:
1) the iteration number, the size of a neighborhood window and an SAD threshold value MinSAD are set, the size of the neighborhood window can be set to be 9 multiplied by 9 or 11 multiplied by 11, and the like, the MinSAD can be set to be 0.015-0.020, and the iteration number can be set to be 1-3;
2) carrying out PCA (principal component analysis) conversion on the satellite-borne hyperspectral reflectivity data, reserving 5 principal components, and normalizing the principal components;
3) calculating the SAD value of each pixel in the central pixel and the neighborhood window pixel by pixel based on the main component normalized in the previous step, and determining the pixel in the window, the SAD value of which is less than MinSAD with the central pixel, as a neighborhood similar pixel;
4) based on the reflectivity data, calculating the average value of the similar pixels of the neighborhood as a central point value;
5) repeating 1) to 4) until the iteration number meets the requirement.
7. The method for identifying the bare soil target of the satellite-borne hyperspectral farmland based on the spectral library as claimed in claim 1, wherein the spectral features extracted in the step E comprise spectral indexes, reflectivity at specific wavelength positions, reflectivity absorption depth, specific wavelength range reflectivity summation and reflectivity slopes.
8. The method for identifying the bare soil target of the satellite-borne hyperspectral farmland based on the spectral library as claimed in claim 1, wherein in the step F, when the training samples are amplified, for each pixel in the neighborhood window of each type of initial sample, the SAD and Euclidean distances between the pixel and each type of initial sample pixel are calculated, and if the initial sample type corresponding to the minimum value of the pixel and each type of initial sample pixel is the same, the pixel is added into the training samples; and if the number of the training samples of a certain category after amplification is less than the characteristic number of the spectral feature set, gradually expanding the window by taking 2 as a unit, and continuing to amplify the category samples until the number of the training samples of all categories meets the requirement.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111193687.1A CN113902717B (en) | 2021-10-13 | 2021-10-13 | Satellite-borne hyperspectral farmland bare soil target identification method based on spectrum library |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111193687.1A CN113902717B (en) | 2021-10-13 | 2021-10-13 | Satellite-borne hyperspectral farmland bare soil target identification method based on spectrum library |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113902717A true CN113902717A (en) | 2022-01-07 |
| CN113902717B CN113902717B (en) | 2022-04-26 |
Family
ID=79191879
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111193687.1A Active CN113902717B (en) | 2021-10-13 | 2021-10-13 | Satellite-borne hyperspectral farmland bare soil target identification method based on spectrum library |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113902717B (en) |
Citations (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050100220A1 (en) * | 2002-11-06 | 2005-05-12 | Keaton Patricia A. | Method and apparatus for automatically extracting geospatial features from multispectral imagery suitable for fast and robust extraction of landmarks |
| CN101814148A (en) * | 2010-04-30 | 2010-08-25 | 霍振国 | Remote sensing hyperspectral image classification method based on semi-supervised kernel adaptive learning |
| CN101950361A (en) * | 2010-09-06 | 2011-01-19 | 中国科学院遥感应用研究所 | Adaptive extraction method of remote sensing image thematic information based on spectrum matching degree |
| CN102012528A (en) * | 2010-11-23 | 2011-04-13 | 北京理工大学 | Hyperspectral remote sensing oil-gas exploration method for vegetation sparse area |
| CN102930293A (en) * | 2012-10-19 | 2013-02-13 | 中国科学院自动化研究所 | Object-oriented hyperspectral characteristic processing method |
| CN103646246A (en) * | 2013-12-19 | 2014-03-19 | 交通运输部天津水运工程科学研究所 | Decision tree model based multispectral remote sensing image river information extraction method |
| CN104463203A (en) * | 2014-12-03 | 2015-03-25 | 复旦大学 | Hyper-spectral remote sensing image semi-supervised classification method based on ground object class membership grading |
| CN104463881A (en) * | 2014-12-12 | 2015-03-25 | 西安电子科技大学 | Multi-spectral remote sensing image change detection method based on spectral reflectivity neighborhood difference chart and neighborhood probability fusion |
| CN105975912A (en) * | 2016-04-27 | 2016-09-28 | 天津大学 | Hyperspectral image nonlinearity solution blending method based on neural network |
| US20160328838A1 (en) * | 2015-05-01 | 2016-11-10 | Applied Research LLC. | Automatic target recognition system with online machine learning capability |
| CN106339674A (en) * | 2016-08-17 | 2017-01-18 | 中国地质大学（武汉） | Hyperspectral image classification method based on edge preservation and graph cut model |
| CN106447688A (en) * | 2016-03-31 | 2017-02-22 | 大连海事大学 | Method for effectively segmenting hyperspectral oil-spill image |
| US20180031422A1 (en) * | 2015-01-14 | 2018-02-01 | The University Of Sydney | Hyperspectral imager method and apparatus |
| KR101873003B1 (en) * | 2018-04-10 | 2018-07-02 | (주)아세아항측 | Detection of Heavy Metal in Soil Using hyperspectral Image and Spectral Angle Variables |
| US20190293489A1 (en) * | 2017-10-24 | 2019-09-26 | Korea Institute Of Ocean Science & Technology | Method for detecting shipwrecked vessel and drown victims by using aerial hyperspectral image |
| JP2019220176A (en) * | 2018-06-15 | 2019-12-26 | 大学共同利用機関法人情報・システム研究機構 | Image processing device and method |
| US20200019817A1 (en) * | 2018-07-11 | 2020-01-16 | Harbin Institute Of Technology | Superpixel classification method based on semi-supervised k-svd and multiscale sparse representation |
| CN110796038A (en) * | 2019-10-15 | 2020-02-14 | 南京理工大学 | Hyperspectral remote sensing image classification method combined with rapid region growing superpixel segmentation |
| CN112396029A (en) * | 2020-12-03 | 2021-02-23 | 宁波大学 | Clustering segmentation and coupling end member extraction synergistic hyperspectral coastal wetland subpixel change detection method |
| CN112990368A (en) * | 2021-04-26 | 2021-06-18 | 湖南大学 | Polygonal structure guided hyperspectral image single sample identification method and system |
-
2021
- 2021-10-13 CN CN202111193687.1A patent/CN113902717B/en active Active
Patent Citations (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050100220A1 (en) * | 2002-11-06 | 2005-05-12 | Keaton Patricia A. | Method and apparatus for automatically extracting geospatial features from multispectral imagery suitable for fast and robust extraction of landmarks |
| CN101814148A (en) * | 2010-04-30 | 2010-08-25 | 霍振国 | Remote sensing hyperspectral image classification method based on semi-supervised kernel adaptive learning |
| CN101950361A (en) * | 2010-09-06 | 2011-01-19 | 中国科学院遥感应用研究所 | Adaptive extraction method of remote sensing image thematic information based on spectrum matching degree |
| CN102012528A (en) * | 2010-11-23 | 2011-04-13 | 北京理工大学 | Hyperspectral remote sensing oil-gas exploration method for vegetation sparse area |
| CN102930293A (en) * | 2012-10-19 | 2013-02-13 | 中国科学院自动化研究所 | Object-oriented hyperspectral characteristic processing method |
| CN103646246A (en) * | 2013-12-19 | 2014-03-19 | 交通运输部天津水运工程科学研究所 | Decision tree model based multispectral remote sensing image river information extraction method |
| CN104463203A (en) * | 2014-12-03 | 2015-03-25 | 复旦大学 | Hyper-spectral remote sensing image semi-supervised classification method based on ground object class membership grading |
| CN104463881A (en) * | 2014-12-12 | 2015-03-25 | 西安电子科技大学 | Multi-spectral remote sensing image change detection method based on spectral reflectivity neighborhood difference chart and neighborhood probability fusion |
| US20180031422A1 (en) * | 2015-01-14 | 2018-02-01 | The University Of Sydney | Hyperspectral imager method and apparatus |
| US20160328838A1 (en) * | 2015-05-01 | 2016-11-10 | Applied Research LLC. | Automatic target recognition system with online machine learning capability |
| CN106447688A (en) * | 2016-03-31 | 2017-02-22 | 大连海事大学 | Method for effectively segmenting hyperspectral oil-spill image |
| CN105975912A (en) * | 2016-04-27 | 2016-09-28 | 天津大学 | Hyperspectral image nonlinearity solution blending method based on neural network |
| CN106339674A (en) * | 2016-08-17 | 2017-01-18 | 中国地质大学（武汉） | Hyperspectral image classification method based on edge preservation and graph cut model |
| US20190293489A1 (en) * | 2017-10-24 | 2019-09-26 | Korea Institute Of Ocean Science & Technology | Method for detecting shipwrecked vessel and drown victims by using aerial hyperspectral image |
| KR101873003B1 (en) * | 2018-04-10 | 2018-07-02 | (주)아세아항측 | Detection of Heavy Metal in Soil Using hyperspectral Image and Spectral Angle Variables |
| JP2019220176A (en) * | 2018-06-15 | 2019-12-26 | 大学共同利用機関法人情報・システム研究機構 | Image processing device and method |
| US20200019817A1 (en) * | 2018-07-11 | 2020-01-16 | Harbin Institute Of Technology | Superpixel classification method based on semi-supervised k-svd and multiscale sparse representation |
| CN110796038A (en) * | 2019-10-15 | 2020-02-14 | 南京理工大学 | Hyperspectral remote sensing image classification method combined with rapid region growing superpixel segmentation |
| CN112396029A (en) * | 2020-12-03 | 2021-02-23 | 宁波大学 | Clustering segmentation and coupling end member extraction synergistic hyperspectral coastal wetland subpixel change detection method |
| CN112990368A (en) * | 2021-04-26 | 2021-06-18 | 湖南大学 | Polygonal structure guided hyperspectral image single sample identification method and system |
Non-Patent Citations (7)
| Title |
|---|
| FUBIAO FENG 等: "Dimensionality Reduction of Hyperspectral Image with Graph-Based Discriminant Analysis Considering Spectral Similarity", 《REMOTE SENSING》 * |
| KUN SHANG 等: "Comparison of bare soil extraction methods in black soil zone for AHSI/GF-5 remote sensing data", 《PROCEEDINGS OF SPIE》 * |
| W. DEAN HIVELY 等: "Use of Airborne Hyperspectral Imagery to Map Soil Properties in Tilled Agricultural Fields", 《APPLIED AND ENVIRONMENTAL SOIL SCIENCE》 * |
| 冯维一 等: "基于三维光谱角统计的光谱图像信息提取方法", 《光谱学与光谱分析》 * |
| 冯维一: "快照式压缩光谱集成成像技术研究", 《中国博士学位论文全文数据库 信息科技辑》 * |
| 王春阳: "基于信息熵的自训练半监督高光谱遥感影像分类研究", 《中国博士学位论文全文数据库 信息科技辑》 * |
| 甘甫平 等: "深空对地观测创新工程进展与主要成果", 《中国地质调查》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113902717B (en) | 2022-04-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Lu | The potential and challenge of remote sensing‐based biomass estimation | |
| Lu et al. | Assessment of atmospheric correction methods for Landsat TM data applicable to Amazon basin LBA research | |
| CN112541921B (en) | Urban green land vegetation information data accurate determination method | |
| Lu et al. | Application of spectral mixture analysis to Amazonian land-use and land-cover classification | |
| Shi et al. | Mapping plastic greenhouses using spectral metrics derived from GaoFen-2 satellite data | |
| CN112164062A (en) | Wasteland information extraction method and device based on remote sensing time sequence analysis | |
| CN112183209A (en) | Regional crop classification method and system based on multi-dimensional feature fusion | |
| CN112285710B (en) | Multi-source remote sensing reservoir water storage capacity estimation method and device | |
| CN112669363B (en) | Method for measuring three-dimensional green space of urban green space | |
| Manakos et al. | Comparison between atmospheric correction modules on the basis of worldview-2 imagery and in situ spectroradiometric measurements | |
| Ahmad | A review of remote sensing data change detection: Comparison of Faisalabad and Multan Districts, Punjab Province, Pakistan | |
| CN107688777A (en) | A kind of urban green space extracting method for cooperateing with multi-source Remote Sensing Images | |
| CN106934405A (en) | A kind of method of vinyl house extraction and cover degree inverting in intermediate-resolution remote sensing image | |
| CN113869370A (en) | Tree species refined classification method based on deep learning algorithm and time sequence sentinel image | |
| CN112906537A (en) | Crop identification method and system based on convolutional neural network | |
| CN105678280B (en) | Mulching film mulching farmland remote sensing monitoring method based on textural features | |
| Khudhur et al. | Comparison of the accuracies of different spectral indices for mapping the vegetation covers in Al-Hawija district, Iraq | |
| CN113902717B (en) | Satellite-borne hyperspectral farmland bare soil target identification method based on spectrum library | |
| Jusoff et al. | Mapping of individual oil palm trees using airborne hyperspectral sensing: an overview | |
| CN112577954B (en) | Urban green land biomass estimation method | |
| CN113902759B (en) | Space-spectrum information combined satellite-borne hyperspectral image segmentation and clustering method | |
| Wheeler et al. | Crop classification with Landsat multispectral scanner data II | |
| England et al. | A bispectral method for the automatic determination of parameters for use in imaging satellite cloud retrievals | |
| CN112179853A (en) | Fruit tree canopy nitrogen content remote sensing inversion method and system based on image shadow removal | |
| CN112215094B (en) | Method and device for extracting accumulated snow from satellite remote sensing image |
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 |