CN108460400B - Hyperspectral image classification method combining various characteristic information - Google Patents

Hyperspectral image classification method combining various characteristic information Download PDF

Info

Publication number
CN108460400B
CN108460400B CN201810002038.0A CN201810002038A CN108460400B CN 108460400 B CN108460400 B CN 108460400B CN 201810002038 A CN201810002038 A CN 201810002038A CN 108460400 B CN108460400 B CN 108460400B
Authority
CN
China
Prior art keywords
hyperspectral image
dictionary
hyperspectral
sparse
space
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
CN201810002038.0A
Other languages
Chinese (zh)
Other versions
CN108460400A (en
Inventor
杨明
张会敏
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.)
Nanjing Ciku Network Information Technology Co ltd
Original Assignee
Nanjing Normal University
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 Nanjing Normal University filed Critical Nanjing Normal University
Priority to CN201810002038.0A priority Critical patent/CN108460400B/en
Publication of CN108460400A publication Critical patent/CN108460400A/en
Application granted granted Critical
Publication of CN108460400B publication Critical patent/CN108460400B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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
    • G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
    • 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
    • G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
    • Y02A40/10—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in agriculture

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Multimedia (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses a hyperspectral image classification method combining multiple kinds of characteristic information, which comprises the following steps: step 1, extracting various characteristic data of spectrum, gradient, texture and shape of a hyperspectral image to be classified: step 2, segmenting the hyperspectral images to be classified by using a watershed segmentation algorithm, and dividing the hyperspectral images into a plurality of spatial neighborhood groups; step 3, learning by using an MFKSADL model to obtain a dictionary and sparse codes; and 4, training an SVM classifier by using the coding coefficient, and predicting the label of the hyperspectral image test set. The method can solve the problems of same-object different spectrums, same-spectrum foreign matters and the like in the hyperspectral image, and can effectively improve the classification precision of the hyperspectral image.

Description

Hyperspectral image classification method combining various characteristic information
Technical Field
The invention belongs to the field of hyperspectral image processing, and particularly relates to a hyperspectral image classification method combining various feature information.
Background
Each pixel point of the hyperspectral remote sensing image is represented by hundreds of spectral values, the spectral values correspond to different narrow wavelengths from a visible spectrum to infrared rays, and the values can provide fine spectral differences among different ground objects, so that the possibility is provided for detecting and distinguishing various ground objects with higher precision. Therefore, the hyperspectral image classification is widely applied in a plurality of fields, including environmental protection, land utilization monitoring, urban planning, fire detection in deep forests, atmospheric monitoring, military operations and the like. Abundant spectral information in the hyperspectral image also contains a plurality of challenges and problems, such as the problem of high-dimensional small sample classification, the phenomenon of 'same object, different spectrum, same spectrum and foreign matter', and the like.
The sparse representation method inspired by a sparse coding mechanism of a human visual system is applied to the field of hyperspectral image classification. Although the pixels of the hyperspectral image are high-dimensional signals, the pixels of the same kind are usually located in the same low-dimensional space and are represented by the same dictionary atom. Thus, each pixel may be sparsely represented by a few atoms in the entire training dictionary, with the sparsity factor containing the location and weight of the selected atom. The class label for each pixel may then be determined by the corresponding sparse coefficient. In the sparse representation method, the discriminative power of the dictionary is quite important. Using a training set as a dictionary may have two significant drawbacks. First, the raw data typically includes noise, which degrades classification performance. Second, the computational cost of finding sparse codes from a large-scale dictionary is high, which may limit its practical application. To address these issues, in recent years, learning a dictionary from training samples has proven to provide additional performance improvements for the classification task. The dictionary learning method improves performance by encoding coefficients and dictionaries. However, a general dictionary learning method is only based on single spectrum feature information, and high-level feature information in hyperspectrum is not considered.
A single spectral feature describes a hyperspectral image from only one perspective, while different types of features have different resolving power and can provide relevant and complementary information. The invention provides a novel dictionary learning algorithm based on multi-feature information to solve the problem of classification of hyperspectral images. The hyperspectral image is first divided into a number of spatial groups and various characteristic information data (spectrum, gradient, texture and shape) of the hyperspectral image are extracted. And then, learning to obtain a dictionary with discrimination and sparse codes based on the hypothesis constraint that the sparse codes corresponding to different types of feature data in the same space group have the same sparse mode. In addition, if the data of the hyperspectral image is linearly inseparable in the raw feature space or the feature encodings are similar, the kernel method can project the raw data into a higher dimensional space to improve separability.
Disclosure of Invention
The invention aims to provide a hyperspectral image classification method combining various characteristic information, which can solve the problems of same-object different spectrums, same-spectrum foreign matters and the like existing in a hyperspectral image and can effectively improve the hyperspectral image classification precision.
In order to achieve the above purpose, the solution of the invention is:
a hyperspectral image classification method combining multiple kinds of characteristic information comprises the following steps:
step 1, extracting various characteristic data of spectrum, gradient, texture and shape of a hyperspectral image to be classified:
step 2, segmenting the hyperspectral images to be classified by using a watershed segmentation algorithm, and dividing the hyperspectral images into a plurality of spatial neighborhood groups;
step 3, learning by using an MFKSADL model to obtain a dictionary and sparse codes;
and 4, training an SVM classifier by using the coding coefficient, and predicting the label of the hyperspectral image test set.
The details of the step 2 are as follows: firstly, extracting a gradient image of a hyperspectral image, then segmenting the gradient image by using a watershed segmentation algorithm to obtain a segmentation map of the hyperspectral image, and dividing the segmentation map into a plurality of space groups.
In step 3, the MFKSADL model is obtained by:
a, setting the hyperspectral image represented by the s-th class feature data
Figure BDA0001537431150000021
In the nuclear space is represented as
Figure BDA0001537431150000022
Wherein S is 1,2, …, and S represents the S-th feature; n is the number of pixel points; bsDimension of the s-th class feature; the pixel points in the same characteristic space are linearly expressed by other pixel points, and the expression model is as follows: phi (X)s)＝Φ(Ds)As，Φ(Ds) Representing the s-th class feature data dictionary in a kernel space; a. thesIs a corresponding coding matrix;
b, setting a hyperspectral image to be divided into a plurality of space groupsg1,…,gGS-th feature data representing a spatial group giMiddle pixel point set
Figure BDA0001537431150000023
Wherein | gi| represents a space group giThe number of middle pixel points and the corresponding expression coefficient matrix
Figure BDA0001537431150000024
And c, combining the step a and the step b to obtain the model:
Figure BDA0001537431150000031
Figure BDA0001537431150000032
wherein the content of the first and second substances,
Figure BDA0001537431150000033
is DsThe ith atom of (c).
In the step 3, the specific contents of the dictionary and the sparse code obtained by learning the MFKSADL model are as follows:
the model was transformed into the following equivalent forms:
Figure BDA0001537431150000034
Figure BDA0001537431150000035
definition of
Figure BDA0001537431150000036
Wherein a isiAnd bjIth and jth columns of matrices a and B, respectively; the specific steps of step 3 are:
step 31, initialize the image matrix X, dictionary { D }s}s＝1,…,S：
Step 32, fix dictionary { Ds}s＝1,…,SAnd updating the sparse code gamma, and solving the original problem in the following way for each space group in a separating way:
Figure BDA0001537431150000037
step 33, fixing the sparse code gamma, updating the dictionary { Ds}s＝1,…,S：
Figure BDA0001537431150000038
Figure BDA0001537431150000039
Step 34, repeating the steps 32-33 until an iteration termination condition is met, and outputting the dictionary { D obtained by learnings}s＝1,…,SAnd sparse coding Γ.
The details of the step 4 are as follows: training by using the sparse codes obtained in the step 3 corresponding to the training set to obtain a Support Vector Machine (SVM); inputting sparse codes of pixel points to be classified into a trained SVM classifier to obtain corresponding class labels; and classifying all the pixel points to be classified according to a method, and obtaining a final hyperspectral image classification result.
After the scheme is adopted, the invention is the image classification method specially provided for hyperspectral image processing. Compared with the prior art, the invention has the following characteristics:
firstly, the hyperspectral images are divided into space groups, and space information is combined for auxiliary classification, so that the space information of the hyperspectral images is well utilized, and the classification effect of the hyperspectral images is improved;
then, by fusing multiple kinds of characteristic information, the relevance and complementarity of the multiple kinds of characteristic information are effectively utilized, the classification accuracy is improved, and the classification robustness is enhanced;
moreover, a dictionary with higher discrimination capacity is learned by utilizing a representation theory combining with various characteristic information, the discrimination capacity of sparse coding is further improved, a class label of a sample to be classified is indirectly obtained by utilizing the sparse coding, and the classification precision of the hyperspectral image is improved. Meanwhile, the linear separability of the data is improved through a kernel method, and the problems of same-object different spectrums and same-spectrum foreign matters existing in a hyperspectral image are effectively solved, so that the method has high use value.
Drawings
FIG. 1 is an overall flow diagram of the present invention;
FIG. 2 is a flow chart of the present invention for using MFKSADL learning to derive dictionaries and sparse codes.
Detailed Description
The technical solution and the advantages of the present invention will be described in detail with reference to the accompanying drawings.
As shown in FIG. 1, the invention provides a hyperspectral image classification method combined with multiple feature information, which comprises the following steps:
step 1, extracting various characteristic data of spectrum, gradient, texture and shape of a hyperspectral image to be classified: and (4) extracting various hyperspectral characteristic information by utilizing the prior art to obtain sample data of different characteristic spaces, and laying a cushion for the step 3. The multiple kinds of characteristic information have relevance and complementarity, more effective information is provided for the correct classification of the hyperspectral images, and the classification precision is further improved.
Step 2, segmenting the hyperspectral images to be classified by using a watershed segmentation method, and dividing the hyperspectral images into a plurality of spatial neighborhood groups: firstly, extracting a gradient image of a hyperspectral image by using the prior art, then segmenting the gradient image by using a watershed segmentation algorithm to obtain a segmentation map of the hyperspectral image, and dividing the segmentation map into a plurality of space groups. The intra-group spatial information has consistency and contributes significantly to classification.
And 3, learning by using an MFKSADL model to obtain a dictionary and sparse codes: and iteratively updating the dictionary and the sparse codes until an iteration termination condition is met, and outputting learning to obtain the codes corresponding to the dictionary with discrimination and the sample set data.
Step 4, training an SVM classifier by using the coding coefficient, predicting a hyperspectral image test set label: and (4) training by using the sparse codes obtained in the step (3) corresponding to the training set to obtain the SVM (support vector machine). And inputting the sparse codes of the pixel points to be classified into the trained SVM classifier to obtain the corresponding class labels. And classifying all the pixel points to be classified according to a method, and obtaining a final hyperspectral image classification result.
It should be noted that the core steps of the present invention lie in obtaining the dictionary and sparse code by using the MFKSADL model learning, and the description of the specific embodiment mainly focuses on step 3, and steps 1,2 and step 4 can be implemented by using the prior art.
In step 3, the MFKSADL model can be obtained by:
(1) with hyperspectral image represented by class s characteristic data
Figure BDA0001537431150000051
In the nuclear space is represented as
Figure BDA0001537431150000052
Wherein S is 1,2, …, S represents the S-th feature; n is the number of pixel points; bsIs the dimension of the s-th class of features. The pixel points in the same characteristic space can have other pixel points to perform linear representation, and the representation model is as follows: phi (X)s)＝Φ(Ds)As，Φ(Ds) Representing the s-th class feature data dictionary in a kernel space; a. thesIs the corresponding coding matrix. In practical applications, the probability of complete reconstruction is relatively small, and we hope that the reconstruction error is small
Figure BDA0001537431150000053
As small as possible.
(2) Dividing the hyperspectral image into a plurality of space groups { g1,…,gGS-th feature data representing a spatial group giMiddle pixel point set
Figure BDA0001537431150000054
Wherein | giI represents a space group giAnd the number of the middle pixels. Corresponding matrix of representation coefficients
Figure BDA0001537431150000055
The probability that the pixel points in the same space group belong to the same class is high, and the pixel points can be linearly represented by similar dictionary atoms. And each pixel is formed by dictionary atoms similar to the pixel as much as possible. In order to utilize neighborhood space information, the representation coefficients corresponding to the pixel points in the same spatial group are considered to have row sparsity. Since different feature data corresponding to the same spatial group have information similarity, in order to effectively utilize complementary and related information of a plurality of kinds of feature information, it is considered that the representation coefficients of different feature data corresponding to the same spatial group have line sparsity.
(3) Combining the ideas in step (1) and step (2), such a model can be obtained:
Figure BDA0001537431150000056
Figure BDA0001537431150000057
wherein the content of the first and second substances,
Figure BDA0001537431150000058
Figure BDA0001537431150000059
is DsThe ith atom of (c). The space similarity is combined with various feature information, a dictionary with strong discriminability is learned, more detail information is effectively reserved, and the obtained corresponding sparse coding classification capability is higher.
As shown in fig. 2, the specific steps of learning and obtaining the dictionary and the sparse code by applying the MFKSADL model are as follows:
the model was transformed into the following equivalent forms:
Figure BDA0001537431150000061
Figure BDA0001537431150000062
here, we define
Figure BDA0001537431150000063
Wherein a isiAnd bjI-th and j-th columns of matrices a and B, respectively.
Step 31, initialize the image matrix X, dictionary { D }s}s＝1,…,S：
Step 32, fix dictionary { Ds}s＝1,…,SAnd updating the sparse code gamma, and solving the original problem in the following way for each space group in a separating way:
Figure BDA0001537431150000064
step 33, fixing the sparse code gamma, updating the dictionary { Ds}s＝1,…,S：
Figure BDA0001537431150000065
Figure BDA0001537431150000066
Step 34, repeating the steps 32-33 until an iteration termination condition is met, and outputting the dictionary { D obtained by learnings}s＝1,…,SAnd sparse coding Γ.
In summary, the invention provides a hyperspectral image classification method combined with various feature information, which adopts a Dictionary Learning model combined with MFKSADL (Multi feature Kernel Spatial-Aware Learning), makes full use of the correlation and complementarity of various feature information of a hyperspectral image, combines the Spatial neighborhood similarity of the hyperspectral image, and adopts a Dictionary Learning method, thereby providing a new hyperspectral image classification method. The method provided by the invention achieves the purpose of remarkably improving the classification precision in both subjective vision and objective evaluation indexes. In addition, the linear separability of the data is improved through a kernel method, and the problems of same-object different spectrums and same-spectrum foreign matters existing in a hyperspectral image are effectively solved, so that the method has high use value.
The above embodiments are only for illustrating the technical idea of the present invention, and the protection scope of the present invention is not limited thereby, and any modifications made on the basis of the technical scheme according to the technical idea of the present invention fall within the protection scope of the present invention.

Claims (3)

1. A hyperspectral image classification method combining multiple kinds of characteristic information is characterized by comprising the following steps:
step 1, extracting various characteristic data of spectrum, gradient, texture and shape of a hyperspectral image to be classified:
step 2, segmenting the hyperspectral images to be classified by using a watershed segmentation algorithm, and dividing the hyperspectral images into a plurality of spatial neighborhood groups;
step 3, learning by using an MFKSADL model to obtain a dictionary and sparse codes;
in step 3, the MFKSADL model is obtained by:
a, setting the hyperspectral image represented by the s-th class characteristic data
Figure FDA0003388109480000011
In the nuclear space is represented as
Figure FDA0003388109480000012
Wherein S is 1,2, …, and S represents the S-th feature; n is the number of pixel points; bsDimension of the s-th class feature; the pixel points in the same characteristic space are linearly expressed by other pixel points, and the expression model is as follows: phi (X)s)＝Φ(Ds)As，Φ(Ds) Representing the s-th class feature data dictionary in a kernel space; a. thesIs a corresponding coding matrix;
b, dividing the hyperspectral image into a plurality of space groups { g1,…,gGS-th feature data representing a spatial group giMiddle pixel point set
Figure FDA0003388109480000013
Wherein | giI represents a space group giThe number of middle pixel points and the corresponding expression coefficient matrix
Figure FDA0003388109480000014
And c, combining the step a and the step b to obtain the model:
Figure FDA0003388109480000015
Figure FDA0003388109480000016
wherein, the first and the second end of the pipe are connected with each other,
Figure FDA0003388109480000017
Figure FDA0003388109480000018
is DsThe ith atom of (a);
in step 3, the specific contents of the dictionary and the sparse code obtained by learning with the MFKSADL model are as follows:
the model was transformed into the following equivalent forms:
Figure FDA0003388109480000019
Figure FDA00033881094800000110
definition of
Figure FDA00033881094800000111
Wherein a isiAnd bjIth and jth columns of matrices a and B, respectively; the specific steps of step 3 are:
step 31, initialize the image matrix X, dictionary { D }s}s＝1,…,S：
Step 32, fix dictionary { Ds}s＝1,…,SAnd updating the sparse code gamma, and solving the original problem in the following way for each space group in a separating way:
Figure FDA0003388109480000021
step 33, fixing the sparse code gamma, updating the dictionary { Ds}s＝1,…,S：
Figure FDA0003388109480000022
Figure FDA0003388109480000023
Step 34, repeating the steps 32-33 until an iteration termination condition is met, and outputting the dictionary { D obtained by learnings}s＝1,…,SAnd sparse coding Γ;
and 4, training an SVM classifier by using the coding coefficient, and predicting the label of the hyperspectral image test set.
2. The hyperspectral image classification method combining multiple pieces of feature information according to claim 1, wherein: the details of the step 2 are as follows: firstly, extracting a gradient image of a hyperspectral image, then segmenting the gradient image by using a watershed segmentation algorithm to obtain a segmentation map of the hyperspectral image, and dividing the segmentation map into a plurality of space groups.
3. The hyperspectral image classification method combining multiple pieces of feature information according to claim 1, wherein: the details of the step 4 are as follows: training by using the sparse codes obtained in the step 3 corresponding to the training set to obtain a Support Vector Machine (SVM); inputting sparse codes of pixel points to be classified into a trained SVM classifier to obtain corresponding class labels; and classifying all the pixel points to be classified according to the method to obtain a final hyperspectral image classification result.
CN201810002038.0A 2018-01-02 2018-01-02 Hyperspectral image classification method combining various characteristic information Active CN108460400B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810002038.0A CN108460400B (en) 2018-01-02 2018-01-02 Hyperspectral image classification method combining various characteristic information

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810002038.0A CN108460400B (en) 2018-01-02 2018-01-02 Hyperspectral image classification method combining various characteristic information

Publications (2)

Publication Number Publication Date
CN108460400A CN108460400A (en) 2018-08-28
CN108460400B true CN108460400B (en) 2022-05-20

Family

ID=63220654

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810002038.0A Active CN108460400B (en) 2018-01-02 2018-01-02 Hyperspectral image classification method combining various characteristic information

Country Status (1)

Country Link
CN (1) CN108460400B (en)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109410223A (en) * 2018-11-07 2019-03-01 电子科技大学 A kind of SAR image segmentation method based on watershed algorithm and dictionary learning
CN110688890B (en) * 2019-08-13 2023-07-25 南京航空航天大学 Hyperspectral image classification method based on self-adaptive kernel sparse representation and multi-feature
CN112633045A (en) * 2019-10-09 2021-04-09 华为技术有限公司 Obstacle detection method, device, equipment and medium
CN111046844B (en) * 2019-12-27 2020-11-27 中国地质大学（北京） Hyperspectral image classification method based on neighborhood selection constraint
CN113065403A (en) * 2021-03-05 2021-07-02 浙江大学 Hyperspectral imaging-based machine learning cell classification method and device
CN115546790B (en) * 2022-11-29 2023-04-07 深圳智能思创科技有限公司 Document layout segmentation method, device, equipment and storage medium

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101930547A (en) * 2010-06-24 2010-12-29 北京师范大学 Method for automatically classifying remote sensing image based on object-oriented unsupervised classification
CN103208011A (en) * 2013-05-05 2013-07-17 西安电子科技大学 Hyperspectral image space-spectral domain classification method based on mean value drifting and group sparse coding
CN103593853A (en) * 2013-11-29 2014-02-19 武汉大学 Remote-sensing image multi-scale object-oriented classification method based on joint sparsity representation
CN103971123A (en) * 2014-05-04 2014-08-06 南京师范大学 Hyperspectral image classification method based on linear regression Fisher discrimination dictionary learning (LRFDDL)
CN104123555A (en) * 2014-02-24 2014-10-29 西安电子科技大学 Super-pixel polarimetric SAR land feature classification method based on sparse representation
CN104281855A (en) * 2014-09-30 2015-01-14 西安电子科技大学 Hyperspectral image classification method based on multi-task low rank
CN106022358A (en) * 2016-05-11 2016-10-12 湖南大学 Hyper-spectral image classification method and hyper-spectral image classification device
CN106778768A (en) * 2016-11-22 2017-05-31 广西师范大学 Image scene classification method based on multi-feature fusion
CN107358249A (en) * 2017-06-07 2017-11-17 南京师范大学 The hyperspectral image classification method of dictionary learning is differentiated based on tag compliance and Fisher
CN107527023A (en) * 2017-08-07 2017-12-29 西安理工大学 Classification of Polarimetric SAR Image method based on super-pixel and topic model

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9946931B2 (en) * 2015-04-20 2018-04-17 Los Alamos National Security, Llc Change detection and change monitoring of natural and man-made features in multispectral and hyperspectral satellite imagery

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101930547A (en) * 2010-06-24 2010-12-29 北京师范大学 Method for automatically classifying remote sensing image based on object-oriented unsupervised classification
CN103208011A (en) * 2013-05-05 2013-07-17 西安电子科技大学 Hyperspectral image space-spectral domain classification method based on mean value drifting and group sparse coding
CN103593853A (en) * 2013-11-29 2014-02-19 武汉大学 Remote-sensing image multi-scale object-oriented classification method based on joint sparsity representation
CN104123555A (en) * 2014-02-24 2014-10-29 西安电子科技大学 Super-pixel polarimetric SAR land feature classification method based on sparse representation
CN103971123A (en) * 2014-05-04 2014-08-06 南京师范大学 Hyperspectral image classification method based on linear regression Fisher discrimination dictionary learning (LRFDDL)
CN104281855A (en) * 2014-09-30 2015-01-14 西安电子科技大学 Hyperspectral image classification method based on multi-task low rank
CN106022358A (en) * 2016-05-11 2016-10-12 湖南大学 Hyper-spectral image classification method and hyper-spectral image classification device
CN106778768A (en) * 2016-11-22 2017-05-31 广西师范大学 Image scene classification method based on multi-feature fusion
CN107358249A (en) * 2017-06-07 2017-11-17 南京师范大学 The hyperspectral image classification method of dictionary learning is differentiated based on tag compliance and Fisher
CN107527023A (en) * 2017-08-07 2017-12-29 西安理工大学 Classification of Polarimetric SAR Image method based on super-pixel and topic model

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
Fast Multifeature Joint Sparse Representation for Hyperspectral Image Classification;Erlei Zhang 等;《IEEE GEOSCIENCE AND REMOTE SENSING LETTERS》;20150306;第12卷(第7期);第1397-1401页 *
Spatial-Aware Dictionary Learning for Hyperspectral Image Classification;Ali Soltani-Farani 等;《IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING》;20140702;第53卷(第1期);第527-541页 *
Weighted multifeature hyperspectral image classification via kernel joint sparse representation;Erlei Zhang 等;《Neurocomputing》;20151106;第71-86页 *
一种多特征转换的高光谱影像自适应分类方法;王凯 等;《武汉大学学报 信息科学版》;20150128;第40卷(第5期);第612-616页 *
基于分水岭分割和稀疏表示的高光谱图像分类方法;舒速 等;《计算机科学》;20160215;第43卷(第2期);第89-94页 *

Also Published As

Publication number Publication date
CN108460400A (en) 2018-08-28

Similar Documents

Publication Publication Date Title
CN108460400B (en) Hyperspectral image classification method combining various characteristic information
CN110321963B (en) Hyperspectral image classification method based on fusion of multi-scale and multi-dimensional space spectrum features
CN109375186B (en) Radar target identification method based on depth residual error multi-scale one-dimensional convolution neural network
CN104317902B (en) Image search method based on local holding iterative quantization Hash
CN107145836B (en) Hyperspectral image classification method based on stacked boundary identification self-encoder
Zhou et al. CIMFNet: Cross-layer interaction and multiscale fusion network for semantic segmentation of high-resolution remote sensing images
CN104331698B (en) Remote sensing type urban image extracting method
CN107563442B (en) Hyperspectral image classification method based on sparse low-rank regular graph tensor embedding
CN103208011B (en) Based on average drifting and the hyperspectral image space-spectral domain classification method organizing sparse coding
CN109871454B (en) Robust discrete supervision cross-media hash retrieval method
CN110929080A (en) Optical remote sensing image retrieval method based on attention and generation countermeasure network
CN111475622A (en) Text classification method, device, terminal and storage medium
CN112163114B (en) Image retrieval method based on feature fusion
CN109034213B (en) Hyperspectral image classification method and system based on correlation entropy principle
Duarte-Carvajalino et al. Multiscale representation and segmentation of hyperspectral imagery using geometric partial differential equations and algebraic multigrid methods
Sun et al. Band selection using sparse nonnegative matrix factorization with the thresholded earth’s mover distance for hyperspectral imagery classification
CN115019104A (en) Small sample remote sensing image classification method and system based on multi-source domain self-attention
Wu et al. Weighted Sparse Image Classification Based on Low Rank Representation.
Ni et al. Scene classification from remote sensing images using mid-level deep feature learning
CN114821340A (en) Land utilization classification method and system
CN108805280B (en) Image retrieval method and device
Low et al. Multi-modal aerial view object classification challenge results-PBVS 2023
CN115984585A (en) Attention mechanism-introduced multi-scale hyperspectral image feature extraction method
Dang et al. Spectral‐Spatial Attention Transformer with Dense Connection for Hyperspectral Image Classification
CN109886315A (en) A kind of Measurement of Similarity between Two Images method kept based on core

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
TR01 Transfer of patent right

Effective date of registration: 20240105

Address after: Room B1015, 10th Floor, North Campus Teaching Complex Building, Nanjing University Gulou Campus, No. 22 Hankou Road, Gulou District, Nanjing City, Jiangsu Province, 210093

Patentee after: NANJING CIKU NETWORK INFORMATION TECHNOLOGY CO.,LTD.

Address before: No. 122, Ninghai Road, Gulou District, Nanjing, Jiangsu 210023

Patentee before: NANJING NORMAL University

TR01 Transfer of patent right