CN109191443B - Hyperspectral image waveband selection method based on sequence information and waveband quality - Google Patents

Hyperspectral image waveband selection method based on sequence information and waveband quality Download PDF

Info

Publication number
CN109191443B
CN109191443B CN201810991085.2A CN201810991085A CN109191443B CN 109191443 B CN109191443 B CN 109191443B CN 201810991085 A CN201810991085 A CN 201810991085A CN 109191443 B CN109191443 B CN 109191443B
Authority
CN
China
Prior art keywords
band
sequence
sub
quality
representing
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.)
Active
Application number
CN201810991085.2A
Other languages
Chinese (zh)
Other versions
CN109191443A (en
Inventor
陈尉钊
杨志景
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guangdong University of Technology
Original Assignee
Guangdong University of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Guangdong University of Technology filed Critical Guangdong University of Technology
Priority to CN201810991085.2A priority Critical patent/CN109191443B/en
Publication of CN109191443A publication Critical patent/CN109191443A/en
Application granted granted Critical
Publication of CN109191443B publication Critical patent/CN109191443B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • 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
    • 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
    • G06T2207/20—Special algorithmic details
    • G06T2207/20212—Image combination
    • G06T2207/20221—Image fusion; Image merging
    • 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
    • G06T2207/30168—Image quality inspection

Landscapes

  • Engineering & Computer Science (AREA)
  • Quality & Reliability (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Investigating Or Analysing Materials By Optical Means (AREA)
  • Spectrometry And Color Measurement (AREA)

Abstract

The invention provides a hyperspectral image band selection method based on sequence information and band quality, which comprises the following steps of: dividing a hyperspectral image space plane into sequence sub-blocks with sequence relation; representing each layer wave band of each sequence sub-block as a column vector, and carrying out standardization processing on the column vector; constructing a spectral band quality evaluation search criterion; fusing each sequence sub-block with a search criterion to construct a sub-block measurement matrix; fusing the sub-block measurement matrix with a determinant point process, and selecting a spectrum band subset with low redundancy and good separability; and outputting the index of the selected spectral band subset. According to the hyperspectral image band selection method based on sequence information and band quality, the redundancy of spectral bands is measured from different spatial domains, and the measurement accuracy is improved; by constructing a search criterion, the quality of the spectral band is effectively evaluated, and the quality of the selected spectral band is improved.

Description

Hyperspectral image waveband selection method based on sequence information and waveband quality
Technical Field
The invention relates to the technical field of hyperspectral images, in particular to a hyperspectral image waveband selection method based on sequence information and waveband quality.
Background
The hyperspectral image is data of a three-dimensional structure, and compared with the traditional image, the hyperspectral image can provide abundant spectral information through the spectral dimension, and is widely applied to agriculture, geology and atmospheric research. However, abundant spectral bands also bring a series of problems, firstly, data redundancy exists, excessive bands cause a huge data set, the computational complexity is high, and a common processor is difficult to effectively process. Secondly, in the excessive spectral bands, there are bands with poor quality, which also affects the classification accuracy.
The traditional spectral band selection method generally comprises two steps, firstly, designing a search criterion, namely designing a search criterion for selecting a spectral band meeting the criterion; the second is a search method for determining how to select spectral bands in the raw data set. Common methods include forward-based search methods. In each search, a band is added to the existing search subset, so that the bands of the subset meet the search criteria. Two disadvantages exist in this method, one is that the method is a traversal method, and has high computational complexity and long time. Secondly, this method lacks consideration of the whole hyperspectral image, but the measure of redundancy is not accurate from the consideration of each selected partial spectral band.
The common spectral band selection method also comprises a sorting-based method, wherein the correlation between each band and the whole data is calculated, sorting is carried out according to the magnitude of the correlation, and the spectral band in the front sorting is selected.
Disclosure of Invention
The invention provides a hyperspectral image band selection method based on sequence information and band quality, aiming at overcoming the technical defects that the redundancy of spectral bands is measured inaccurately and the quality of the spectral bands cannot be evaluated in the existing spectral band selection method.
In order to solve the technical problems, the technical scheme of the invention is as follows:
the hyperspectral image band selection method based on sequence information and band quality comprises the following steps:
s1: dividing a hyperspectral image space plane into sequence sub-blocks with sequence relation;
s2: representing each layer wave band of each sequence sub-block as a column vector, and carrying out standardization processing on the column vector;
s3: constructing a spectral band quality evaluation search criterion;
s4: fusing each sequence sub-block with a search criterion by using a Gaussian radial basis function to construct different sub-block measurement matrixes;
s5: fusing the obtained sub-block measurement matrix with a determinant point process, and selecting a spectrum band subset with low redundancy and good separability;
s6: and outputting the index of the selected spectral band subset.
In step S1, the raw hyperspectral data represents:
B＝{b1,b2,b3,…,bl}∈Rn×l,bi∈Rn×1；
wherein n is the total number of pixel points in each layer of spectral band,l is the total number of spectral bands; bi(1 ≦ i ≦ l) for the ith layer spectral band;
the mth sequence subblock is represented as:
Figure GDA0003465377590000021
wherein, each subblock contains r pixel points.
In step S2, each layer band of each sequence sub-block is expressed as a one-dimensional column vector, specifically:
Figure GDA0003465377590000022
the calculation formula for normalizing the column vectors is:
Figure GDA0003465377590000023
wherein, the step S3 specifically includes:
s31: estimating the information entropy of each layer of spectral band, wherein the calculation formula is as follows:
Figure GDA0003465377590000024
wherein p isiRepresenting the pixel point of the ith layer; p (p)i) Probability distribution estimation values representing pixel points, estimated using a histogram or estimated using a Parzen window, HiInformation entropy representing the ith layer;
s32: collecting sample pixel points and calculating mutual information of the category labels, wherein the collected sample pixel points are expressed as:
Figure GDA0003465377590000031
wherein
Figure GDA0003465377590000032
Pixel sample points representing acquisitions of the ith layer of optical tape, n representing the number of acquisitions, the corresponding labels being expressed as: c ═ C1,c2,c3,…,cn]T；
Therefore, the formula for calculating the mutual information is as follows:
Figure GDA0003465377590000033
wherein the content of the first and second substances,
Figure GDA0003465377590000034
representing a sample set
Figure GDA0003465377590000035
The α 1 st sample of (a); c. Cα2The α 2-th label representing the label set C;
s33: establishing a spectral band quality evaluation search criterion according to the information entropy and the mutual information, wherein the maximization formula is as follows:
maxQi＝Hi+Mi；
wherein Q isiRepresenting the quality of the spectral band, QiThe larger the value, the better the spectral band quality.
In step S4, the similarity matrix constructed by each sequence of sub-blocks is represented as:
Figure GDA0003465377590000036
wherein the content of the first and second substances,
Figure GDA0003465377590000037
indicating the correlation of the ith and jth layer wave bands calculated according to the mth and sequence information; the correlation relationship between every two of all the bands can be expressed as a matrix LmL is the total number of wave bands;
Figure GDA0003465377590000038
sigma is an adjusting parameter and is set to be between 0 and 1;
fusing similar matrixes constructed by different sub-blocks to construct a measurement matrix S with weight, wherein the calculation formula of the measurement matrix is as follows:
Figure GDA0003465377590000039
wherein alpha ismAs weight value, representing the importance of the sub-block, take
Figure GDA00034653775900000310
Wherein, the step S5 specifically includes:
s51: performing characteristic decomposition on the measurement matrix S, and selecting k characteristic values and corresponding characteristic vectors from the measurement matrix S;
s52: from the selected feature vectors, indices of k spectral bands are selected.
Wherein, the step S51 specifically includes:
s511: performing characteristic decomposition on the measurement matrix S to obtain characteristic values and corresponding characteristic vectors
Figure GDA0003465377590000041
S512: computing a characteristic polynomial
Figure GDA0003465377590000042
Wherein the characteristic polynomial is expressed as
Figure GDA0003465377590000043
S513: let h be k, n be l;
s514: let n be n-1, judge whether u is less than
Figure GDA0003465377590000044
If yes, go to step S515; if not, repeatedly executing the step S514;
s515: storing the collected characteristic values and indexes of corresponding characteristic vectors into a set S, and assigning a parameter h-1 to a parameter h;
s516: judging whether the parameter h is 0, if so, outputting an index set S of the set characteristic vectors; if not, go to step S514.
Wherein, the step S52 specifically includes:
s521: calculating { V | Vn}n∈SInitializing a spectrum band index set Y;
s522: judging whether the | V | is greater than 0; if yes, go to step S53; if not; step S55 is executed;
s523: judging whether the following formula is satisfied:
Figure GDA0003465377590000045
pr (i) is the probability of the ith layer spectrum being selected; u is a random variable uniformly distributed from [0,1 ];
if yes, go to step S54; if not, the step S53 is executed repeatedly;
s524: storing the wave band index i into a spectrum wave band index set Y, and assigning orthogonal and standardized V to the spectrum wave band index set Y
Figure GDA0003465377590000046
S525: and outputting a spectral band index set Y.
Wherein the step S6 specifically includes: and (4) storing the indexes of the spectral band sets selected in the step (S5) to obtain the spectral band subsets with low redundancy and high quality.
Compared with the prior art, the technical scheme of the invention has the beneficial effects that:
according to the hyperspectral image band selection method based on sequence information and band quality, provided by the invention, the spatial sequence information of a hyperspectral image is effectively utilized, the redundancy of spectral bands is measured from different spatial domains, and the measurement accuracy is improved; by constructing a search criterion, the quality of the spectral band is effectively evaluated, and the quality of the selected spectral band is improved.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Fig. 2 is a flow chart of an algorithm for selecting k eigenvectors.
Fig. 3 is a flow chart of an algorithm for selecting k spectral bands.
Detailed Description
The drawings are for illustrative purposes only and are not to be construed as limiting the patent;
for the purpose of better illustrating the embodiments, certain features of the drawings may be omitted, enlarged or reduced, and do not represent the size of an actual product;
it will be understood by those skilled in the art that certain well-known structures in the drawings and descriptions thereof may be omitted.
The technical solution of the present invention is further described below with reference to the accompanying drawings and examples.
As shown in fig. 1, the hyperspectral image band selection method based on sequence information and band quality includes the following steps:
s1: dividing a hyperspectral image space plane into sequence sub-blocks with sequence relation;
s2: representing each layer wave band of each sequence sub-block as a column vector, and carrying out standardization processing on the column vector;
s3: constructing a spectral band quality evaluation search criterion;
s4: fusing each sequence sub-block with a search criterion by using a Gaussian radial basis function to construct different sub-block measurement matrixes;
s5: fusing the obtained sub-block measurement matrix with a determinant point process, and selecting a spectrum band subset with low redundancy and good separability;
s6: and outputting the index of the selected spectral band subset.
More specifically, in step S1, the raw hyperspectral data represents:
B＝{b1,b2,b3,…,bl}∈Rn×l,bi∈Rn×1；
wherein n is the total number of pixel points of each layer of spectral band, and l is the total number of spectral bands; bi(1 ≦ i ≦ l) for the ith layer spectral band;
the mth sequence subblock is represented as:
Figure GDA0003465377590000061
wherein, each subblock contains r pixel points.
More specifically, in step S2, each layer band of each sequence sub-block is represented as a one-dimensional column vector, specifically:
Figure GDA0003465377590000062
the calculation formula for normalizing the column vectors is:
Figure GDA0003465377590000063
more specifically, the step S3 specifically includes:
s31: estimating the information entropy of each layer of spectral band, wherein the calculation formula is as follows:
Figure GDA0003465377590000064
wherein p isiRepresenting the pixel point of the ith layer; p (p)i) Probability distribution estimation values representing pixel points, estimated using a histogram or estimated using a Parzen window, HiInformation entropy representing the ith layer;
s32: collecting sample pixel points and calculating mutual information of the category labels, wherein the collected sample pixel points are expressed as:
Figure GDA0003465377590000065
wherein
Figure GDA0003465377590000066
Pixel sample points representing acquisitions of the ith layer of optical tape, n representing the number of acquisitions, the corresponding labels being expressed as: c ═ C1,c2,c3,…,cn]T；
Therefore, the formula for calculating the mutual information is as follows:
Figure GDA0003465377590000067
wherein the content of the first and second substances,
Figure GDA0003465377590000068
representing a sample set
Figure GDA0003465377590000069
The α 1 st sample of (a); c. Cα2The α 2-th label representing the label set C;
s33: establishing a spectral band quality evaluation search criterion according to the information entropy and the mutual information, wherein the maximization formula is as follows:
maxQi＝Hi+Mi；
wherein Q isiRepresenting the quality of the spectral band, QiThe larger the value, the better the spectral band quality.
More specifically, in step S4, the similarity matrix constructed by each sequence of sub-blocks is represented as:
Figure GDA0003465377590000071
wherein the content of the first and second substances,
Figure GDA0003465377590000072
indicating the correlation of the ith and jth layer wave bands calculated according to the mth and sequence information; the correlation relationship between every two of all the bands can be expressed as a matrix LmL is the total number of wave bands;
Figure GDA0003465377590000073
sigma is an adjusting parameter and is set to be between 0 and 1;
fusing similar matrixes constructed by different sub-blocks to construct a measurement matrix S with weight, wherein the calculation formula of the measurement matrix is as follows:
Figure GDA0003465377590000074
wherein alpha ismAs weight value, representing the importance of the sub-block, take
Figure GDA0003465377590000075
More specifically, the step S5 specifically includes:
s51: performing characteristic decomposition on the measurement matrix S, and selecting k characteristic values and corresponding characteristic vectors from the measurement matrix S;
s52: from the selected feature vectors, indices of k spectral bands are selected.
More specifically, the step S51 specifically includes:
s511: performing characteristic decomposition on the measurement matrix S to obtain characteristic values and corresponding characteristic vectors
Figure GDA0003465377590000076
S512: computing a characteristic polynomial
Figure GDA0003465377590000077
Wherein the characteristic polynomial is expressed as
Figure GDA0003465377590000078
S513: let h be k, n be l;
s514: let n be n-1, judge whether u is less than
Figure GDA0003465377590000079
If yes, go to step S515; if not, repeatedly executing the step S514;
s515: storing the collected characteristic values and indexes of corresponding characteristic vectors into a set S, and assigning a parameter h-1 to a parameter h;
s516: judging whether the parameter h is 0, if so, outputting an index set S of the set characteristic vectors; if not, go to step S514.
More specifically, the step S52 specifically includes:
s521: calculating { V | Vn}n∈SInitializing a spectrum band index set Y;
s522: judging whether the | V | is greater than 0; if yes, go to step S53; if not; step S55 is executed;
s523: judging whether the following formula is satisfied:
Figure GDA0003465377590000081
pr (i) is the probability of the ith layer spectrum being selected; u is a random variable uniformly distributed from [0,1 ];
if yes, go to step S54; if not, the step S53 is executed repeatedly;
s524: storing the wave band index i into a spectrum wave band index set Y, and assigning orthogonal and standardized V to the spectrum wave band index set Y
Figure GDA0003465377590000082
S525: and outputting a spectral band index set Y.
More specifically, the step S6 specifically includes: and (4) storing the indexes of the spectral band sets selected in the step (S5) to obtain the spectral band subsets with low redundancy and high quality.
In a specific implementation process, the hyperspectral image band selection method based on sequence information and band quality effectively utilizes spatial sequence information of a hyperspectral image to measure the redundancy of spectral bands from different spatial domains, and improves the measurement accuracy; by constructing a search criterion, the quality of the spectral band is effectively evaluated, and the quality of the selected spectral band is improved.
It should be understood that the above-described embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. Other variations and modifications will be apparent to persons skilled in the art in light of the above description. And are neither required nor exhaustive of all embodiments. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the claims of the present invention.

Claims (5)

1. The hyperspectral image band selection method based on sequence information and band quality is characterized by comprising the following steps of:
s1: dividing a hyperspectral image space plane into sequence sub-blocks with sequence relation;
s2: representing each layer wave band of each sequence sub-block as a column vector, and carrying out standardization processing on the column vector;
s3: constructing a spectral band quality evaluation search criterion;
s4: fusing each sequence sub-block with a search criterion by using a Gaussian radial basis function to construct different sub-block measurement matrixes;
s5: fusing the obtained sub-block measurement matrix with a determinant point process, and selecting a spectrum band subset with low redundancy and good separability;
s6: outputting the index of the selected spectral band subset;
in step S1, the raw hyperspectral data represents:
B＝{b1,b2,b3,…,bl}∈Rn×l,bi∈Rn×1；
wherein n is the total number of pixel points of each layer of spectral band, and l is the total number of spectral bands; biThe spectral band of the ith layer is expressed, wherein i is more than or equal to 1 and less than or equal to l;
the mth sequence subblock is represented as:
Figure FDA0003465377580000011
wherein each sub-block contains r pixel points;
in step S2, each layer band of each sequence sub-block is expressed as a one-dimensional column vector, specifically:
Figure FDA0003465377580000012
the calculation formula for normalizing the column vectors is:
Figure 1
；
the step S3 specifically includes:
s31: estimating the information entropy of each layer of spectral band, wherein the calculation formula is as follows:
Figure FDA0003465377580000021
wherein p isiRepresenting the pixel point of the ith layer; p (p)i) Probability distribution estimation values representing pixel points, estimated using a histogram or estimated using a Parzen window, HiInformation entropy representing the ith layer;
s32: collecting sample pixel points and calculating mutual information of the category labels, wherein the collected sample pixel points are expressed as:
Figure FDA0003465377580000022
wherein
Figure FDA0003465377580000023
Pixel sample points representing acquisitions of the ith layer of optical tape, n representing the number of acquisitions, the corresponding labels being expressed as: c ═ C1,c2,c3,…,cn]T；
Therefore, the formula for calculating the mutual information is as follows:
Figure FDA0003465377580000024
wherein the content of the first and second substances,
Figure FDA0003465377580000025
representing a sample set
Figure FDA0003465377580000026
The α 1 st sample of (a); c. Cα2The α 2-th label representing the label set C;
s33: establishing a spectral band quality evaluation search criterion according to the information entropy and the mutual information, wherein the maximization formula is as follows:
max Qi＝Hi+Mi；
wherein Q isiRepresenting the quality of the spectral band, QiThe larger the value, the better the spectral band quality;
in step S4, the similarity matrix constructed by each sequence of sub-blocks is represented as:
Figure FDA0003465377580000027
wherein the content of the first and second substances,
Figure FDA0003465377580000028
is according to the m-thThe correlation relation of the ith and jth layer wave bands calculated by the sequence information; the correlation relationship between every two of all the bands can be expressed as a matrix LmL is the total number of wave bands;
Figure FDA0003465377580000029
sigma is an adjusting parameter and is set to be between 0 and 1;
fusing similar matrixes constructed by different sub-blocks to construct a measurement matrix S with weight, wherein the calculation formula of the measurement matrix is as follows:
Figure FDA0003465377580000031
wherein alpha ismAs weight value, representing the importance of the sub-block, take
Figure FDA0003465377580000032
2. The hyperspectral image band selection method based on sequence information and band quality according to claim 1, wherein the step S5 specifically is:
s51: performing characteristic decomposition on the measurement matrix S, and selecting k characteristic values and corresponding characteristic vectors from the measurement matrix S;
s52: from the selected feature vectors, indices of k spectral bands are selected.
3. The hyperspectral image band selection method based on sequence information and band quality according to claim 2, wherein the step S51 specifically is:
s511: performing characteristic decomposition on the measurement matrix S to obtain characteristic values and corresponding characteristic vectors
Figure FDA0003465377580000033
S512: computing a characteristic polynomial
Figure FDA0003465377580000034
Wherein h is 0,1,2, …, k; n is 0,1,2, …, l;
wherein the characteristic polynomial is expressed as
Figure FDA0003465377580000035
S513: let h be k, n be l;
s514: let n be n-1, judge whether u is less than
Figure FDA0003465377580000036
If yes, go to step S515; if not, repeatedly executing the step S514;
s515: storing the collected characteristic values and indexes of corresponding characteristic vectors into a set S, and assigning a parameter h-1 to a parameter h;
s516: judging whether the parameter h is 0, if so, outputting an index set S of the set characteristic vectors; if not, go to step S514.
4. The hyperspectral image band selection method based on sequence information and band quality according to claim 3, wherein the step S52 specifically comprises:
s521: calculating { V | Vn}n∈SInitializing a spectrum band index set Y;
s522: judging whether the | V | is greater than 0; if yes, go to step S523; if not; step S525 is performed;
s523: judging whether the following formula is satisfied:
Figure FDA0003465377580000041
pr (i) is the probability of the ith layer spectrum being selected; u is a random variable uniformly distributed from [0,1 ];
if yes, go to step S524; if not, the step S523 is executed repeatedly;
s524: storing the wave band index i into a spectrum wave band index set Y, and assigning orthogonal and standardized V to the spectrum wave band index set Y
Figure FDA0003465377580000042
S525: and outputting a spectral band index set Y.
5. The hyperspectral image band selection method based on sequence information and band quality according to claim 4, wherein the step S6 is specifically as follows: and (4) storing the indexes of the spectral band sets selected in the step (S5) to obtain the spectral band subsets with low redundancy and high quality.
CN201810991085.2A 2018-08-28 2018-08-28 Hyperspectral image waveband selection method based on sequence information and waveband quality Active CN109191443B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810991085.2A CN109191443B (en) 2018-08-28 2018-08-28 Hyperspectral image waveband selection method based on sequence information and waveband quality

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810991085.2A CN109191443B (en) 2018-08-28 2018-08-28 Hyperspectral image waveband selection method based on sequence information and waveband quality

Publications (2)

Publication Number Publication Date
CN109191443A CN109191443A (en) 2019-01-11
CN109191443B true CN109191443B (en) 2022-03-04

Family

ID=64916722

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810991085.2A Active CN109191443B (en) 2018-08-28 2018-08-28 Hyperspectral image waveband selection method based on sequence information and waveband quality

Country Status (1)

Country Link
CN (1) CN109191443B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109871768B (en) * 2019-01-18 2022-04-29 西北工业大学 Hyperspectral optimal waveband selection method based on shared nearest neighbor
CN110766619A (en) * 2019-09-19 2020-02-07 北京航空航天大学 Unsupervised band selection algorithm based on band quality analysis
CN111031390B (en) * 2019-12-17 2022-10-21 南京航空航天大学 Method for summarizing process video of outputting determinant point with fixed size
CN113505846A (en) * 2021-07-26 2021-10-15 云南电网有限责任公司电力科学研究院 Hyperspectral band selection method based on mutual information

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106033545A (en) * 2015-03-10 2016-10-19 中国科学院西安光学精密机械研究所 Wave band selecting method for determinant point process
CN107220661A (en) * 2017-05-16 2017-09-29 沈阳航空航天大学 Spectral band system of selection based on multi-modal fusion

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101853392B (en) * 2010-04-21 2012-05-23 河海大学 Remote sensing hyperspectral image band selection method based on conditional mutual information
US9429476B2 (en) * 2011-06-03 2016-08-30 Frederick S. Solheim Correcting noncontact infrared thermometer data by removing contamination of the intervening atmosphere
CN104268582B (en) * 2014-08-26 2017-07-14 中国科学院遥感与数字地球研究所 The band selection method and device of a kind of high spectrum image
US10113910B2 (en) * 2014-08-26 2018-10-30 Digimarc Corporation Sensor-synchronized spectrally-structured-light imaging
CN104484681B (en) * 2014-10-24 2018-03-13 西安电子科技大学 Hyperspectral Remote Sensing Imagery Classification method based on spatial information and integrated study
CN107563324B (en) * 2017-08-30 2021-04-16 广东工业大学 Hyperspectral image classification method and device of ultralimit learning machine with composite nuclear structure

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106033545A (en) * 2015-03-10 2016-10-19 中国科学院西安光学精密机械研究所 Wave band selecting method for determinant point process
CN107220661A (en) * 2017-05-16 2017-09-29 沈阳航空航天大学 Spectral band system of selection based on multi-modal fusion

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
A New Band Selection Method for Hyperspectral Image Based on Data Quality;Kang Sun;《IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING》;20140630;第7卷(第6期);第2697-2703页 *

Also Published As

Publication number Publication date
CN109191443A (en) 2019-01-11

Similar Documents

Publication Publication Date Title
CN109191443B (en) Hyperspectral image waveband selection method based on sequence information and waveband quality
CN109948693B (en) Hyperspectral image classification method based on superpixel sample expansion and generation countermeasure network
CN111738124B (en) Remote sensing image cloud detection method based on Gabor transformation and attention
CN107515895B (en) Visual target retrieval method and system based on target detection
CN108596154B (en) Remote sensing image classification method based on high-dimensional feature selection and multilevel fusion
CN114926746B (en) SAR image change detection method based on multiscale differential feature attention mechanism
CN103440505B (en) The Classification of hyperspectral remote sensing image method of space neighborhood information weighting
CN107239759B (en) High-spatial-resolution remote sensing image transfer learning method based on depth features
CN103208011B (en) Based on average drifting and the hyperspectral image space-spectral domain classification method organizing sparse coding
CN109543723B (en) Robust image clustering method
CN112507901A (en) Unsupervised pedestrian re-identification method based on pseudo tag self-correction
CN107194936B (en) Hyperspectral image target detection method based on superpixel combined sparse representation
CN109359525B (en) Polarized SAR image classification method based on sparse low-rank discrimination spectral clustering
CN112633202B (en) Hyperspectral image classification algorithm based on dual denoising combined multi-scale superpixel dimension reduction
CN109784205B (en) Intelligent weed identification method based on multispectral inspection image
CN111783884A (en) Unsupervised hyperspectral image classification method based on deep learning
CN111062428A (en) Hyperspectral image clustering method, system and equipment
CN113408398A (en) Remote sensing image cloud detection method based on channel attention and probability up-sampling
CN106156728B (en) HYPERSPECTRAL IMAGERY dimension reduction method and system based on spectral space decomposition and noise contribution analysis
CN114693577A (en) Infrared polarization image fusion method based on Transformer
CN102024153A (en) Hyperspectral image supervised classification method
CN109460772B (en) Spectral band selection method based on information entropy and improved determinant point process
CN106101732B (en) The vector quantization scheme of Fast Compression bloom spectrum signal
CN113095218A (en) Hyperspectral image target detection algorithm
CN111666999A (en) Remote sensing image classification method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant