CN101236106A - Light spectrum and spatial information bonded high spectroscopic data classification method - Google Patents
Light spectrum and spatial information bonded high spectroscopic data classification method Download PDFInfo
- Publication number
- CN101236106A CN101236106A CNA2008100561176A CN200810056117A CN101236106A CN 101236106 A CN101236106 A CN 101236106A CN A2008100561176 A CNA2008100561176 A CN A2008100561176A CN 200810056117 A CN200810056117 A CN 200810056117A CN 101236106 A CN101236106 A CN 101236106A
- Authority
- CN
- China
- Prior art keywords
- mei
- structural element
- expansion
- spatial information
- spectral data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Image Analysis (AREA)
Abstract
Disclosed is a hyperspectral data classification method which is combined spectrum and spatial information. The steps comprises (1) reading the hypersectral data, (2) confirming the minimum size of structural element, (3) calculating differentiation between picture elements in neighborhood of each structural element by extended mathematical morphology expansion and corrosion operation, (4) obtaining exponential value of morphology eccentricity by the extended expansion and the corrosion operation of step (3), (5), constantly repeating the above steps with the adding of the size of the structural element to achieve the maximum size of the structural element, (6), constantly updating the exponential value MEI of morphology eccentricity in iteration process via the obtained new value, and generating a final exponential value MEI of morphology eccentricity after the iteration process is finished, (7) realizing the extraction of the data characteristic by the image of the exponential value MEI of morphology eccentricity, namely generating ground object type information, and realizing sophisticated category of the ground object by a minimum-distance classifier. The method is an unsupervised classification method for hyperspectral ground object with strong stability, high reliability and high accuracy.
Description
(1) technical field
The present invention relates to a kind of high-spectral data sorting technique of utilizing spectrum and spatial information simultaneously, belong to high-spectral data disposal route and applied technical field, be applicable to that high-spectral data does not have the theoretical method and the application technical research of supervised classification.
(2) background technology
Hyperspectral imager is a kind of novel remote sensing load, its spectrum has tight, continuous characteristics, can write down the spectrum and the spatial information feature of tested same atural object simultaneously, the material that can not survey in broadband remote sensing originally can be detected in high-spectrum remote-sensing.Target detection and terrain classification are one of main directions of high-spectrum remote sensing data application, and the development of such technology can promote the application of high-spectral data greatly, and constantly expands the application degree of depth and the range of high-spectral data.
Mathematical morphology is as a kind of non-linear space information processing technology of classics, and purpose is to analyze the spatial relationship between the pixel.Mathematical morphology be one based on sets theory and topology, the shape of geometrical concept and the science of structure, at present, mathematical morphology has become high-efficiency method of image processing field, and become a strong mathematical theory, it successfully has been applied to diverse discipline, as mineralogy, medical diagnosis, computer vision, Digital Image Processing and pattern-recognition.Two basic operations are corrosion and expand that these two operations all are to be defined in bianry image at first, but have expanded in the gray level image at present in the mathematical morphology.In the gray scale mathematical morphology, image is used as the successive value process of aggregation.If (x y) is input picture to f, and (x y) is structural element to b, and is the subimage function.With structural element b gray scale expansion and the corrosion that image f carries out is defined as follows respectively:
d(x,y)=(fb)(x,y)=Max{f(x-s,y-t)+b(s,t)|(x-s),(y-t)∈D
f;(s,t)∈D
b}
e(x,y)=(fb)(x,y)=Min{f(x+s,y+t)-b(s,t)|(x+s),(y+t)∈D
f;(s,t)∈D
b}
Wherein, D
fAnd D
bBe respectively the field of definition of f and b, (x-s), (x+s), (y-t) and (y+t) must be in the field of definition of f, and also s and t must be in the field of definition of b.Topmost task is how to calculate maximum or minimum value in the neighborhood of each pixel in image in the gray scale morphology operations, and this size and shape with the structural element of definition is closely related.Easy for algorithm, easily realize, the general structural element of only considering to meet convex function, establish b usually and be square and b (s, t)=0, (s, t) ∈ D
b
In the gray scale mathematical morphology, the numerical values recited of pixel is carried out the calculating of maximum or minimum gradation value as ordering relation, and in high spectrum image, each pixel all is a multidimensional, can not simple and direct compare their size.Therefore, it is exactly to define a suitable ordering relation element in the N gt is sorted that mathematical morphology is expanded to challenge maximum in the high spectrum image, determines maximum/least member.At present, the expansion Mathematical Morphology Method has received the concern of research fields such as the extraction of high-spectral data end member, data dimensionality reduction and classification.
(3) summary of the invention
The objective of the invention is to propose the high-spectral data sorting technique of a kind of spectrum and spatial information combination, it has overcome existing atural object does not have supervised classification carries out terrain classification etc. from the single aspect of data spectrum or space or characteristic information deficiency, suppressed background influence effectively, it is that the high spectrum atural object that a kind of stability is strong, reliability is high, degree of accuracy is high does not have supervised classification method.
Technical solution of the present invention is: a kind of under target and background condition of unknown, and utilize spectrum and spatial information to realize the method for high-spectral data classification simultaneously.This method mainly is based on mathematics morphology theory, utilizes expansion to expand and the corrosion operation, and carries out the feature extraction of high-spectral data by calculating form eccentricity index, utilizes the minor increment method to realize the classification of high-spectral data at last.Form excentricity index is a rectangular projection divergence of utilizing the computational data as a result of expansion expansion and corrosion, thereby the type of ground objects feature extraction is come out, and in the present invention for fear of since in the morphology size of structural element to the restriction of algorithm performance, adopt the method for iteration, the size that changes structural element realizes.
The high-spectral data sorting technique of a kind of spectrum of the present invention and spatial information combination, its step is as follows:
(1) high-spectral data reads in;
(2) determine the minimum dimension of structural element;
(3) by the pixel differences in the expansion of expansion mathematical morphology and each structural element neighborhood of corrosion operational computations;
(4) calculate form eccentricity index (Morphological Eccentricity Index, MEI) value by step (3) expansion expansion and Corrosion results;
(5) above-mentioned steps increases constantly repetition with the size of structural element, up to the full-size that reaches structural element;
(6) form eccentricity index M EI value is brought in constant renewal in by the new value that obtains in iterative process, and final form eccentricity index M EI finishes the back in iterative process and produces;
(7) by the feature extraction of form eccentricity index M EI image realization data, promptly produce type of ground objects information, and realize the sophisticated category of atural object by minimum distance classifier.
Wherein, the structural element minimum dimension described in the step (2) is 3 * 3;
Wherein, step (3) described " by the pixel differences in the expansion of expansion mathematical morphology and each structural element neighborhood of corrosion operational computations ", its expansion mathematical morphology expansion and corrosion are operated and are defined as follows:
Arg_Max, arg_Min represent to make D reach respectively and arrive most and minimum pixel vectors; D is in order to determine the ordering relation according to the multi-C vector of target and background difference, the tolerance operator of a multi-C vector of introducing.This tolerance operator is calculated by each pixel accumulation distance in the structural element, is defined as follows:
Wherein, dist is the pointwise linear range of measuring N dimensional vector.In order to effectively utilize spectrum and the spatial information that high-spectral data provides, the present invention adopts the rectangular projection divergence, and (Orthogonal ProjectionDivergence OPD) calculates this distance, considers two N dimension spectral signal s
i=[s
I1, s
I2..., s
IN]
T, s
j=[s
J1, s
J2..., s
JN]
T, then N ties up spectral signal s
iAnd s
jBetween rectangular projection divergence OPD be expressed as:
Therefore, accumulation distance D can be according to the vector in the difference size ordering structure element of target and background.By above the analysis showed that, what expand to that the expansion results of high spectrum image obtains is pixel bigger with the background subtraction opposite sex in structural element, and what Corrosion results obtained is pixel similar to background in structural element.
Wherein, " calculating form eccentricity exponential quantity " described in the step (4) by step (3) expansion expansion and Corrosion results, its implication is described as follows: by expanding and Corrosion results in the rectangular projection divergence fusion structure element neighborhood, calculate form eccentricity index M EI, computing formula is as follows:
MEI(x,y)=OPD[d(x,y),e(x,y)]
Wherein, the full-size of structural element is 9 * 9 in the step (5).
Wherein, " form eccentricity index M EI value is brought in constant renewal in by the new value that obtains in iterative process " described in the step (6), its implication is described as follows: establish MEI
i(x y) is pixel point (x, MEI value y), the MEI that the i time iterative computation obtains
I-1(x, y) be the i-1 time iterative computation obtain (if replacement criteria is MEI for x, MEI value y)
i(x, y)<MEI
I-1(x, y), MEI then
i(x, y)=MEI
I-1(x, y).
Wherein, described in the step (7) " realize the feature extraction of data by form eccentricity index M EI image; promptly produce type of ground objects information; and realize the sophisticated category of atural object by minimum distance classifier ", its implication is described as follows: its type of ground objects classification is to utilize the minor increment method utilizing MEI to carry out finishing on the result images of feature extraction.Minimum distance classifier is a kind of special case of discriminant function type sorter, is based on the sorter that minimum distance criterion makes up, and it has certain advantage on calculating, and realizes on computers easily.Suppose to have defined R point, v at feature space
1, v
2..., v
RBe class w
1, w
2..., w
RSample mode, minimum distance classifier will treat that classification mode x is identified as its nearest sample mode place classification.Minimum distance classifier is defined as follows:
The present invention's advantage compared with prior art is: overcome traditional high-spectral data do not have supervised classification method from spectrum/space/the single aspect of feature space carries out the limitation of data processing, spectrum and spatial information that this method has utilized high-spectral data to provide have simultaneously been realized the type of ground objects sophisticated category under the no supervision situation.It has following advantage: (1) based on mathematics morphology theory and spectrum similarity measurement principle, and the spectrum and the spatial information that have utilized high-spectral data to provide have simultaneously improved the reliability of algorithm; (2) measure by the otherness of expanding morphology expansion and corrosion operation realization target and background; (3) utilize the rectangular projection divergence to merge expansion and expand and Corrosion results, calculate form eccentricity index, realized the high-spectral data feature extraction, suppressed of the influence of factors such as background effectively the feature extraction result; (4) adopt the method for iteration, the size that changes structural element overcome since in the morphology size of structural element to the restriction of algorithm performance.
(4) description of drawings
Fig. 1 is expansion mathematics morphological operation result in the structural element, wherein:
Fig. 1 (a) is expansion expansive working result of the present invention;
Fig. 1 (b) is an expansion Corrosion results of the present invention.
Fig. 2 (a) is the terrestrial information reference picture of use data of the present invention;
Fig. 2 (b) is the terrain classification result of the inventive method.
Symbol description is as follows among the figure:
The b-structural element; Expansion expansion results in the d-structural element neighborhood;
Expansion Corrosion results in the e-structural element neighborhood.
C4-represents paddy rice; V2-represents Ipomoea batatas; V13-represents caraway; W2-represents the pond;
T6, T7-represent the trees of different cultivars.
(5) embodiment
The spectrum that the present invention relates to for better explanation and the high-spectral data sorting technique of spatial information combination utilize PHI aviation high-spectral data to carry out area, Fang Lu tea plantation, Jiangsu crops sophisticated category.The high-spectral data sorting technique of a kind of spectrum of the present invention and spatial information combination, the specific implementation step is as follows:
(1) reading in of high-spectral data: read in Fang Lu tea plantation PHI high-spectral data;
(2) determine the minimum dimension of structural element: according to data and algorithm characteristic, the structural element minimum dimension is 3 * 3;
(3) by the pixel differences in the expansion of expansion mathematical morphology and each structural element neighborhood of corrosion operational computations;
In order to realize reliable more, stable, accurate high-spectral data classification, the method among the present invention has been utilized spectrum and spatial information simultaneously, has introduced expansion of expansion mathematical morphology and corrosion operation thus, is defined as follows:
Wherein, arg_Max, arg_Min represent to make D reach respectively and arrive most and minimum pixel vectors; D is in order to determine the ordering relation according to the multi-C vector of target and background difference, the tolerance operator of a multi-C vector of introducing.This tolerance operator is calculated by each pixel accumulation distance in the structural element, is defined as follows:
Wherein, dist is the pointwise linear range of measuring N dimensional vector.In order to effectively utilize spectrum and the spatial information that high-spectral data provides, the present invention adopts rectangular projection divergence OPD to calculate this distance, considers two N dimension spectral signal s
i=[s
I1, s
I2..., s
IN]
T, s
j=[s
J1, s
J2..., s
JN]
T, then N ties up spectral signal s
iAnd s
jBetween rectangular projection divergence OPD be expressed as:
Wherein,
, k=i, j, and I
N * NIt is the unit matrix of N*N dimension.
Therefore, accumulation distance D can be according to the vector in the difference size ordering structure element of target and background.By above the analysis showed that, what expand to that the expansion results of high spectrum image obtains is pixel bigger with the background subtraction opposite sex in structural element, what Corrosion results obtained is pixel similar to background in structural element, and the expansion mathematical morphology expands and corrodes operational computations shown in Fig. 1 (a) and Fig. 1 (b).
(4) calculate form eccentricity exponential quantity by step (3) expansion expansion and Corrosion results;
To utilize single expansion or artificial noise and the background influence of introducing of corrosion operation in order overcoming, and, to have adopted form eccentricity index for the otherness between quantification target and the background.By expanding and Corrosion results in the rectangular projection divergence fusion structure element neighborhood, calculate form eccentricity index M EI, realize the feature extraction of high-spectral data.
MEI(x,y)=OPD[d(x,y),e(x,y)]
(5) above-mentioned steps increases constantly with the size of structural element and repeats, and up to the full-size that reaches structural element: the full-size of structural element is chosen as 9 * 9 can reach classifying quality preferably;
(6) form eccentricity index M EI value is brought in constant renewal in by the new value that obtains in iterative process, and final form eccentricity index M EI finishes the back in iterative process and produces;
If EMI
i(x y) is pixel point (x, MEI value y), the MEI that the i time iterative computation obtains
I-1(x, y) be the i-1 time iterative computation obtain (if replacement criteria is MEI for x, MEI value y)
i(x, y)<MEI
I-1(x, y), MEI then
i(x, y)=MEI
I-1(x, y).
(7) by the feature extraction of form eccentricity index M EI image realization data, promptly produce type of ground objects information, and realize the sophisticated category of atural object by minimum distance classifier.
Type of ground objects classification is to utilize the minor increment method utilizing MEI to carry out finishing on the result images of feature extraction.Minimum distance classifier is a kind of special case of discriminant function type sorter, is based on the sorter that minimum distance criterion makes up, and it has certain advantage on calculating, and realizes on computers easily.Suppose to have defined R point, v at feature space
1, v
2..., v
RBe class w
1, w
2..., w
RSample mode, minimum distance classifier will treat that classification mode x is identified as its nearest sample mode place classification.Minimum distance classifier is defined as follows:
High-spectral data sorting technique by a kind of spectrum of the present invention and spatial information combination, utilize PHI aviation high-spectral data to carry out area, Fang Lu tea plantation, Jiangsu crops sophisticated category, Fig. 2 (a) has provided the ground reference information of data, wherein, reference identification C4 represents paddy rice, and V2 represents Ipomoea batatas, and V13 represents caraway, W2 represents the pond, and it is background that T6, T7 represent the trees of different cultivars.Fig. 2 (b) has provided the unsupervised classification result of the method that the present invention relates to, and has realized the high-spectral data crops sophisticated category under the situation of any terrestrial information of the unknown.By the comparison and analysis of Fig. 2 (a) and Fig. 2 (b), the present invention relates to as can be seen not have misclassification in the classification results of method, nicety of grading is higher.
Claims (7)
1, the high-spectral data sorting technique of a kind of spectrum and spatial information combination, it is characterized in that: it comprises following steps:
(1) high-spectral data reads in;
(2) determine the minimum dimension of structural element;
(3) by the pixel differences in the expansion of expansion mathematical morphology and each structural element neighborhood of corrosion operational computations;
(4) calculate form eccentricity exponential quantity by step (3) expansion expansion and Corrosion results;
(5) above-mentioned steps increases constantly repetition with the size of structural element, up to the full-size that reaches structural element;
(6) form eccentricity exponential quantity is brought in constant renewal in by the new value that obtains in iterative process, and final form eccentricity index finishes the back in iterative process and produces;
(7) by the feature extraction of form eccentricity index image realization data, promptly produce type of ground objects information, and realize the sophisticated category of atural object by minimum distance classifier.
2, the high-spectral data sorting technique of a kind of spectrum according to claim 1 and spatial information combination is characterized in that: the structural element minimum dimension described in the step (2) is 3 * 3.
3, the high-spectral data sorting technique of a kind of spectrum according to claim 1 and spatial information combination, it is characterized in that: " by the pixel differences in the expansion of expansion mathematical morphology and each structural element neighborhood of corrosion operational computations " described in the step (3), its expansion mathematical morphology expands and corrosion is operated as follows:
Wherein, arg_Max, arg_Min represent to make D reach respectively and arrive most and minimum pixel vectors; D is in order to determine the ordering relation according to the multi-C vector of target and background difference, the tolerance operator of a multi-C vector of introducing; This tolerance operator is calculated by each pixel accumulation distance in the structural element, is defined as follows:
Wherein, dist is the pointwise linear range of measuring N dimensional vector; In order to effectively utilize spectrum and the spatial information that high-spectral data provides, the present invention adopts the rectangular projection divergence to calculate this distance, considers two N dimension spectral signal s
i=[s
I1, s
I2..., s
IN]
T, s
j=[s
J1, s
J2..., s
JN]
T, then N ties up spectral signal s
iAnd s
jBetween rectangular projection divergence OPD be expressed as:
Wherein,
, k=i, j, and I
N * NIt is the unit matrix of N*N dimension.
4, the high-spectral data sorting technique that combines of a kind of spectrum according to claim 1 and spatial information, it is characterized in that: " the calculating form eccentricity exponential quantity " described in its step (4) by step (3) expansion expansion and Corrosion results, its implication is described as follows: by expanding and Corrosion results in the rectangular projection divergence fusion structure element neighborhood, calculate form eccentricity index M EI, computing formula is: and MEI (x, y)=OPD[d (x, y), e (x, y)].
5, the high-spectral data sorting technique that combines of a kind of spectrum according to claim 1 and spatial information is characterized in that: the full-size of structural element is 9 * 9 in its step (5).
6, the high-spectral data sorting technique that combines of a kind of spectrum according to claim 1 and spatial information, it is characterized in that: " form eccentricity index M EI value is brought in constant renewal in by the new value that obtains in iterative process " described in the step (6), its implication is described as follows: establish MEI
i(x y) is pixel point (x, MEI value y), the MEI that the i time iterative computation obtains
I-1(x, y) be the i-1 time iterative computation obtain (if replacement criteria is MEI for x, MEI value y)
i(x, y)<MEI
I-1(x, y), MEI then
i(x, y)=MEI
I-1(x, y).
7, the high-spectral data sorting technique that combines of a kind of spectrum according to claim 1 and spatial information, it is characterized in that: described in the step (7) " realize the feature extraction of data by form eccentricity index M EI image; promptly produce type of ground objects information; and realize the sophisticated category of atural object by minimum distance classifier ", its implication is described as follows: its type of ground objects classification is to utilize the minor increment method utilizing MEI to carry out finishing on the result images of feature extraction; Suppose to have defined R point, v at feature space
1, v
2..., v
RBe class w
1, w
2..., w
RSample mode, minimum distance classifier will treat that classification mode x is identified as its nearest sample mode place classification, minimum distance classifier is defined as:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN200810056117A CN100590402C (en) | 2008-01-11 | 2008-01-11 | Light spectrum and spatial information bonded high spectroscopic data classification method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN200810056117A CN100590402C (en) | 2008-01-11 | 2008-01-11 | Light spectrum and spatial information bonded high spectroscopic data classification method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN101236106A true CN101236106A (en) | 2008-08-06 |
CN100590402C CN100590402C (en) | 2010-02-17 |
Family
ID=39919883
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN200810056117A Expired - Fee Related CN100590402C (en) | 2008-01-11 | 2008-01-11 | Light spectrum and spatial information bonded high spectroscopic data classification method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN100590402C (en) |
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101504315B (en) * | 2009-02-23 | 2011-01-05 | 北京航空航天大学 | Expansion morphology and orthogonal subspace projection combined end member automatic extraction method |
CN102073879A (en) * | 2010-12-02 | 2011-05-25 | 南京大学 | Method for identifying characteristic land categories of ocean remote sensing images of coast on basis of semi-supervised learning |
CN102314677A (en) * | 2010-07-07 | 2012-01-11 | 中国科学院地理科学与资源研究所 | Classification-based high time definition and high space definition remote sensing data quantitative fusing method |
CN102521601A (en) * | 2011-11-17 | 2012-06-27 | 西安电子科技大学 | Method for classifying hyperspectral images based on semi-supervised conditional random field |
CN102521603A (en) * | 2011-11-17 | 2012-06-27 | 西安电子科技大学 | Method for classifying hyperspectral images based on conditional random field |
CN102592134A (en) * | 2011-11-28 | 2012-07-18 | 北京航空航天大学 | Multistage decision fusing and classifying method for hyperspectrum and infrared data |
CN102829867A (en) * | 2012-08-22 | 2012-12-19 | 哈尔滨工业大学 | Simulation method of hyperspectral remote sensing data embedded based on target empty spectral information |
CN102938072A (en) * | 2012-10-20 | 2013-02-20 | 复旦大学 | Dimension reducing and sorting method of hyperspectral imagery based on blocking low rank tensor analysis |
CN103197017A (en) * | 2013-05-02 | 2013-07-10 | 云南烟草科学研究院 | Cigarette smoke colour spectrum baseline correction MPLS (mathematically proven learning system) method |
CN103927756A (en) * | 2014-04-28 | 2014-07-16 | 中国国土资源航空物探遥感中心 | Spectral characteristic index extraction method based on spectral characteristic space centralization |
CN104318252A (en) * | 2014-11-02 | 2015-01-28 | 西安电子科技大学 | Hyperspectral image classification method based on stratified probability model |
CN105427319A (en) * | 2015-11-27 | 2016-03-23 | 杭州电子科技大学 | Automated morphological endmember extraction based hyperspectral image data unmixing method |
CN106408034A (en) * | 2016-10-17 | 2017-02-15 | 大连海事大学 | Hyperspectral image classification method based on spatial feature iteration |
CN109271874A (en) * | 2018-08-23 | 2019-01-25 | 广东工业大学 | A kind of high spectrum image feature extracting method merging spatially and spectrally information |
CN109446965A (en) * | 2018-10-22 | 2019-03-08 | 北京师范大学 | A kind of tobacco field automatic identifying method based on Unmanned Aerial Vehicle Data |
WO2019218313A1 (en) * | 2018-05-17 | 2019-11-21 | 五邑大学 | Progressive dynamic hyperspectral image classification method |
CN112434685A (en) * | 2021-01-28 | 2021-03-02 | 中国科学院空天信息创新研究院 | Remote sensing image information fine extraction method based on earth surface elements |
-
2008
- 2008-01-11 CN CN200810056117A patent/CN100590402C/en not_active Expired - Fee Related
Cited By (27)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101504315B (en) * | 2009-02-23 | 2011-01-05 | 北京航空航天大学 | Expansion morphology and orthogonal subspace projection combined end member automatic extraction method |
CN102314677B (en) * | 2010-07-07 | 2014-03-26 | 中国科学院地理科学与资源研究所 | Classification-based high time definition and high space definition remote sensing data quantitative fusing method |
CN102314677A (en) * | 2010-07-07 | 2012-01-11 | 中国科学院地理科学与资源研究所 | Classification-based high time definition and high space definition remote sensing data quantitative fusing method |
CN102073879A (en) * | 2010-12-02 | 2011-05-25 | 南京大学 | Method for identifying characteristic land categories of ocean remote sensing images of coast on basis of semi-supervised learning |
CN102521603B (en) * | 2011-11-17 | 2013-09-25 | 西安电子科技大学 | Method for classifying hyperspectral images based on conditional random field |
CN102521601B (en) * | 2011-11-17 | 2013-12-25 | 西安电子科技大学 | Method for classifying hyperspectral images based on semi-supervised conditional random field |
CN102521603A (en) * | 2011-11-17 | 2012-06-27 | 西安电子科技大学 | Method for classifying hyperspectral images based on conditional random field |
CN102521601A (en) * | 2011-11-17 | 2012-06-27 | 西安电子科技大学 | Method for classifying hyperspectral images based on semi-supervised conditional random field |
CN102592134A (en) * | 2011-11-28 | 2012-07-18 | 北京航空航天大学 | Multistage decision fusing and classifying method for hyperspectrum and infrared data |
CN102829867A (en) * | 2012-08-22 | 2012-12-19 | 哈尔滨工业大学 | Simulation method of hyperspectral remote sensing data embedded based on target empty spectral information |
CN102938072A (en) * | 2012-10-20 | 2013-02-20 | 复旦大学 | Dimension reducing and sorting method of hyperspectral imagery based on blocking low rank tensor analysis |
CN102938072B (en) * | 2012-10-20 | 2016-05-11 | 复旦大学 | A kind of high-spectrum image dimensionality reduction and sorting technique based on the tensor analysis of piecemeal low-rank |
CN103197017A (en) * | 2013-05-02 | 2013-07-10 | 云南烟草科学研究院 | Cigarette smoke colour spectrum baseline correction MPLS (mathematically proven learning system) method |
CN103927756A (en) * | 2014-04-28 | 2014-07-16 | 中国国土资源航空物探遥感中心 | Spectral characteristic index extraction method based on spectral characteristic space centralization |
CN103927756B (en) * | 2014-04-28 | 2017-04-19 | 中国国土资源航空物探遥感中心 | Spectral characteristic index extraction method based on spectral characteristic space centralization |
CN104318252A (en) * | 2014-11-02 | 2015-01-28 | 西安电子科技大学 | Hyperspectral image classification method based on stratified probability model |
CN104318252B (en) * | 2014-11-02 | 2017-11-14 | 西安电子科技大学 | Hyperspectral image classification method based on stratified probability model |
CN105427319B (en) * | 2015-11-27 | 2018-10-02 | 杭州电子科技大学 | A kind of hyperspectral image data solution mixing method based on automatic morphology Endmember extraction |
CN105427319A (en) * | 2015-11-27 | 2016-03-23 | 杭州电子科技大学 | Automated morphological endmember extraction based hyperspectral image data unmixing method |
CN106408034A (en) * | 2016-10-17 | 2017-02-15 | 大连海事大学 | Hyperspectral image classification method based on spatial feature iteration |
WO2019218313A1 (en) * | 2018-05-17 | 2019-11-21 | 五邑大学 | Progressive dynamic hyperspectral image classification method |
CN109271874A (en) * | 2018-08-23 | 2019-01-25 | 广东工业大学 | A kind of high spectrum image feature extracting method merging spatially and spectrally information |
CN109271874B (en) * | 2018-08-23 | 2022-02-11 | 广东工业大学 | Hyperspectral image feature extraction method fusing spatial and spectral information |
CN109446965A (en) * | 2018-10-22 | 2019-03-08 | 北京师范大学 | A kind of tobacco field automatic identifying method based on Unmanned Aerial Vehicle Data |
CN109446965B (en) * | 2018-10-22 | 2021-07-13 | 北京师范大学 | Automatic tobacco field identification method based on unmanned aerial vehicle data |
CN112434685A (en) * | 2021-01-28 | 2021-03-02 | 中国科学院空天信息创新研究院 | Remote sensing image information fine extraction method based on earth surface elements |
CN112434685B (en) * | 2021-01-28 | 2021-05-28 | 中国科学院空天信息创新研究院 | Remote sensing image information fine extraction method based on earth surface elements |
Also Published As
Publication number | Publication date |
---|---|
CN100590402C (en) | 2010-02-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN100590402C (en) | Light spectrum and spatial information bonded high spectroscopic data classification method | |
Li et al. | Automatic pavement crack detection by multi-scale image fusion | |
Jia et al. | Feature mining for hyperspectral image classification | |
CN101980250B (en) | Method for identifying target based on dimension reduction local feature descriptor and hidden conditional random field | |
US20030068082A1 (en) | Systems and methods for automatic scale selection in real-time imaging | |
Bhattacharya et al. | Outlier detection using neighborhood rank difference | |
Guilherme et al. | Petroleum well drilling monitoring through cutting image analysis and artificial intelligence techniques | |
Vakalopoulou et al. | Graph-based registration, change detection, and classification in very high resolution multitemporal remote sensing data | |
Gesto-Diaz et al. | Feature matching evaluation for multimodal correspondence | |
Carneiro et al. | Flexible spatial configuration of local image features | |
Zhu et al. | Advances and challenges in multimodal remote sensing image registration | |
Lan et al. | An edge-located uniform pattern recovery mechanism using statistical feature-based optimal center pixel selection strategy for local binary pattern | |
Zhu et al. | Parallel multi-temporal remote sensing image change detection on GPU | |
Celik et al. | Change detection without difference image computation based on multiobjective cost function optimization | |
He et al. | Anomaly detection in hyperspectral imagery based on maximum entropy and nonparametric estimation | |
Feng et al. | Target recognition of SAR images via hierarchical fusion of complementary features | |
Sadeghian et al. | Automatic target recognition using discrimination based on optimal transport | |
WANG et al. | A deep double-channel dense network for hyperspectral image classification | |
Gupta et al. | Unsupervised change detection in optical satellite images using binary descriptor | |
González et al. | Multi-Resolution spectral graph matching | |
Baum et al. | OSPA barycenters for clustering set-valued data | |
Jiang et al. | Hyperspectral image classification using set-to-set distance | |
Gupta et al. | Spatial neighborhood mutual information based satellite image change detection | |
Wan et al. | Squeeze excitation densely connected residual convolutional networks for specific emitter identification based on measured signals | |
Li et al. | Graph construction for hyperspectral data unmixing |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
C17 | Cessation of patent right | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20100217 Termination date: 20140111 |