CN106971156A - Rare earth mining area remote sensing information extraction method based on object-oriented classification - Google Patents
Rare earth mining area remote sensing information extraction method based on object-oriented classification Download PDFInfo
- Publication number
- CN106971156A CN106971156A CN201710173837.XA CN201710173837A CN106971156A CN 106971156 A CN106971156 A CN 106971156A CN 201710173837 A CN201710173837 A CN 201710173837A CN 106971156 A CN106971156 A CN 106971156A
- Authority
- CN
- China
- Prior art keywords
- rule
- rare earth
- classification
- area
- spectrum
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30181—Earth observation
- G06T2207/30184—Infrastructure
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/194—Terrestrial scenes using hyperspectral data, i.e. more or other wavelengths than RGB
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Astronomy & Astrophysics (AREA)
- Remote Sensing (AREA)
- Multimedia (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The invention provides a rare earth mining area remote sensing information extraction method based on object-oriented classification, which is characterized in that the object-oriented classification method is applied to extract the remote sensing information of the rare earth mining area; and the extraction result of the object-oriented classification method is compared and analyzed with the extraction result of the traditional supervision classification method, the feasibility and the superiority of the object-oriented classification method in the extraction of the rare earth mine information are discussed, and a foundation is laid for the follow-up researches such as the current rare earth investigation and the dynamic monitoring. The method of the invention comprises the following steps: 1) firstly, preprocessing a remote sensing data source of a selected research area, and then, carrying out image segmentation by adopting an edge-based segmentation algorithm; 2) then, a series of rules are established by combining topographic information, spectral information and geometric information to realize feature extraction; 3) and finally, carrying out object-oriented classification and extracting the rare earth mining area of the research area.
Description
Technical field
It is more particularly to a kind of based on the dilute of object oriented classification the invention belongs to geological and mineral the Study on Resources technical field
Native mining area's remote sensing information extracting method.
Background technology
Classification of remote-sensing images is a kind of method that image information is extracted, and is also one of widest field of remote sensing application.Pass
The image classification method of system mainly includes supervised classification and unsupervised classification method, and this foundational development fuzzy classifier method,
Support vector cassification method and Decision tree classification etc..But these methods are mainly built upon the classification on pixel level, not
There are the information such as space, the texture of consideration object.In recent years, high-resolution remote sensing image technology is continued to develop, traditional classification
Method can not meet the classificating requirement of high-resolution remote sensing image.In order to fully excavate high resolution image information, towards
Object basis arises at the historic moment and is gradually widely used.From unlike conventional sorting methods, object oriented classification method point
In class process, processing unit is not single pixel, but the pixel set with same or similar homogeneous feature, i.e. homogeneous
Object.The foundation of classification in addition to spectral information, in addition to spatial relationship between object size and shape, object and object etc.
Level attribute etc..When being classified by object oriented classification method to high-resolution remote sensing image, it can avoid conventional based on picture
" spiced salt " phenomenon caused by first sorting technique, obtains higher precision.
Ion adsorption type rare earth ore is a kind of Novel rare-earth ore found in China Jiangxi in 1969, and the ore deposit has reserves rich
The rich, feature such as radioactivity is low, rare earth partition is complete, exploitation is simple, is the mineral resources that China shows unique characteristics.In recent years, rare earth
The effect that resource is played in economic, society is increasingly notable, and demand is continuously increased, it has also become important strategic resources.State
Inside and outside scholar has carried out certain research for Rare-earth Mine information extraction, makes some progress, such as Sun Yaping, Lei Guojing
Jiangxi Area rare-earth mining area typical feature is classified Deng using Maximum likelihood classification, carries out dynamic monitoring research；Dai Jing
Crystalline substance carries out dynamic monitoring research using Spectral angle mapper classification to the regional rare-earth mining area in Xunwu of Jiangxi；Zhang Hang uses two points of pixel
Method, vegetation coverage is extracted with reference to TM/ETM+ and HJ-1/CCD data；After Li Hengkai uses fractal texture subsidiary classification to classification
Comparison method is improved, and two kinds of points of shape windows of separable index and J-M distances and waveband selection model is constructed, to ridge backlands area
Carry out dynamic monitoring in rare-earth mining area.There is presently no use object oriented classification method to rare earth mining area remote sensing information extracting method
The report of research.
The content of the invention
The purpose of the present invention is to propose to a kind of rare earth mining area remote sensing information extracting method based on object oriented classification, with
It is distant with reference to IKONOS high-resolution using ENVI Feature Extraction softwares as platform exemplified by Xunwu of Jiangxi rare-earth mining area
Data are felt, using object-oriented classification method Extraction of rare earth mining area remote sensing information；The present invention will also apply object oriented classification
The extraction result of method and the extraction result of traditional supervised classification are analyzed, and inquire into object oriented classification method in rare earth
Feasibility and dominance in mine information extraction, are that the follow-up studies such as rare earth on-site investigation, dynamic monitoring lay the foundation.
The technical scheme is that：
1. a kind of rare earth mining area remote sensing information extracting method based on object oriented classification, it is characterised in that including with
Lower step：1) after the remotely-sensed data source in the research area of selection is pre-processed first, entered using the partitioning algorithm based on edge
Row Image Segmentation；2) series of rules is set up then in conjunction with terrain information, spectral information and geological information, realizes feature extraction；
3) classification of object-oriented is finally carried out, the rare earth mining area in research area is extracted.
2. described in remotely-sensed data source remotely-sensed data be on July 24th, 2011 IKONOS image datas, including light more than 4
Wave band (blue, green, red, near-infrared) and 1 panchromatic wave-band are composed, wherein multispectral resolution rate is 4m, and panchromatic wave-band is 1m.
3. described in pretreatment include by the remotely-sensed data carry out atmospheric correction, ortho-rectification, image co-registration, geometric correction
A series of images processing work.
4. described in the partitioning algorithm based on edge refer to the sharp point detected is linked to be into wheel according to certain rule
Exterior feature, constitutes cut zone；During Image Segmentation, it is first determined segmentation yardstick, initial segmentation is carried out to image；It is then determined that
Merger yardstick so that same or analogous object merger is a class, reduces the number of objects of initial segmentation, further corrects object
Partitioning boundary.
5. described in segmentation range scale be 0-100, select as 50；The merger range scale is 0-100, is selected as 90.
6. described in step 2) in, it is described set up rule refer to according between imaged object feature and atural object relation set up one
Set of constraints condition, including geometrical rule, spectrum rule, landform rule；The geometrical rule refers to according to the several of imaged object
The rule what feature is set up；The spectrum rule refers to the atural object that different atural objects are recognized by the difference of electromagnetic spectrum feature
Feature, the rule that spectrum and reflectivity are set up as characters of ground object；The landform rule refers to that the numeral in research on utilization area is high
Journey model (DEM) data, a series of landform characterization factors of the description topography and landform character calculated and the rule set up.
7. described in the landform rule feature factor include elevation, the gradient, earth's surface depth of cut, topographic relief amplitude；The spectrum
The rule feature factor includes normalized differential vegetation index, the reflected value of atural object, the curve of spectrum tendency of atural object；The geometrical rule is special
Levy including imaged object area.
8. described in step 3) in, using membership function method carry out object-oriented classification, built using membership function
Rule set, landform rule, spectrum rule and geometrical rule are brought into rule set, different weights are assigned respectively, are entered
Row classification, draws the classification results in rare earth mining area.
The technique effect of the present invention：
A kind of rare earth mining area remote sensing information extracting method based on object oriented classification proposed by the present invention, is sought with Jiangxi
Black rare-earth mining area is research area, using IKONOS images as data source, using ENVI Feature Extraction softwares as platform,
Using object-oriented classification method Extraction of rare earth mining area, the also extraction result with traditional supervised classification is analyzed,
Feasibility and dominance of the object oriented classification method in Rare-earth Mine information extraction are inquired into, is rare earth on-site investigation, dynamic prison
The follow-up studies such as survey lay the foundation.
The characteristics of present invention is directed to rare earth mining area remote sensing information, selects the partitioning algorithm based on edge to carry out image point
Cut；Combining with terrain information, spectral information and geological information set up rule set, carry out feature extraction；Finally use membership function
Method realizes object oriented classification, it was demonstrated that the feasibility of object oriented classification method Extraction of rare earth mining area, it is possible to obtain following
Conclusion：
(1) nicety of grading of object oriented classification method is significantly improved, compared with traditional supervised classification, such as with tradition
Spectral angle mapper classification be analyzed, as a result show, object oriented classification method Extraction of rare earth mining area of the invention
Overall accuracy has reached that 92.49%, Kappa coefficients are 0.8576, than the nicety of grading of traditional supervised classification Spectral angle mapper method
It is high by about 10%, show object oriented classification method Extraction of rare earth mining area more advantage.
(2) Image Segmentation is the basis of object oriented classification method, therefore, and the selection for splitting yardstick is particularly significant, segmentation knot
The quality of fruit will directly influence the precision of classification results.If segmentation yardstick is too small, image fragmentation degree height is easily caused, point
The broken spot of class result is serious, if otherwise segmentation yardstick is excessive, it is indefinite to easily cause atural object border, " mixed point " phenomenon occurs, because
This present invention is during Image Segmentation is carried out, it is considered to the characteristics of type of ground objects, reasonable selection segmentation yardstick, improves towards right
As the nicety of grading of classification.
(3) present invention not only make use of spectral information, also by landform with object oriented classification method Extraction of rare earth mining area
Geomorphologic conditions, geometric space relation also serve as classification indicators, have taken into full account the spectrum of atural object, texture, several how many
Information, sets up geometrical rule, spectrum rule and landform rule, and formation rule collection reduces " the different spectrum of jljl " in assorting process
" same object different images " phenomenon is produced, it is to avoid " spiced salt " phenomenon occurs caused by being based on pixel sorting technique.
Brief description of the drawings
Fig. 1 is the Technology Roadmap of the method for the present invention.
Fig. 2 is research area geographical position figure of the invention.
Fig. 3 is Image Segmentation result figure of the invention.
Fig. 4 a meet elevation features because of subconditional binary map for research area.
Fig. 4 b meet the binary map of gradient characterization factor condition for research area.
Fig. 4 c meet the binary map of earth's surface depth of cut characterization factor condition for research area.
Fig. 4 d meet the binary map of topographic relief amplitude characterization factor condition for research area.
Fig. 5 is rare earth mining area of the research area result figure based on object oriented classification method.
Fig. 6 is rare earth mining area of the research area result figure based on supervised classification.
Reference lists as follows：1- mining areas, 2- buildings, 3- vegetation.
Embodiment
Embodiments of the invention are described in further detail below in conjunction with accompanying drawing.
As shown in figure 1, the Technology Roadmap of the method for the present invention.A kind of rare earth mining area based on object oriented classification
Remote sensing information extracting method, comprises the following steps：1) after the remotely-sensed data source in the research area of selection is pre-processed first, adopt
Image Segmentation is carried out with the partitioning algorithm based on edge；2) one is set up then in conjunction with terrain information, spectral information and geological information
Series rule, realizes feature extraction；3) classification of object-oriented is finally carried out, the rare earth mining area in research area is extracted.
As shown in Fig. 2 being the geographical position figure in the research area of selection.The research area of selection is located at Jiangxi Province Xunwu County, face
About 100 sq-kms of product, research area's tectonics is located at South China Caledonian orogenic belt eastern section, and the stratum of exposure mainly includes late dwarf
Sieve world first separate unit granite porphyry and chicken coop screen-like mountain peak group flow liner matter Porphyroclastic lava, dacite etc..Study the granite warp in area
Slacking, which forms the weathering crust that area is wide, thickness is big, be very beneficial to rare earth element is enriched with to form ion adsorption type rare earth ore.
Xunwu Rare Earth Mine is typical low yttrium cerium-rich rare earth ore deposit, be also in the weathered superficial leaching rare-earth ore bed having now been found that it is unique with
Rare earth mineral deposit based on light rare earth.
On step 1) in data source and pretreatment：Remotely-sensed data used herein is on July 24th, 2011
IKONOS image datas, including 4 multi light spectrum hands (blue, green, red, near-infrared) and 1 panchromatic wave-band, wherein multispectral resolution
Rate is 4m, and panchromatic wave-band is 1m.Because the data used are 1A DBMSs, it is therefore desirable to pre-processed, including atmospheric correction,
The a series of images processing work such as ortho-rectification, image co-registration, geometric correction, is object-oriented point to improve the quality of image
Data basis is established in class research.
Atmospheric correction：In remote sensing image imaging process, due to sunshine in transmitting procedure by atmospheric molecule, aerosol
The influence of scattering and absorption with cloud particle etc., the atural object amount of radiation that sensor is received with certain non-targeted atural object into
As information.Therefore, in order to eliminate influence of the air to remote sensing image, it is necessary to carry out atmospheric correction.Atmospheric correction is most commonly used that
Equation of radiative transfer method, mainly uses electromagnetic wave atmospheric radiative transfer principle, sets up atmospheric correction models, such as 6S models,
LOWTRAN models, MORTRAN models, FLAASH models etc..Atmospheric correction is carried out using FLAASH models herein.
Ortho-rectification：Satellite image is in imaging process, because hypsography can cause each picture point in image to produce difference
The geometry deformation of degree and distortion, therefore in order to eliminate topographic projection's difference firstly the need of carrying out ortho-rectification.Ortho-rectification is base
Direct transferred the RPC files of appearance rail data production and the dem data of research area's 30m resolution ratio in satellite, and RPC is used under ENVI platforms
Orthorectification Workflow instruments are realized.
Image co-registration：At present for panchromatic and multi light spectrum hands image co-registration method, mainly including principal component transform,
Brovey conversion, HIS conversion, product of transformation, Gram-Schmidt conversion and wavelet transformation etc..Pass through the experiment to fusion method
Analysis, final herein to use Gram-Schmidt Spectral Sharpening methods, the fusion method can be protected preferably
The colourity and saturation degree of former image are held, while all more prominent in spectral information and the aspect syncretizing effect of spatial information two.
Geometric correction：Geometric correction is drawn to eliminate sensor imaging mode, attitude of satellite change and flying height etc.
The geometric distortion risen.Geometric correction, as reference images, manually chooses geographical feature in image using Google Earth data
Obvious point, using quadratic polynomial matching algorithm, is corrected as ground control point to IKONOS data.The control of selection
Point is evenly distributed in image as far as possible, and control point is no less than 20, and by overall error control in 1 pixel.
Carry out after data prediction, object oriented classification process includes three steps：Image Segmentation, extracting object；Set up
Rule, feature extraction；Image classification.
Image Segmentation：Image Segmentation is, according to the heterogeneous standard of homogeney, to be divided into several mutual by whole imagery zone
There is same or analogous characteristic inside the process for the non-NULL subregion not overlapped, the same area.The dividing method commonly used at present
Substantially can be summarized as the partitioning algorithm based on region, Threshold Segmentation Algorithm, the partitioning algorithm based on border, based on active contour mould
Type partitioning algorithm etc..Image Segmentation is as the basis of object oriented classification method, and the selection for splitting yardstick is particularly significant, can direct shadow
Ring the precision of classification.
As shown in figure 3, being the Image Segmentation result figure of the present invention.Use the algorithm split based on edge, the algorithm
The sharp point detected is linked to be profile according to certain rule, so as to constitute cut zone.In cutting procedure, first
It is determined that segmentation yardstick, initial segmentation is carried out to image, image degree of crushing now is higher；It is then determined that merger yardstick so that
Same or analogous object merger is a class, reduces the number of objects of initial segmentation, further the partitioning boundary of amendment object.Through
Cross and test repeatedly, when it is 90 (scope 0-100) that to split yardstick, which be 50 (scope is 0-100), merges yardstick, can obtain preferable
Segmentation result, as shown in figure 3, carry out Image Segmentation after, mining area 1, building 2, vegetation 3 boundary profile it is clear.
After Image Segmentation, imaged object becomes information carrier, it is possible to obtain a series of related to imaged object
Information characteristics, the foundation rule refers to that, according to the relation between imaged object feature and atural object, a series of constraints can be set up
Condition, forms series of rules, so as to realize feature extraction.Feature extraction refers to according to the characteristics of each class categories, never
The various features that the category is different from other classifications are extracted with aspect.The process for setting up rule realizes feature extraction
Process.The rule set up herein mainly includes geometrical rule, spectrum rule and regular three classes of landform.Wherein, the geometry rule
Then refer to the rule set up according to the geometric properties of imaged object；The spectrum rule refers to the difference by electromagnetic spectrum feature
The different characters of ground object to recognize different atural objects, the rule that spectrum and reflectivity are set up as characters of ground object；The landform rule
Refer to digital elevation model (DEM) data in research on utilization area, a series of landform of the description topography and landform character calculated are special
The rule levied the factor and set up.
Geometrical rule：In Rare Earth Mine recovery process, pond leaching process and dump leaching method are required for peeling off the plant that mine surface is covered
Quilt and soil layer, therefore rare earth mining area shows as the massif exposed area of large area.And the region such as intermountain path, bare area near mining area
Due to the influence of manpower factor, also without vegetative coverage, wrong point phenomenon in assorting process is easily caused.After Image Segmentation, divided
The imaged object heterogeneity for being segmented into the same area reaches minimum, and contains a variety of same characteristic features that can be used for classifying, such as
Spectral Characteristic, shape, it is several how.Use geometric properties first herein, set up geometrical rule.ENVI Feature
The geometric properties attribute that Extraction is provided has AREA (area), LENGTH (length), SQLIDITY (integrity degree), RECT_
14 indexes such as FIT (rectangular degree), that mainly use herein is area AREA., will according to remote sensing image and field actual conditions
Area control is more than 1000m2In the range of, the influence of road, bare area etc. can be removed.
Spectrum rule：Different atural objects have the different spectral charactersiticss of targets in nature, therefore can be connect by sensor
The difference of the electromagnetic spectrum feature received recognizes different atural object, therefore spectrum and reflectivity can also be used as characters of ground object
Set up rule.The normalization for calculating image by the red wave band of remote sensing image and the clutter reflections rate of near infrared band first is planted
By index NDVI (NDVI=(NIR-R)/(NIR+R)), understand that the NDVI of vegetation is all higher than 0.35, water body with reference to NDVI results
NDVI be respectively less than 0, control NDVI threshold range, the atural objects such as vegetation, water body can be rejected.With reference to the curve of spectrum of image
Understand, the reflected value of the green wave band in rare earth mining area should be greater than 2000, and the reflected value of red wave band should be greater than 3000, near infrared band
Reflected value be more than 4000, can distinguish the typical features such as building, cement road according to this condition.In addition, passing through statistics
The clutter reflections value of image understand, the reflected value of partial white building is similar to rare earth mining area, but red wave band with closely it is red
Wave section, the reflectivity of red wave band is less than near infrared band, has obvious difference with the reflection spectrum curve of rare earth.It is special according to this
Levy, these can be rejected and obscure atural object.Spectrum rule specific requirement is as shown in table 1.
Spectrum rule in the object oriented classification method of table 1
Landform rule：Ion adsorption type rare earth ore is into ore deposit except with having outside the Pass into ore deposit protolith, topography and geomorphology is also very heavy
The minerogentic condition wanted.Topography and geomorphology can generally be described by a series of landform characterization factors, such as microcosmic factor elevation, slope
Degree, macroscopical factor earth's surface depth of cut, topographic relief amplitude etc., these indexs have an impact to the development and preservation of weathering crust.It is high
Journey refers to distance of the ground any point along plumb line direction to absolute datum；The gradient refers to the degree that ground table unit delays suddenly, i.e.,
The section of earth's surface certain point and the angle of level ground；Earth's surface depth of cut refers to the mean height of ground neighborhood of a point scope
Journey and the difference (D of the minimum elevation in the contiguous rangei=Hmean-Hmin)；Topographic relief amplitude is maximum high in designated area
Difference (the RF of journey and minimum elevationi=Hmax-Hmin)。
Understood with reference to the research of forefathers, the strong orographic condition of Xunwu area Ree Metallization is mainly：Elevation 100-500m,
0 ° -20 ° of the gradient, earth's surface depth of cut 20-150m, topographic relief amplitude 100-400m.The digital elevation model in research on utilization area
(DEM) data, can calculate the terrain factors such as elevation, the gradient, earth's surface depth of cut, topographic relief amplitude, so as to set up landform
Rule.If Fig. 4 a~Fig. 4 d are using the ASTER GEDM2 dem datas that spatial resolution is 30m, with ArcGIS spaces point
Analysis and ENVI band maths, that obtains meets the binary map of every terrain factor condition.Wherein, Fig. 4 a meet elevation for research area
The binary map of characterization factor condition, Fig. 4 b meet the binary map of gradient characterization factor condition for research area, and Fig. 4 c are research area symbol
The binary map of earth's surface depth of cut characterization factor condition is closed, Fig. 4 d meet the two of topographic relief amplitude characterization factor condition for research area
Value figure.
Conventional object-oriented classification method has two kinds：Membership function method and nearest neighbour classification.Membership function method
It is to construct membership function using fuzzy mathematics method, each feature is estimated and being subordinate in the range of 0-1 is returned
Angle value, the belonging kinds of object are determined according to the logic operation result for being respectively subordinate to angle value, it is adaptable to which a few feature can area
The atural object divided.Therefore membership function method is used herein to be classified, rule set is built using membership function, will be above-mentioned
Landform rule, spectrum rule and geometrical rule bring into rule set, and for anti-leak-stopping point and wrong point of phenomenon hair
It is raw, be according to the importance of each influence of rule to classification results, and different weights are assigned respectively, classified, obtained
The classification results in rare earth mining area, as shown in figure 5, being rare earth mining area of the research area result figure based on object oriented classification method.
Object-oriented classification method and traditional supervised classification method comparative analysis：
Supervised classification refers to the sample provided according to known training center, by calculating selection characteristic parameter, sets up and differentiates letter
Several methods classified to image to be sorted.Conventional supervised classification method has parallel pipe Dow process, maximum likelihood method, spectral modeling
Charting method etc..Exercised supervision classification from Spectral angle mapper method herein, the mining area for studying area is extracted, extraction result is such as
Shown in Fig. 6, and contrasted with object oriented classification method extraction result.
With reference to Object-Oriented Method classification results (Fig. 5) and supervised classification classification results (Fig. 6), from the point of view of qualitative,
Object oriented classification result precision, which is significantly greater than in supervised classification result, supervised classification result, is clearly present wrong point phenomenon, such as Fig. 6
Shown in middle square frame a, found through field investigation, the region is exposed for the earth's surface that construction is caused, not rare earth mining area, and supervising
Broken spot in classification results is superintended and directed more, as shown in round frame b, c in Fig. 6.And broken spot is seldom in object oriented classification result in Fig. 5, this
The substantial amounts of time is saved for the post-classification comparison work in later stage.
On the basis of qualitative evaluation is carried out to achievement, quantitative accuracy evaluation is further carried out.Conventional precision test method
Inspection-classification precision has two kinds, i.e. confusion matrix and ROC curve, and the precision of classification results can be shown in one by confusion matrix
Inside confusion matrix, and ROC curve is then graphically to express nicety of grading, compared with confusion matrix method, ROC curve
Method is more abstract, therefore uses confusion matrix method herein.Evaluated by confusion matrix method, the nicety of grading of traditional supervised classification
For 83..2168%, Kappa coefficients are 0.6923, and the nicety of grading of object oriented classification result is 92.49%, Kappa coefficients
For 0.8576, as shown in table 2 and table 3, hence it is evident that object oriented classification result is higher, therefore object oriented classification method classifying quality is more
It is good.
The supervised classification confusion matrix of table 2
The object oriented classification method confusion matrix of table 3
It should be pointed out that embodiment described above can make those skilled in the art that the present invention is more fully understood
Create, but do not limit the invention in any way is created.Therefore, although this specification and embodiment have been carried out to the invention
Detailed description, it will be understood by those skilled in the art, however, that still can be modified to the invention or equivalent
Replace；And technical scheme and its improvement of all spirit and scope for not departing from the invention, it is encompassed by wound of the present invention
Make among the protection domain of patent.
Claims (8)
1. a kind of rare earth mining area remote sensing information extracting method based on object oriented classification, it is characterised in that including following step
Suddenly：1) after the remotely-sensed data source in the research area of selection is pre-processed first, shadow is carried out using the partitioning algorithm based on edge
As segmentation；2) series of rules is set up then in conjunction with terrain information, spectral information and geological information, realizes feature extraction；3) most
The classification of object-oriented is carried out afterwards, extracts the rare earth mining area in research area.
2. according to the method described in claim 1, it is characterised in that the remotely-sensed data in the remotely-sensed data source is in July, 2011
The IKONOS image datas of 24 days, including 4 multi light spectrum hands (blue, green, red, near-infrared) and 1 panchromatic wave-band, wherein light more
Spectral resolution is 4m, and panchromatic wave-band is 1m.
3. method according to claim 2, it is characterised in that the pretreatment includes the remotely-sensed data carrying out air
Correction, ortho-rectification, image co-registration, a series of images processing work of geometric correction.
4. method according to claim 3, it is characterised in that the partitioning algorithm based on edge refers to according to certain
The sharp point detected is linked to be profile by rule, constitutes cut zone；During Image Segmentation, it is first determined segmentation chi
Degree, initial segmentation is carried out to image；It is then determined that merger yardstick so that same or analogous object merger is a class, is reduced just
The partitioning boundary for the number of objects split, further amendment object of beginning.
5. method according to claim 4, it is characterised in that the segmentation range scale is 0-100, is selected as 50；Institute
Merger range scale is stated for 0-100, is selected as 90.
6. the method according to one of claim 1 to 5, it is characterised in that the foundation rule refers to according to imaged object
A series of constraintss that relation between feature and atural object is set up, form series of rules；The geometrical rule refers to according to shadow
The rule set up as the geometric properties of object；The spectrum rule refers to recognize difference by the difference of electromagnetic spectrum feature
The characters of ground object of atural object, the rule that spectrum and reflectivity are set up as characters of ground object；The landform rule refers to research on utilization
Digital elevation model (DEM) data in area, a series of landform characterization factors of the description topography and landform character calculated and set up
Rule.
7. method according to claim 6, it is characterised in that the landform rule feature factor include elevation, the gradient,
Table depth of cut, topographic relief amplitude；The spectrum rule feature factor include normalized differential vegetation index, the reflected value of atural object,
The curve of spectrum tendency of thing；The geometrical rule feature includes imaged object area.
8. method according to claim 7, it is characterised in that the step 3) in, face is carried out using membership function method
To the classification of object, rule set is built using membership function, landform rule, spectrum rule and geometrical rule are brought into
In rule set, different weights are assigned respectively, are classified, draw the classification results in rare earth mining area.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710173837.XA CN106971156A (en) | 2017-03-22 | 2017-03-22 | Rare earth mining area remote sensing information extraction method based on object-oriented classification |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710173837.XA CN106971156A (en) | 2017-03-22 | 2017-03-22 | Rare earth mining area remote sensing information extraction method based on object-oriented classification |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106971156A true CN106971156A (en) | 2017-07-21 |
Family
ID=59330102
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710173837.XA Pending CN106971156A (en) | 2017-03-22 | 2017-03-22 | Rare earth mining area remote sensing information extraction method based on object-oriented classification |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106971156A (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107527364A (en) * | 2017-07-25 | 2017-12-29 | 浙江大学 | A kind of seaweed growing area monitoring method based on remote sensing images and lace curtaining information |
| CN107609526A (en) * | 2017-09-21 | 2018-01-19 | 吉林大学 | Rule-based fine dimension city impervious surface rapid extracting method |
| CN108132220A (en) * | 2017-12-25 | 2018-06-08 | 中国林业科学研究院资源信息研究所 | The BRDF normalization methods of the airborne push-broom type Hyperspectral imaging in forest zone |
| CN108509570A (en) * | 2018-03-26 | 2018-09-07 | 武汉智博创享科技股份有限公司 | The method and device of molecular formula annotation is drawn on the electronic map |
| CN108875615A (en) * | 2018-06-07 | 2018-11-23 | 中国石油天然气股份有限公司 | Deposition region remote sensing recognition method, apparatus, electronic equipment and storage medium |
| CN109143401A (en) * | 2018-10-10 | 2019-01-04 | 河北地质大学 | Ion adsorption type rare earth ore methods of prospecting for ore deposits and device based on remote sensing technology |
| CN109359621A (en) * | 2018-11-02 | 2019-02-19 | 中国地质科学院矿产资源研究所 | Pegmatite type lithium ore prospecting method based on multi-source remote sensing data |
| CN109581412A (en) * | 2019-01-17 | 2019-04-05 | 合肥工业大学 | A kind of method of quick carry out soil erosion Dynamic Change by Remote Sensing monitoring |
| CN110806605A (en) * | 2019-11-15 | 2020-02-18 | 中国地质科学院矿产综合利用研究所 | Remote sensing ore searching method for rare earth-uranium ore in high-latitude and high-cold area |
| CN111274968A (en) * | 2020-01-20 | 2020-06-12 | 广州市城市规划自动化中心 | Object-oriented road information extraction method and device and electronic equipment |
| CN111428627A (en) * | 2020-03-23 | 2020-07-17 | 西北大学 | Mountain landform remote sensing extraction method and system |
| CN111950361A (en) * | 2020-07-07 | 2020-11-17 | 内蒙古农业大学 | Beet identification method based on single-time-sequence NDVI |
| CN113446992A (en) * | 2021-06-28 | 2021-09-28 | 中国水利水电科学研究院 | Method for optimizing distribution of measuring points in topographic survey |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101930547A (en) * | 2010-06-24 | 2010-12-29 | 北京师范大学 | Method for automatically classifying remote sensing image based on object-oriented unsupervised classification |
| CN102902978A (en) * | 2012-08-31 | 2013-01-30 | 电子科技大学 | Object-oriented high-resolution remote-sensing image classification method |
| CN105427309A (en) * | 2015-11-23 | 2016-03-23 | 中国地质大学（北京） | Multiscale hierarchical processing method for extracting object-oriented high-spatial resolution remote sensing information |
| CN106446875A (en) * | 2016-11-16 | 2017-02-22 | 航天恒星科技有限公司 | County remote-sensing scale-oriented crop planting area information extraction method and device |
-
2017
- 2017-03-22 CN CN201710173837.XA patent/CN106971156A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101930547A (en) * | 2010-06-24 | 2010-12-29 | 北京师范大学 | Method for automatically classifying remote sensing image based on object-oriented unsupervised classification |
| CN102902978A (en) * | 2012-08-31 | 2013-01-30 | 电子科技大学 | Object-oriented high-resolution remote-sensing image classification method |
| CN105427309A (en) * | 2015-11-23 | 2016-03-23 | 中国地质大学（北京） | Multiscale hierarchical processing method for extracting object-oriented high-spatial resolution remote sensing information |
| CN106446875A (en) * | 2016-11-16 | 2017-02-22 | 航天恒星科技有限公司 | County remote-sensing scale-oriented crop planting area information extraction method and device |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107527364A (en) * | 2017-07-25 | 2017-12-29 | 浙江大学 | A kind of seaweed growing area monitoring method based on remote sensing images and lace curtaining information |
| CN107609526A (en) * | 2017-09-21 | 2018-01-19 | 吉林大学 | Rule-based fine dimension city impervious surface rapid extracting method |
| CN108132220A (en) * | 2017-12-25 | 2018-06-08 | 中国林业科学研究院资源信息研究所 | The BRDF normalization methods of the airborne push-broom type Hyperspectral imaging in forest zone |
| CN108509570B (en) * | 2018-03-26 | 2020-12-04 | 武汉智博创享科技股份有限公司 | Method and device for drawing molecular formula annotation on electronic map |
| CN108509570A (en) * | 2018-03-26 | 2018-09-07 | 武汉智博创享科技股份有限公司 | The method and device of molecular formula annotation is drawn on the electronic map |
| CN108875615A (en) * | 2018-06-07 | 2018-11-23 | 中国石油天然气股份有限公司 | Deposition region remote sensing recognition method, apparatus, electronic equipment and storage medium |
| CN108875615B (en) * | 2018-06-07 | 2021-04-30 | 中国石油天然气股份有限公司 | Deposition area remote sensing identification method and device, electronic equipment and storage medium |
| CN109143401A (en) * | 2018-10-10 | 2019-01-04 | 河北地质大学 | Ion adsorption type rare earth ore methods of prospecting for ore deposits and device based on remote sensing technology |
| CN109359621A (en) * | 2018-11-02 | 2019-02-19 | 中国地质科学院矿产资源研究所 | Pegmatite type lithium ore prospecting method based on multi-source remote sensing data |
| CN109581412A (en) * | 2019-01-17 | 2019-04-05 | 合肥工业大学 | A kind of method of quick carry out soil erosion Dynamic Change by Remote Sensing monitoring |
| CN110806605A (en) * | 2019-11-15 | 2020-02-18 | 中国地质科学院矿产综合利用研究所 | Remote sensing ore searching method for rare earth-uranium ore in high-latitude and high-cold area |
| CN111274968A (en) * | 2020-01-20 | 2020-06-12 | 广州市城市规划自动化中心 | Object-oriented road information extraction method and device and electronic equipment |
| CN111274968B (en) * | 2020-01-20 | 2023-04-14 | 广州市城市规划自动化中心 | Object-oriented road information extraction method and device and electronic equipment |
| CN111428627A (en) * | 2020-03-23 | 2020-07-17 | 西北大学 | Mountain landform remote sensing extraction method and system |
| CN111428627B (en) * | 2020-03-23 | 2023-03-24 | 西北大学 | Mountain landform remote sensing extraction method and system |
| CN111950361A (en) * | 2020-07-07 | 2020-11-17 | 内蒙古农业大学 | Beet identification method based on single-time-sequence NDVI |
| CN113446992A (en) * | 2021-06-28 | 2021-09-28 | 中国水利水电科学研究院 | Method for optimizing distribution of measuring points in topographic survey |
| CN113446992B (en) * | 2021-06-28 | 2023-06-16 | 中国水利水电科学研究院 | Method for optimizing distribution of topographic survey points in topographic survey |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106971156A (en) | Rare earth mining area remote sensing information extraction method based on object-oriented classification | |
| Wei et al. | Cloud detection for Landsat imagery by combining the random forest and superpixels extracted via energy-driven sampling segmentation approaches | |
| Pu et al. | A comparative analysis of high spatial resolution IKONOS and WorldView-2 imagery for mapping urban tree species | |
| Bunting et al. | The delineation of tree crowns in Australian mixed species forests using hyperspectral Compact Airborne Spectrographic Imager (CASI) data | |
| Vasuki et al. | An interactive image segmentation method for lithological boundary detection: A rapid mapping tool for geologists | |
| CN103364781A (en) | Remote sensing data and geographical information system-based grainfield ground reference point screening method | |
| CN105447274B (en) | A method of seashore wetland drawing being carried out to intermediate resolution remote sensing images using object oriented classification technology | |
| CN103646246A (en) | Decision tree model based multispectral remote sensing image river information extraction method | |
| CN102005034A (en) | Remote sensing image segmentation method based on region clustering | |
| CN110390255A (en) | High-speed rail environmental change monitoring method based on various dimensions feature extraction | |
| CN113033670A (en) | Method for extracting rice planting area based on Sentinel-2A/B data | |
| Dehvari et al. | Comparison of object-based and pixel based infrared airborne image classification methods using DEM thematic layer | |
| CN106340005A (en) | High-resolution remote sensing image unsupervised segmentation method based on scale parameter automatic optimization | |
| Zhang et al. | The analysis of object-based change detection in mining area: A case study with pingshuo coal mine | |
| CN114283213A (en) | House self-adaptive vectorization method combining LiDAR point cloud and optical image | |
| Lei et al. | A novel algorithm of individual tree crowns segmentation considering three-dimensional canopy attributes using UAV oblique photos | |
| YU et al. | Object oriented land cover classification using ALS and GeoEye imagery over mining area | |
| Chehata et al. | Object-based forest change detection using high resolution satellite images | |
| Chang et al. | An object-oriented analysis for characterizing the rainfall-induced shallow landslide | |
| You et al. | Segmentation of individual mangrove trees using UAV-based LiDAR data | |
| Kumar et al. | Multi-sensor multi-resolution image fusion for improved vegetation and urban area classification | |
| CN102231190B (en) | Automatic extraction method for alluvial-proluvial fan information | |
| Koc-San et al. | Plastic and glass greenhouses detection and delineation from worldview-2 satellite imagery | |
| CN117475314A (en) | Geological disaster hidden danger three-dimensional identification method, system and medium | |
| Nelson et al. | Spatial statistical techniques for aggregating point objects extracted from high spatial resolution remotely sensed imagery |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170721 |
|
| RJ01 | Rejection of invention patent application after publication |