CN110070137A - A kind of hyperspectral image classification method based on adaptive manifold filtering and domain transfer standard convolutional filtering - Google Patents
A kind of hyperspectral image classification method based on adaptive manifold filtering and domain transfer standard convolutional filtering Download PDFInfo
- Publication number
- CN110070137A CN110070137A CN201910345789.7A CN201910345789A CN110070137A CN 110070137 A CN110070137 A CN 110070137A CN 201910345789 A CN201910345789 A CN 201910345789A CN 110070137 A CN110070137 A CN 110070137A
- Authority
- CN
- China
- Prior art keywords
- filtering
- data collection
- information
- spatial
- domain transfer
- 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
- 238000001914 filtration Methods 0.000 title claims abstract description 51
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000012546 transfer Methods 0.000 title claims abstract description 29
- 230000003044 adaptive effect Effects 0.000 title claims abstract description 22
- 238000013480 data collection Methods 0.000 claims abstract description 50
- 238000001228 spectrum Methods 0.000 claims abstract description 25
- 230000004927 fusion Effects 0.000 claims abstract description 10
- 238000012549 training Methods 0.000 claims description 25
- 238000012360 testing method Methods 0.000 claims description 6
- 238000002790 cross-validation Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 abstract description 4
- VMXUWOKSQNHOCA-UKTHLTGXSA-N ranitidine Chemical compound [O-][N+](=O)\C=C(/NC)NCCSCC1=CC=C(CN(C)C)O1 VMXUWOKSQNHOCA-UKTHLTGXSA-N 0.000 description 14
- 241001466077 Salina Species 0.000 description 11
- 230000006978 adaptation Effects 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000012729 kappa analysis Methods 0.000 description 2
- 230000003595 spectral effect Effects 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 238000010521 absorption reaction Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000007500 overflow downdraw method Methods 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 241000894007 species Species 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- 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
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/40—Analysis of texture
-
- 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
- G06T2207/10036—Multispectral image; Hyperspectral image
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of hyperspectral image classification methods based on adaptive manifold filtering and domain transfer standard convolutional filtering, include the following steps: to input high spectrum image to be processed and to the high-spectral data collection normalized of setting wave band, obtains image data set W;Image data set W is subjected to PCA dimensionality reduction, the data of preceding n dimension in the data set W is selected to form new data set D;Adaptive manifold filtering is carried out to data set D, obtains spatial texture information Ksc;High-spectral data collection W is filtered with domain transfer standard convolutional filtering, obtains spatial correlation information Kst;By spatial texture information KscWith spatial correlation information KstLinear superposition merges to obtain fused spatial information data collection K;Data set K is trained and is classified using large-spacing Distributed learning machine.The present invention obtains spatial signature information by the filtering of adaptive manifold and domain transfer standard convolutional filtering and merges, and makes the extraction of characteristic information more sufficiently, completely;Classified with spatial information data collection of the large-spacing Distributed learning machine to fusion, improves nicety of grading.
Description
Technical field
The present invention relates to classification hyperspectral imagery fields, more particularly, to one kind based on the filtering of adaptive manifold and domain
The hyperspectral image classification method of transfer standard convolutional filtering.
Background technique
Bloom spectrum sensor obtains the reflected radiation information of atural object by a spectrum channels up to a hundred, and wavelength band covers
From visible light to near-infrared or even LONG WAVE INFRARED region, high spectrum image contain spatial information, reflection or the spoke of atural object simultaneously
Information and spectral information are penetrated, feature is commonly known as " collection of illustrative plates ".And spectral image data provides nearly continuity
Spectrum sample information, can recorde the reflection differences of atural object spectrally very little.The diagnosis that this characteristic is referred to as atural object is special
Property, it can be used as the foundation classified to atural object and detected.Classification hyperspectral imagery new technology is studied, there is important theory
Meaning and application value.
The technology of current classification hyperspectral imagery is primarily present following problems: not obtaining sufficiently during hyperspectral classification empty
Between feature, be easily lost using filter texture feature extraction atural object spatial coherence information, do not consider spatial texture spy
Spatial coherence Fusion Features of seeking peace get up to constitute more complete space characteristics, in classification hyperspectral imagery, it is traditional most
The disaggregated model of large-spacing model optimization cannot represent the interval point of entire training dataset both for some single interval
Cloth, it is difficult to further increase nicety of grading.Therefore the EO-1 hyperion based on adaptive manifold filtering and domain transfer standard convolutional filtering
Image classification is still the direction for being worth research.
Summary of the invention
The present invention is that overcome the above-mentioned spatial signature information of high spectrum image in the prior art to extract insufficient, it is imperfect, point
The low at least one defect of class precision provides a kind of EO-1 hyperion based on adaptive manifold filtering and domain transfer standard convolutional filtering
Image classification method.
The present invention is directed to solve above-mentioned technical problem at least to a certain extent.
Primary and foremost purpose of the invention is in order to solve the above technical problems, technical scheme is as follows:
A kind of hyperspectral image classification method based on adaptive manifold filtering and domain transfer standard convolutional filtering, including such as
Lower step:
S1: inputting high spectrum image to be processed, and the high-spectral data collection that high spectrum image medium wave number of segment is l is returned
One change processing, obtains the hyperspectral image data collection W of information content redistribution;
S2: carrying out PCA dimensionality reduction for high-spectral data collection W, and the data of preceding n dimension in the data set W is selected to form new number
According to collection D;
S3: adaptive manifold filtering is carried out to data set D, obtains spatial texture information Ksc;
S4: domain transfer standard convolutional filtering is made to high-spectral data collection W with domain transfer standard convolutional filtering, obtains space phase
Close information Kst;
S5: by spatial texture information KscWith spatial correlation information KstLinear superposition is carried out to merge to obtain fused space
Message data set K;
S6: fusion spatial information data collection K is trained and is classified using large-spacing Distributed learning machine.
Further, normalized described in step S1 is calculated according to following formula:
Wherein, R represents hyperspectral image data collection reflected intensity numerical value, and μ, σ are respectively mean value and variance.
Further, PCA dimensionality reduction is carried out to high-spectral data collection in step S2, formula is as follows:
D=Pca (W).
Further, adaptive manifold filtering is carried out to data set D, obtains spatial texture information Ksc, formula is as follows:
Ksc=F (D).
Further, domain transfer standard convolutional filtering is made to high-spectral data collection W with domain transfer standard convolutional filtering, obtains
Spatial correlation information Kst, formula is as follows:
Kst=T (W).
Further, by spatial information texture information KscWith spatial correlation information KstLinear superposition is carried out to merge
Spatial information data collection K after conjunction, formula are as follows:
K=Ksc+Kst。
Further, the step for fusion spatial information data collection K being trained and being classified using large-spacing Distributed learning machine
It is rapid as follows:
S6.1: training set K is randomly selected from spatial information data collection K ratio D% at randoms, remaining (1-D%) part conduct
Training set Kt;
S6.2: large-spacing Distributed learning machine cross validation is used, optimal parameter combination is found;
S6.3: with large-spacing Distributed learning machine to KsIt is trained, obtains training pattern;
S6.4: after obtaining training pattern, with large-spacing Distributed learning machine to test set KtClassify.
Compared with prior art, the beneficial effect of technical solution of the present invention is:
The present invention obtains spatial signature information by the filtering of adaptive manifold and domain transfer standard convolutional filtering and carries out
Fusion, keeps feature information extraction more abundant, more completely;Space is minimized simultaneously with maximization space characteristics interval averages
The large-spacing Distributed learning machine of significant interval Variance feature classifies to fused spatial information data collection, improves nicety of grading.
Detailed description of the invention
Fig. 1 is the method for the present invention flow chart.
Fig. 2 is nicety of grading statistical form of the different classifications method to Indian agricultural hyperspectral image data collection.
Fig. 3 is using different classifications method to the classifying quality figure of Indian agricultural hyperspectral image data collection.
Fig. 4 is nicety of grading statistical form of the different classifications method to Salinas mountain valley hyperspectral image data collection.
Fig. 5 is using figure different classifications method to the classifying quality figure of Salinas mountain valley hyperspectral image data collection.
Fig. 6 is Indian agricultural hyperspectral image data collection classification results precision histogram under different classifications method.
Fig. 7 is Salinas mountain valley hyperspectral image data collection classification results precision histogram under different classifications method.
Fig. 8 be Indian woods hyperspectral image data collection different training sample ratios classification after OA, AA and Kappa roll over
Line chart.
Fig. 9 be Salinas mountain valley hyperspectral image data collection different training sample ratios classification after OA, AA and
Kappa line chart.
Specific embodiment
The following further describes the technical solution of the present invention with reference to the accompanying drawings and examples.
Embodiment 1
The method of the present invention is tested using the high-spectral data image from two different topographic characteristics in the present embodiment
Card, specific as follows:
Indian agricultural hyperspectral image data collection is to be received in the Indian agricultural in the state of Indiana northwestward using spectrometer
The high-spectrum remote sensing collected, the image that this image data is concentrated have 20 meters of spatial resolution, and it includes 144x144
Pixel, 220 wave bands, since the factors such as noise and water absorption remove 20 wave bands therein, remaining 200 wave bands include 16
Kind of atural object, specifically species not and number of samples referring to fig. 2.
Salinas mountain valley hyperspectral image data collection is the image being collected into Salinas mountain valley using spectrometer, this
The image that image data is concentrated has 3.7 meters of spatial resolution, and it includes 512x217 pixel, 224 wave bands, due to making an uproar
Sound and water the factors such as absorb and remove 20 wave bands therein, remaining 204 wave bands, include 16 kinds of atural objects, specific vegetation classification and
Number of samples is referring to fig. 4.
The present embodiment carries out comparison using 12 kinds of classification methods and the method for the present invention, and the specific method is as follows:
SVM: the primitive character based on high spectrum image is directly classified using support vector machines.
EPF: classifying to high spectrum image by SVM, then carries out edge preserving filter to each probability graph, most
The class of each pixel is selected based on maximum probability afterwards.
IFRF: being based on image co-registration and recursion filter, obtains classification results using SVM.
LDM: the primitive character based on high spectrum image is directly classified using large-spacing distribution machine.
LDM-FL: classified using recursion filter combination large-spacing distribution machine.
PCA-EPFs: it is stacked by the spatial information constructed first using edge preserving filter special to form fusion
Sign, and the size of the classifier by PCA dimensionality reduction reduction SVM, classify.
GFDN: space characteristics are extracted in first three principal component of high spectrum image by Gabor filter, form fusion
Feature carries out depth network class then in conjunction with high spectrum image primitive character.
AMF-SVM: combination supporting vector machine is filtered by self adaptation stream and is classified.
AMF-LDM: large-spacing Distributed learning machine is combined to classify by self adaptation stream filtering.
DTNCF-SVM: classified by domain transfer standard convolutional filtering combination supporting vector machine.
DTNCF-LDM: classified by domain transfer standard convolutional filtering combination large-spacing Distributed learning machine.
AMFDTNCF-LDM: classification method of the invention.
AMFDTNCF-SVM: the classification method of the extraction of spatial information fusion method combination supporting vector machine in the present invention.
The present embodiment is using whole nicety of grading (Overall accuracy, OA), average nicety of grading (Average
Accuracy, AA) and Kappa statistics coefficient (Kappa statistic, Kappa) Lai Hengliang classification method precision, in order to
Random deviation is avoided, each experiment is repeated 10 times record average result, and verification platform uses Matlab R2012b, i7-6700
The experiment porch of CPU, 8GBRAM.
The specific steps of the present invention are as follows:
S1: inputting high spectrum image to be processed, and the high-spectral data collection that high spectrum image medium wave number of segment is l is returned
One change processing, obtains the hyperspectral image data collection W of information content redistribution;
Wherein, the normalized is calculated according to following formula:
Wherein, R represents hyperspectral image data collection reflected intensity numerical value, and μ, σ are respectively mean value and variance.
S2: carrying out PCA dimensionality reduction for high-spectral data collection W, and the data of preceding n dimension in the data set W is selected to form new number
According to collection, formula is handled are as follows: D=Pca (W);
S3: adaptive manifold filtering is carried out to data set D, obtains spatial texture information Ksc, formula is as follows: Ksc=F (D);
S4: domain transfer standard convolutional filtering is made to high-spectral data collection W with domain transfer standard convolutional filtering, obtains space phase
Close information Kst, formula is as follows: Kst=T (W);
S5: by spatial information texture information KscWith spatial correlation information KstProgress linear superposition merges to obtain fused
Spatial information data collection K, formula are as follows: K=Ksc+Kst;
S6: fusion spatial information data collection K is trained and is classified using large-spacing Distributed learning machine.
S6.1: training set K is randomly selected from spatial information data collection K ratio D% at randoms, remaining (1-D%) part conduct
Training set Kt;The ratio for the training set that data set in this implementation for two different landform is taken is as shown in Figure 2 and Figure 4.
S6.2: large-spacing Distributed learning machine cross validation is used, optimal parameter combination is found;
S6.3: with large-spacing Distributed learning machine to KsIt is trained, obtains training pattern;
S6.4: after obtaining training pattern, with large-spacing Distributed learning machine to test set KtClassify.
Indian agricultural hyperspectral image data collection, atural object are distributed as shown in (a) figure in Fig. 3, are chosen complete in atural object
16, portion classification, every class randomly select 4% sample composition have label training set, remaining 96% be used as test set, atural object quantity compared with
Few three classes atural object 20% is used as training set.Fig. 2 is classification of the various classification methods to Indian agriculture hyperspectral image data collection
Precision statistics, the classifying quality using different classifications method are as shown in Figure 3, wherein (b) to (n) indicates that meaning is as follows in Fig. 3:
(c) SVM, OA=76.11%;(c) EPF, OA=86.75%;(d) IFRF, OA=88.61%;(e)PCA-
EPFs, OA=90.67%;(f) LDM, OA=77.05%;(g) LDM-FL, OA=92.36%;(h) GFDN, OA=
95.70%;(i) AMF-SVM, OA=93.05%;(j) AMF-LDM, OA=96.34%;(k) DTNCF-SVM, OA=
92.28%;(l) DTNCF-LDM, OA=96.02%;(m) AMFDTNCF-SVM, OA=94.83%;(n)AMFDTNCF-LDM,
OA=97.23%.
Salinas mountain valley high spectrum image, atural object are distributed as shown in (a) figure in Fig. 5, and 16 kinds of plants are chosen in atural object
By classification, every class randomly select 0.7% sample composition have label training set, remaining 99.3% be used as test set, Fig. 4 be various points
Class method counts the nicety of grading of Salinas mountain valley data set, uses classifying quality such as Fig. 5 of different classifications classification method
It is shown.Wherein, (b) to (n) indicates that meaning is as follows in Fig. 5:
(b) SVM, OA=86.61%;(c) EPF, OA=89.39%;(d) IFRF, OA=97.15%;(e)PCA-
EPFs, OA=96.45%;(f) LDM, OA=88.39%;(g) LDM-FL, OA=98.49%;(h) GFDN, OA=
95.58%;(i) AMF-SVM, OA=92.26%;(j) AMF-LDM, OA=95.91%;(k) DTNCF-SVM, OA=
94.93%;(l) DTNCF-LDM, OA=98.48%;(m) AMFDTNCF-SVM, OA=96.21%;(n)AMFDTNCF-LDM,
OA=98.52%.
Verifying is analyzed as follows:
(1) as shown in Figure 6 and Figure 7, LDM classification method is to Indian agricultural and Salinas mountain valley hyperspectral image data
The OA of collection is respectively 77.05% and 88.39%, is higher by 0.94 and 1.78 percentage points, the classification side AMF-LDM than svm classifier method
Method and AMFDTNCF-LDM classification method are to Indian agricultural nicety of grading OA than classification method AMF-LDM and classification method
AMFDTNCF-LDM is higher by 3.03 and 2.4 percentage points respectively;Equally, AMF-LDM classification method and AMFDTNCF-LDM classification
Method is to Salinas mountain valley hyperspectral image data collection nicety of grading OA ratio AMF-LDM classification method and AMFDTNCF-LDM points
Class method is higher by 3.65 and 3.56 percentage points respectively, demonstrates LDM and is carried out based on maximization interval averagely by space characteristics
Value minimizes the validity of interval variance simultaneously.
The nicety of grading of different training sample testing classification methods is selected, as shown in figure 9, the wherein number mark on curve
Label are OA numerical value.Training sample ratio OA of the Indian woods hyperspectral image data collection overall classification accuracy in training sample 3%
Just reach 96.38%, 6% training sample OA has reached 98.12%;Salinas mountain valley hyperspectral image data collection totally divides
Class precision OA has just reached 83.23% and 94.28% when training sample is 0.1% and 0.2%, demonstrates the method for the present invention and exists
Also preferably nicety of grading can be obtained in the case where low training sample, and has certain stability.
The same or similar label correspond to the same or similar components;
The terms describing the positional relationship in the drawings are only for illustration, should not be understood as the limitation to this patent;
Obviously, the above embodiment of the present invention be only to clearly illustrate example of the present invention, and not be pair
The restriction of embodiments of the present invention.For those of ordinary skill in the art, may be used also on the basis of the above description
To make other variations or changes in different ways.There is no necessity and possibility to exhaust all the enbodiments.It is all this
Made any modifications, equivalent replacements, and improvements etc., should be included in the claims in the present invention within the spirit and principle of invention
Protection scope within.
Claims (7)
1. a kind of hyperspectral image classification method based on adaptive manifold filtering and domain transfer standard convolutional filtering, feature exist
In including the following steps:
S1: inputting high spectrum image to be processed, and the high-spectral data collection that high spectrum image medium wave number of segment is l is normalized
Processing obtains the hyperspectral image data collection W of information content redistribution;
S2: carrying out PCA dimensionality reduction for high-spectral data collection W, and the data of preceding n dimension in the data set W is selected to form new data set
D;
S3: adaptive manifold filtering is carried out to data set D, obtains spatial texture information Ksc;
S4: domain transfer standard convolutional filtering is made to high-spectral data collection W with domain transfer standard convolutional filtering, obtains space correlation letter
Cease Kst;
S5: by spatial texture information KscWith spatial correlation information KstLinear superposition is carried out to merge to obtain fused spatial information number
According to collection K;
S6: fusion spatial information data collection K is trained and is classified using large-spacing Distributed learning machine.
2. a kind of high-spectrum based on adaptive manifold filtering and domain transfer standard convolutional filtering according to claim 1
As classification method, which is characterized in that normalized described in step S1 is calculated according to following formula:
Wherein, R represents hyperspectral image data collection reflected intensity numerical value, and μ, σ are respectively mean value and variance.
3. a kind of high-spectrum based on adaptive manifold filtering and domain transfer standard convolutional filtering according to claim 1
As classification method, which is characterized in that carry out PCA dimensionality reduction to high-spectral data collection in step S2, formula is as follows:
D=Pca (W).
4. a kind of high-spectrum based on adaptive manifold filtering and domain transfer standard convolutional filtering according to claim 1
As classification method, which is characterized in that carry out adaptive manifold filtering to data set D, obtain spatial texture information Ksc, formula is such as
Under:
Ksc=F (D).
5. a kind of high-spectrum based on adaptive manifold filtering and domain transfer standard convolutional filtering according to claim 1
As classification method, which is characterized in that make the filter of domain transfer standard convolution to high-spectral data collection W with domain transfer standard convolutional filtering
Wave obtains spatial correlation information Kst, formula is as follows:
Kst=T (W).
6. a kind of high-spectrum based on adaptive manifold filtering and domain transfer standard convolutional filtering according to claim 1
As classification method, which is characterized in that by spatial information texture information KscWith spatial correlation information KstLinear superposition is carried out to merge
To fused spatial information data collection K, formula is as follows:
K=Ksc+Kst。
7. according to claim 1-6 a kind of based on the filtering of adaptive manifold and domain transfer standard convolutional filtering
Hyperspectral image classification method, which is characterized in that fusion spatial information data collection K is instructed using large-spacing distribution machine LDM
The step of practicing and classifying is as follows:
S6.1: training set K is randomly selected from spatial information data collection K ratio D% at randoms, remaining (1-D%) is partially as training
Collect Kt;
S6.2: large-spacing Distributed learning machine cross validation is used, optimal parameter combination is found;
S6.3: with large-spacing Distributed learning machine to KsIt is trained, obtains training pattern;
S6.4: after obtaining training pattern, with large-spacing Distributed learning machine to test set KtClassify.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910345789.7A CN110070137A (en) | 2019-04-26 | 2019-04-26 | A kind of hyperspectral image classification method based on adaptive manifold filtering and domain transfer standard convolutional filtering |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910345789.7A CN110070137A (en) | 2019-04-26 | 2019-04-26 | A kind of hyperspectral image classification method based on adaptive manifold filtering and domain transfer standard convolutional filtering |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110070137A true CN110070137A (en) | 2019-07-30 |
Family
ID=67369138
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910345789.7A Pending CN110070137A (en) | 2019-04-26 | 2019-04-26 | A kind of hyperspectral image classification method based on adaptive manifold filtering and domain transfer standard convolutional filtering |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110070137A (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106326926A (en) * | 2016-08-23 | 2017-01-11 | 复旦大学 | Hyperspectral image target spectrum learning method |
CN107832693A (en) * | 2017-10-31 | 2018-03-23 | 广东交通职业技术学院 | A kind of high spectrum image vegetation classification method based on spatial autocorrelation information |
CN108399355A (en) * | 2017-02-08 | 2018-08-14 | 广东交通职业技术学院 | A kind of hyperspectral image classification method that spatial information adaptively merges |
CN109598284A (en) * | 2018-10-23 | 2019-04-09 | 广东交通职业技术学院 | A kind of hyperspectral image classification method based on large-spacing distribution and space characteristics |
-
2019
- 2019-04-26 CN CN201910345789.7A patent/CN110070137A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106326926A (en) * | 2016-08-23 | 2017-01-11 | 复旦大学 | Hyperspectral image target spectrum learning method |
CN108399355A (en) * | 2017-02-08 | 2018-08-14 | 广东交通职业技术学院 | A kind of hyperspectral image classification method that spatial information adaptively merges |
CN107832693A (en) * | 2017-10-31 | 2018-03-23 | 广东交通职业技术学院 | A kind of high spectrum image vegetation classification method based on spatial autocorrelation information |
CN109598284A (en) * | 2018-10-23 | 2019-04-09 | 广东交通职业技术学院 | A kind of hyperspectral image classification method based on large-spacing distribution and space characteristics |
Non-Patent Citations (2)
Title |
---|
廖建尚等: "基于自适应流形滤波的高光谱图像分类方法", 《激光与光电子学进展》 * |
廖建尚等: "空间信息自适应融合的高光谱图像分类方法", 《光子学报》 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Wang et al. | Identification of tomato disease types and detection of infected areas based on deep convolutional neural networks and object detection techniques | |
Trier et al. | Tree species classification in Norway from airborne hyperspectral and airborne laser scanning data | |
Suárez et al. | Artificial intelligence in agriculture | |
Hamada et al. | Detecting Tamarisk species (Tamarix spp.) in riparian habitats of Southern California using high spatial resolution hyperspectral imagery | |
Lin et al. | Classification of tree species in overstorey canopy of subtropical forest using QuickBird images | |
CN107871132B (en) | Hyperspectral image classification method for spatial feature adaptive optimization | |
CN110346312B (en) | Winter wheat head gibberellic disease identification method based on Fisher linear discrimination and support vector machine technology | |
CN107066959B (en) | Hyperspectral image classification method based on guided filtering and linear spatial correlation information | |
CN110376202B (en) | Tea tree anthracnose lesion identification method based on imaging hyperspectral technology | |
US11682203B2 (en) | Feature extraction method, model training method, detection method of fruit spectrum | |
US8068132B2 (en) | Method for identifying Guignardia citricarpa | |
Hati et al. | Plant recognition from leaf image through artificial neural network | |
CN110763698A (en) | Hyperspectral citrus leaf disease identification method based on characteristic wavelength | |
CN109598284A (en) | A kind of hyperspectral image classification method based on large-spacing distribution and space characteristics | |
CN111161362A (en) | Tea tree growth state spectral image identification method | |
CN111639587A (en) | Hyperspectral image classification method based on multi-scale spectrum space convolution neural network | |
Mengistu et al. | An automatic coffee plant diseases identification using hybrid approaches of image processing and decision tree | |
Zhou et al. | Identification of the variety of maize seeds based on hyperspectral images coupled with convolutional neural networks and subregional voting | |
CN108399355A (en) | A kind of hyperspectral image classification method that spatial information adaptively merges | |
CN114266961A (en) | Method for integrating, learning and classifying marsh vegetation stacks by integrating hyperspectral and multiband fully-polarized SAR images | |
CN107145831A (en) | Based on vector probabilistic diffusion and markov random file Hyperspectral Image Classification method | |
CN107527001A (en) | A kind of hyperspectral image classification method based on Steerable filter and linear space correlation information | |
CN116124710A (en) | Early pine tree identification method for pine wood nematode disease | |
Cao et al. | Discrimination of tea plant variety using in-situ multispectral imaging system and multi-feature analysis | |
CN111523587A (en) | Woody plant species spectrum identification method based on machine learning |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190730 |