CN116385593A - Hyperspectral remote sensing mineral mapping method based on quantitative semi-supervised learning - Google Patents
Hyperspectral remote sensing mineral mapping method based on quantitative semi-supervised learning Download PDFInfo
- Publication number
- CN116385593A CN116385593A CN202310309520.XA CN202310309520A CN116385593A CN 116385593 A CN116385593 A CN 116385593A CN 202310309520 A CN202310309520 A CN 202310309520A CN 116385593 A CN116385593 A CN 116385593A
- Authority
- CN
- China
- Prior art keywords
- spectrum
- mineral
- pixel
- learner
- ground
- 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
- 229910052500 inorganic mineral Inorganic materials 0.000 title claims abstract description 94
- 239000011707 mineral Substances 0.000 title claims abstract description 94
- 238000013507 mapping Methods 0.000 title claims abstract description 30
- 238000000034 method Methods 0.000 title claims abstract description 26
- 238000001228 spectrum Methods 0.000 claims abstract description 101
- 238000011160 research Methods 0.000 claims abstract description 19
- 238000005516 engineering process Methods 0.000 claims abstract description 8
- 230000003595 spectral effect Effects 0.000 claims description 22
- 238000012549 training Methods 0.000 claims description 14
- YGANSGVIUGARFR-UHFFFAOYSA-N dipotassium dioxosilane oxo(oxoalumanyloxy)alumane oxygen(2-) Chemical compound [O--].[K+].[K+].O=[Si]=O.O=[Al]O[Al]=O YGANSGVIUGARFR-UHFFFAOYSA-N 0.000 claims description 8
- 229910052627 muscovite Inorganic materials 0.000 claims description 8
- 229910021532 Calcite Inorganic materials 0.000 claims description 7
- 229910001919 chlorite Inorganic materials 0.000 claims description 7
- 229910052619 chlorite group Inorganic materials 0.000 claims description 7
- QBWCMBCROVPCKQ-UHFFFAOYSA-N chlorous acid Chemical compound OCl=O QBWCMBCROVPCKQ-UHFFFAOYSA-N 0.000 claims description 7
- 229910052900 illite Inorganic materials 0.000 claims description 5
- VGIBGUSAECPPNB-UHFFFAOYSA-L nonaaluminum;magnesium;tripotassium;1,3-dioxido-2,4,5-trioxa-1,3-disilabicyclo[1.1.1]pentane;iron(2+);oxygen(2-);fluoride;hydroxide Chemical compound [OH-].[O-2].[O-2].[O-2].[O-2].[O-2].[F-].[Mg+2].[Al+3].[Al+3].[Al+3].[Al+3].[Al+3].[Al+3].[Al+3].[Al+3].[Al+3].[K+].[K+].[K+].[Fe+2].O1[Si]2([O-])O[Si]1([O-])O2.O1[Si]2([O-])O[Si]1([O-])O2.O1[Si]2([O-])O[Si]1([O-])O2.O1[Si]2([O-])O[Si]1([O-])O2.O1[Si]2([O-])O[Si]1([O-])O2.O1[Si]2([O-])O[Si]1([O-])O2.O1[Si]2([O-])O[Si]1([O-])O2 VGIBGUSAECPPNB-UHFFFAOYSA-L 0.000 claims description 5
- 238000012360 testing method Methods 0.000 claims description 5
- NLYAJNPCOHFWQQ-UHFFFAOYSA-N kaolin Chemical compound O.O.O=[Al]O[Si](=O)O[Si](=O)O[Al]=O NLYAJNPCOHFWQQ-UHFFFAOYSA-N 0.000 claims description 4
- 229910052622 kaolinite Inorganic materials 0.000 claims description 4
- 229910021646 siderite Inorganic materials 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000012937 correction Methods 0.000 claims description 3
- 238000000354 decomposition reaction Methods 0.000 claims description 3
- 229910000514 dolomite Inorganic materials 0.000 claims description 3
- 239000010459 dolomite Substances 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 229910052674 natrolite Inorganic materials 0.000 claims description 2
- 239000002131 composite material Substances 0.000 claims 1
- 230000004075 alteration Effects 0.000 abstract description 5
- 230000004927 fusion Effects 0.000 abstract description 2
- 239000011435 rock Substances 0.000 description 10
- 230000008569 process Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 4
- 238000005065 mining Methods 0.000 description 4
- 238000010521 absorption reaction Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 3
- 238000011835 investigation Methods 0.000 description 3
- DGAQECJNVWCQMB-PUAWFVPOSA-M Ilexoside XXIX Chemical compound C[C@@H]1CC[C@@]2(CC[C@@]3(C(=CC[C@H]4[C@]3(CC[C@@H]5[C@@]4(CC[C@@H](C5(C)C)OS(=O)(=O)[O-])C)C)[C@@H]2[C@]1(C)O)C)C(=O)O[C@H]6[C@@H]([C@H]([C@@H]([C@H](O6)CO)O)O)O.[Na+] DGAQECJNVWCQMB-PUAWFVPOSA-M 0.000 description 2
- 230000033558 biomineral tissue development Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- PCHJSUWPFVWCPO-UHFFFAOYSA-N gold Chemical compound [Au] PCHJSUWPFVWCPO-UHFFFAOYSA-N 0.000 description 2
- 229910052737 gold Inorganic materials 0.000 description 2
- 239000010931 gold Substances 0.000 description 2
- 239000010445 mica Substances 0.000 description 2
- 229910052618 mica group Inorganic materials 0.000 description 2
- 239000010453 quartz Substances 0.000 description 2
- VYPSYNLAJGMNEJ-UHFFFAOYSA-N silicon dioxide Inorganic materials O=[Si]=O VYPSYNLAJGMNEJ-UHFFFAOYSA-N 0.000 description 2
- 229910052708 sodium Inorganic materials 0.000 description 2
- 239000011734 sodium Substances 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- RYGMFSIKBFXOCR-UHFFFAOYSA-N Copper Chemical compound [Cu] RYGMFSIKBFXOCR-UHFFFAOYSA-N 0.000 description 1
- HCHKCACWOHOZIP-UHFFFAOYSA-N Zinc Chemical compound [Zn] HCHKCACWOHOZIP-UHFFFAOYSA-N 0.000 description 1
- 230000003213 activating effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 229910052802 copper Inorganic materials 0.000 description 1
- 239000010949 copper Substances 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 239000011133 lead Substances 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 229910052751 metal Inorganic materials 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 229910052683 pyrite Inorganic materials 0.000 description 1
- 239000011028 pyrite Substances 0.000 description 1
- NIFIFKQPDTWWGU-UHFFFAOYSA-N pyrite Chemical compound [Fe+2].[S-][S-] NIFIFKQPDTWWGU-UHFFFAOYSA-N 0.000 description 1
- 238000002310 reflectometry Methods 0.000 description 1
- 238000005475 siliconizing Methods 0.000 description 1
- 239000000758 substrate Substances 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- 229910052725 zinc Inorganic materials 0.000 description 1
- 239000011701 zinc Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T11/00—2D [Two Dimensional] image generation
- G06T11/40—Filling a planar surface by adding surface attributes, e.g. colour or texture
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/778—Active pattern-learning, e.g. online learning of image or video features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
- Y02A40/10—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in agriculture
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Investigating, Analyzing Materials By Fluorescence Or Luminescence (AREA)
Abstract
The invention relates to application of fusion ground-pixel spectrum in the field of mineral mapping, in particular to a hyperspectral remote sensing mineral mapping method based on quantitative semi-supervised learning. Identifying a ground mineral sample by extracting the abundance and end members of the main alteration mineral; and judging each pixel in the image based on the semi-supervised learning model, covering the coordinate range of the research area, and judging whether the current pixel is an altered mineral or not by taking the comprehensive spectrum angle value as a quantitative index to finish mineral mapping. The invention provides feasibility for identifying similar mineral spectrum characteristic curves, and improves the reliability of mineral mapping and the portability of related technologies.
Description
Technical Field
The invention relates to application of fusion ground-pixel spectrum in the field of mineral mapping, in particular to a hyperspectral remote sensing mineral mapping method based on quantitative semi-supervised learning.
Background
The mineral mapping is based on mineral identification, and the regional mineral category distribution is understood, so that the method has an important effect on analysis of altered mineral causes. The remote sensing technology actively or passively receives electromagnetic waves reflected by ground objects through the sensor to detect the remote target, and has the characteristics of wide observation range, abundant information quantity, high intelligent processing degree and the like, so that the remote sensing technology is preliminarily applied to the field of mineral mapping. However, the traditional high-resolution multi-spectrum remote sensing technology has low spectral resolution, and accurate description of mineral categories through a feature model is difficult.
The hyperspectral remote sensing utilizes continuous spectral characteristics, so that weak information differences among different components belonging to the same type of ground objects can be well distinguished, and a better theoretical and technical basis is provided for the application of the hyperspectral remote sensing in mineral mapping. However, analysis of mineral distribution throughout the study area relies mainly on costly, time-consuming and laborious geological investigation, and it is difficult to meet regional mineral mapping requirements. Semi-supervised learning is to analyze the mineral category of a research area by learning the ground spectrum of a mineral sample and combining hyperspectral remote sensing images. The spectral angle chart is used as a common method for judging spectral characteristics, and the pixel spectral attribute is analyzed based on the spectral angle value. However, when the spectral angle values lie in the critical range, it is difficult for a single learner to accurately determine their mineral class attributes. According to the invention, the spectrum angle is used as a quantitative index to analyze the mineral category of the pixel spectrum, a semi-supervised learning model is built by combining a plurality of learners to analyze the mineral distribution of a research area, and the mineral identification precision is further improved.
Disclosure of Invention
The invention provides a hyperspectral remote sensing mineral mapping method based on quantitative semi-supervised learning, which is characterized in that ground mineral samples are identified by extracting abundance and end members of main alteration minerals; and judging each pixel in the image based on semi-supervised learning, covering the range of a research area, and judging whether the current pixel is an altered mineral or not by taking the comprehensive spectrum angle value as a quantitative index to finish mineral mapping. The invention provides feasibility for identifying similar mineral spectrum characteristic curves, and improves the reliability of mineral mapping and the portability of related technologies.
According to an aspect of the embodiments, there is provided a hyperspectral remote sensing mineral mapping method based on quantitative semi-supervised learning, including:
step 1, testing a ground mineral sample by using a portable spectrometer to obtain a ground spectrum, and extracting abundance and end members of the mineral sample;
step 2, acquiring a hyperspectral remote sensing image, and extracting a pixel spectrum;
step 3, dividing the ground spectrum into two parts according to the wavelength, and respectively inputting the two parts into a learner 1 and a learner 2 of a semi-supervised learning model based on a three-training algorithm;
step 4, performing pixel-by-pixel traversal on the hyperspectral remote sensing image, and inputting the current pixel spectrum to a learner 3 of the semi-supervised learning model;
step 5, calculating the spectrum angles alpha of the semi-supervised learning models learner 1, learner 2 and learner 3 based on the wavelet spectrum angle drawing respectively 1 、α 2 、α 3 ；
Step 6, spectrum Angle α 1 、α 2 When the values are close, the comprehensive spectrum angles alpha of the learner 1, the learner 2 and the learner 3 are calculated, and the learner 3 is activated to judge the current pixel spectrum;
and 7, if the comprehensive spectrum angle alpha meets the threshold value of the current changed mineral, the pixel is represented by the corresponding changed mineral, otherwise, the pixel is identified as other changed minerals or non-mineral pixels.
And (5) repeating the steps 5 to 7 until all pixels in the image are traversed, and carrying out mineral mapping on the research area.
In the steps 1 and 2, the ground spectrum is collected by adopting an SVC HR1024 ground object spectrometer, the image of the research area is collected by adopting a CASI/SASI platform, and the ground spectrum and the pixel spectrum are collected in the same period so as to avoid local time difference.
In the step 2, the pixel spectrum is obtained by performing atmospheric correction on the image of the research area and decomposing the spectrum containing a plurality of alteration minerals in the image by adopting a mixed pixel decomposition technology.
In the steps 3 and 4, the ground spectrum is divided into two parts according to the wavelength of 950nm-1700nm and 1700nm-2450nm and is input into two learners, and the pixel spectrum with the wavelength range of 950nm-2450nm is input into a third learner.
The ground spectrum and the pixel spectrum have the same wavelength variation range, the ground spectrum has higher spectrum resolution, the ground spectrum and the pixel spectrum are compared one by one, and the ground spectrum and the pixel spectrum keep the same wave band number and sampling interval through wavelength downsampling.
The invention has the beneficial effects that:
(1) According to the invention, the three training algorithms are quantitatively evaluated by using the spectrum angle values, and the ground spectrum of a small amount of mineral samples is trained and pixel spectra are judged one by one, so that the mineral identification precision is further improved;
(2) According to the invention, the wavelet function is integrated into the calculation process of the spectral angle chart, and the difference between spectrums is judged by combining two different scales, so that the tiny differences between absorption characteristics of different changed minerals are effectively distinguished;
(3) The invention combines multisource hyperspectral remote sensing data, fully exerts the advantages of high resolution of a ground spectrometer and wide detection range of an onboard sensor, and has higher portability in the field of mineral mapping.
Drawings
In order to more clearly illustrate the technical solution of the embodiments of the present invention, the drawings of the embodiments will be briefly described below.
Fig. 1 is a flowchart of a hyperspectral remote sensing mineral mapping method based on quantitative semi-supervised learning according to an embodiment of the present invention.
Fig. 2 is a quantitative semi-supervised learning implementation process according to an embodiment of the present invention.
FIG. 3 is a diagram of hyperspectral remote sensing images and mineral mapping results of an investigation region according to an embodiment of the present invention.
Detailed Description
The method for filling the hyperspectral remote sensing minerals based on quantitative semi-supervised learning is described in detail by taking the German-Youxi-Yongtai mining area as a research area.
germany-ulide-Yongtai mining area is located in the city of spring, fujian province at the intersection of the mountain macroloop construction of the coastal dycloud mountain (northwest) in southeast China. The regional dense magma and hydrothermal mineralization are controlled by these faults and their underlying structures. The stratum mainly belongs to metamorphic rock and mesogenic volcanic rock from ancient to mesogenic. The research area has abundant metal mineral resources such as gold, copper, lead, zinc and the like. The main types of deposits are volcanic-hypovolcanic-pyro-hydraulic, metamorphic clastic-rock-pyro-hydraulic and structural-pyro-hydraulic. Bare metamorphic substrates and mesogenic volcanic rocks are typically of higher abundance. The rock pulp invades frequently, mineralization is closely related to shallow low-temperature liquid invaded into rock, and the rock pulp is often in an external contact zone of green-curtain rock, binary-long rock, granite-long rock and square-long rock. The main mineral alteration types include siliconizing (quartz), sericite (sericite, illite), sliming (kaolinite), chlorite petrochemical (chlorite), and partial calcite (calcite).
Fig. 1 shows a flow chart of a hyperspectral remote sensing mineral mapping method based on quantitative semi-supervised learning. The method shown in fig. 1 will be described in detail.
(1) Obtaining ground spectrum, extracting abundance and end member
And testing the ground mineral sample by using a portable ground object spectrometer to obtain a ground spectrum, and extracting the abundance and end members of the main altered mineral sample. The ground spectrum can be acquired by an SVC HR1024 ground object spectrometer. The end members and abundance representing the percentage of the main altered mineral in the spectrum were obtained by geological software.
(2) Acquiring hyperspectral remote sensing images and extracting pixel spectra
The ground spectrum and the pixel spectrum are collected in the same period to avoid local time difference. And acquiring hyperspectral remote sensing images based on a CASI/SASI platform, wherein the acquired images comprise 100 wave bands, and the spectrum coverage range is 950nm to 2450nm. Atmospheric corrections were made using the fast line-of-sight atmospheric analysis model in ENVI 5.3 software. And decomposing the spectrum containing a plurality of changed minerals in the image by adopting a mixed pixel decomposition technology to obtain a pixel spectrum.
(3) Acquiring pixel spectra of nine changed minerals
Spectral signatures containing kaolinite, chlorite, muscovite, calcite, illite, natrolite, dolomite, siderite and polysilicium muscovite are extracted according to abundance and altered mineral coordinates, the signatures for each wavelength representing the reflectivity between the sensor and the object.
(4) Based on three training algorithms, a semi-supervised learning model is constructed
And training by using the ground spectrum, inputting the ground spectrum into the first two learners of the three training algorithms according to the wavelength range, and respectively calculating the spectrum angle to assist the third learner to judge. And judging whether the pixel is an altered mineral or not by taking the difference value of the two spectrum angles as a standard, and determining the category of the pixel. Wherein, the ground spectrum is divided into two parts according to the wavelength of 950nm-1700nm and 1700nm-2450nm and is input into the first two learners of the three training algorithm, and the pixel spectrum with the wavelength range of 950nm-2450nm is input into the third learner.
The training process can utilize most of the effective spectral features due to the averaging calculation involved, making the recognition performance more robust. This shows the performance of semi-supervised learning on small sample data, with the spectral angle as the judgment index, no complex voting is required in the training process, and the quantitative index has a clearer judgment boundary.
(5) Training by using ground spectrum and traversing pixel spectrum for testing
The ground spectrum and the pixel spectrum have the same wavelength variation range, the ground spectrum has higher spectrum resolution, the ground spectrum and the pixel spectrum are compared one by one, and the ground spectrum and the pixel spectrum keep the same wave band number and sampling interval through wavelength downsampling. The ground spectrum is input into two learners for training, pixel-by-pixel traversal is carried out on the images of the research area, and the current pixel spectrum is input into a third learner.
(6) Calculating the spectral angle of an independent learner
The wavelet spectral angle drawing is to use a wavelet function as a kernel function for calculating a spectral angle value, judge the difference between spectrums by combining two different scales, and effectively distinguish the tiny difference between absorption characteristics of different alteration minerals. Thus, the angle between spectrum a and spectrum b is defined as follows:
(7) Calculating the integrated spectral angle
Comprehensive spectrum angles are calculated based on wavelet spectrum angle drawing, and the spectrum angles of three learners are respectively defined as alpha 1 、α 2 、α 3 . If the spectrum angle alpha 1 、α 2 And (3) calculating the comprehensive spectrum angle alpha of the three learners when the values are close, and activating a third learner to judge the current pixel spectrum.
Based on the three training algorithm, the comprehensive spectrum angle alpha formula of the three learners is calculated as follows:
where α (x) is the integrated spectral angle between the current pixel and some kind of altered mineral. When the integrated spectral angle α satisfies the value of the altered mineral, the pixel is represented by the corresponding altered mineral, otherwise it is considered as other altered mineral or non-mineral pixel.
And traversing all pixels in the image, and carrying out mineral mapping on the research area.
(8) Mineral identification accuracy
A surface sample is collected and its mineral properties are determined by spectral absorption characteristics. The identification accuracy of each altered mineral is calculated by dividing the number of correctly identified picture elements by the total number of relevant picture elements, the total number of picture elements involved consisting of correctly identified picture elements, incorrectly identified picture elements and unidentified picture elements. Experimental results indicate that the research area has a large amount of kaolinite, chlorite, muscovite, calcite, illite and a small amount of sodium mica, dolomite, siderite and polysilicic muscovite. In addition, the wavelet function can improve the distinguishing capability of mineral samples, provides feasibility for identifying mineral samples with similar spectrum curves, and combines a three-training algorithm to judge the pixel spectrum in the image by using a small amount of ground spectrum training. According to the data in Table 1, the overall recognition accuracy of nine kinds of altered minerals reaches 82%, and more than 92% of muscovite and calcite are correctly recognized, and the experimental results have higher consistency with the relevant geological survey data.
TABLE 1 identification accuracy of nine kinds of altered minerals
(9) Mineral mapping test
Due to the mining process, deposits and weathering, a range of mines are widely distributed, and a large number of altered minerals, such as muscovite, sodium mica, calcite, etc., are found in research areas. As the main mineralisation area in the middle of the fowler's work, its construction activities are based on faults with complex mechanical properties, usually associated with the causes of quartz and pyrite, a series of polysilicium muscovite and siderite are found in the relevant areas. According to historical data, mining industry is used as a main economic source of research areas, lithology mainly comprises the development of chlorite petrochemical industry and sericite in a gold-containing fracture zone, and a small amount of chlorite and illite are found in the fracture zone. As shown in fig. 3, the mapping result accords with geological conditions and an ore forming mechanism of a research area, has higher identification precision on most minerals, and has certain application value. Note that fig. 3 can be used only in color, and a black-and-white chart cannot effectively reflect the mineral category distribution.
In the embodiment, the German-Youxi-Yongtai is taken as a research area, and a hyperspectral remote sensing mineral mapping method based on quantitative semi-supervised learning is provided. Proved by experimental verification, the method is used for distinguishing the category of the changed minerals, and has higher identification precision. By combining the ground and pixel spectrum, the mineral distribution of the research area can be integrally analyzed, so that a great deal of time-consuming and labor-consuming geological investigation work is avoided, and the method has good adaptability to other research areas.
Claims (6)
1. The hyperspectral remote sensing mineral mapping method based on quantitative semi-supervised learning is characterized by comprising the following steps of:
testing a ground mineral sample by using a portable spectrometer to obtain a ground spectrum, and extracting abundance and end members of the mineral sample;
acquiring a hyperspectral remote sensing image and extracting a pixel spectrum;
dividing the ground spectrum into two parts according to the wavelength, and respectively inputting the two parts into a learner 1 and a learner 2 of a semi-supervised learning model based on a three-training algorithm;
performing pixel-by-pixel traversal on the hyperspectral remote sensing image, and inputting the current pixel spectrum into a learner 3 of a semi-supervised learning model;
semi-supervised learning model learner 1, learner 2 and learner 3 based on wavelet spectral angle drawing calculation 1 、α 2 、α 3 ；
Spectral angle alpha 1 、α 2 When the values are close, the comprehensive spectrum angles alpha of the learner 1, the learner 2 and the learner 3 are calculated, and the learner 3 is activated to judge the current pixel spectrum;
the composite spectral angle α satisfies the threshold value of the current altered mineral, then that pixel is represented by the corresponding altered mineral, otherwise it is considered as other altered mineral or non-mineral pixel.
2. The hyperspectral remote sensing mineral mapping method based on quantitative semi-supervised learning as recited in claim 1, wherein the ground spectrum and the pixel spectrum are acquired at the same time period to avoid local moveout.
3. The hyperspectral remote sensing mineral mapping method based on quantitative semi-supervised learning as recited in claim 1, wherein the ground spectrum has the same wavelength variation range as the pixel spectrum, the ground spectrum has higher spectral resolution, the ground spectrum is compared with the wavelength of the pixel spectrum one by one, and the ground spectrum and the pixel spectrum keep the same wave band number and sampling interval through wavelength downsampling.
4. The hyperspectral remote sensing mineral mapping method based on quantitative semi-supervised learning as set forth in claim 1, wherein the ground spectrum is divided into two parts according to the wavelengths of 950nm-1700nm and 1700nm-2450nm, and is input to two learners, and the pixel spectrum with the wavelength range of 950nm-2450nm is input to a third learner.
5. The hyperspectral remote sensing mineral mapping method based on quantitative semi-supervised learning as set forth in claim 1, wherein the pixel spectrum is obtained by performing atmospheric correction on an image of a research area and decomposing spectrums containing a plurality of changed minerals in the hyperspectral remote sensing image by adopting a mixed pixel decomposition technology.
6. The method of claim 5, wherein the altered mineral comprises one or more of kaolinite, chlorite, muscovite, calcite, illite, natrolite, dolomite, siderite, and polysilicite.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310309520.XA CN116385593A (en) | 2023-03-24 | 2023-03-24 | Hyperspectral remote sensing mineral mapping method based on quantitative semi-supervised learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310309520.XA CN116385593A (en) | 2023-03-24 | 2023-03-24 | Hyperspectral remote sensing mineral mapping method based on quantitative semi-supervised learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116385593A true CN116385593A (en) | 2023-07-04 |
Family
ID=86960882
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310309520.XA Pending CN116385593A (en) | 2023-03-24 | 2023-03-24 | Hyperspectral remote sensing mineral mapping method based on quantitative semi-supervised learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116385593A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117422991A (en) * | 2023-10-07 | 2024-01-19 | 中色地科蓝天矿产（北京）有限公司 | Intelligent mine detection system and method based on big data and readable storage medium |
-
2023
- 2023-03-24 CN CN202310309520.XA patent/CN116385593A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117422991A (en) * | 2023-10-07 | 2024-01-19 | 中色地科蓝天矿产（北京）有限公司 | Intelligent mine detection system and method based on big data and readable storage medium |
| CN117422991B (en) * | 2023-10-07 | 2024-06-04 | 中色地科蓝天矿产（北京）有限公司 | Intelligent mine detection system and method based on big data and readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Hosseinjani et al. | Mapping alteration minerals using sub-pixel unmixing of ASTER data in the Sarduiyeh area, SE Kerman, Iran | |
| Tangestani et al. | Evaluating Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) data for alteration zone enhancement in a semi‐arid area, northern Shahr‐e‐Babak, SE Iran | |
| Molan et al. | Mineral mapping in the Maherabad area, eastern Iran, using the HyMap remote sensing data | |
| CN109614942B (en) | Forest disturbance long-time sequence monitoring method based on cloud computing platform | |
| CN108171210B (en) | Method and system for extracting remote sensing abnormal information of alteration of covered area of planting | |
| Mahboob et al. | Mapping hydrothermal minerals using remotely sensed reflectance spectroscopy data from Landsat | |
| CN115761038B (en) | Tunnel face geological sketch method and system based on image spectrum technology | |
| CN116385593A (en) | Hyperspectral remote sensing mineral mapping method based on quantitative semi-supervised learning | |
| Ni et al. | Mineral identification and mapping by synthesis of hyperspectral VNIR/SWIR and multispectral TIR remotely sensed data with different classifiers | |
| Chen et al. | Hydrocarbon micro-seepage detection by altered minerals mapping from airborne hyper-spectral data in Xifeng Oilfield, China | |
| Omran | Rapid prediction of soil mineralogy using imaging spectroscopy | |
| CN109738369A (en) | A kind of archaeology detection method using Airborne Hyperspectral remote sensing jadeware | |
| Wang et al. | Altered mineral mapping based on ground-airborne hyperspectral data and wavelet spectral angle mapper tri-training model: Case studies from Dehua-Youxi-Yongtai Ore District, Central Fujian, China | |
| CN109145881B (en) | Remote sensing image paste salt information extraction method and device | |
| Ghorbani et al. | Regional scale prospecting for non-sulphide zinc deposits using ASTER data and different spectral processing methods | |
| Kokaly | Spectroscopic remote sensing for material identification, vegetation characterization, and mapping | |
| Canbaz et al. | Hydrothermal alteration mapping using EO-1 Hyperion hyperspectral data in Kösedağ, Central-Eastern Anatolia (Sivas-Turkey) | |
| Agus | Mapping white mica in milled porphyry copper pebbles using hyperspectral imagery: an exploratory study | |
| Wolfe et al. | Hyperspectral analytics in envi target detection and spectral mapping methods | |
| Quental et al. | Identification of materials related to acid mine drainage using multi-source spectra at S. Domingos Mine, southeast Portugal | |
| Wang et al. | Rocks/minerals information extraction from EO-1 Hyperion data base on SVM | |
| Wolfe et al. | Hyperspectral analytics in ENVI | |
| Merrill-Cifuentes et al. | Unsupervised textural classification of rocks in large imagery datasets | |
| Hajaj et al. | A review on hyperspectral imagery application for lithological mapping and mineral prospecting: Machine learning techniques and future prospects | |
| Kumar | Developing innovative spectral and machine learning methods for mineral and lithological classification using multi-sensor datasets |
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 |