CN110428369B - CHNMF remote sensing image unmixing method based on information entropy - Google Patents

CHNMF remote sensing image unmixing method based on information entropy Download PDF

Info

Publication number
CN110428369B
CN110428369B CN201910538105.5A CN201910538105A CN110428369B CN 110428369 B CN110428369 B CN 110428369B CN 201910538105 A CN201910538105 A CN 201910538105A CN 110428369 B CN110428369 B CN 110428369B
Authority
CN
China
Prior art keywords
remote sensing
sensing image
information entropy
chnmf
unmixing
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.)
Expired - Fee Related
Application number
CN201910538105.5A
Other languages
Chinese (zh)
Other versions
CN110428369A (en
Inventor
李杏梅
刘晓杰
王心宇
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.)
China University of Geosciences
Original Assignee
China University of Geosciences
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 China University of Geosciences filed Critical China University of Geosciences
Priority to CN201910538105.5A priority Critical patent/CN110428369B/en
Publication of CN110428369A publication Critical patent/CN110428369A/en
Application granted granted Critical
Publication of CN110428369B publication Critical patent/CN110428369B/en
Expired - Fee Related 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
    • G06T5/00—Image enhancement or restoration
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/70—Denoising; Smoothing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10032—Satellite or aerial image; Remote sensing

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

The invention provides a CHNMF remote sensing image unmixing method based on information entropy, which comprises the following steps: s1, establishing a remote sensing image unmixing algorithm based on NMF for the current remote sensing image; s2, establishing a CNMF remote sensing image unmixing algorithm based on smooth constraint for the current remote sensing image according to the NMF-based remote sensing image unmixing algorithm to obtain L of the CNMF remote sensing image unmixing algorithm based on smooth constraint2A norm; s3, acquiring the information entropy of the current remote sensing image; s4, for the remote sensing image unmixing algorithm based on NMF, using the L obtained in S22And (3) the norm is used as a regularization function to constrain an end-member spectrum matrix M, the information entropy obtained in S3 is used as a regularization function to constrain an abundance matrix S, and the CHNMF remote sensing image unmixing method based on the information entropy is established. Compared with the traditional algorithm, the method has better unmixing effect.

Description

CHNMF remote sensing image unmixing method based on information entropy
Technical Field
The invention relates to a CHNMF remote sensing image unmixing method based on information entropy.
Background
The remote sensing technology is a novel technology which intersects with multiple disciplines such as detection, information detection and the like. With the continuous development of imaging spectrometers, image analysis technology is more and more deep, and the development of remote sensing technology has raised a hot trend, gained the favor of the majority of scientific research personnel, and occupied a place in each field.
The prototype of the remote sensing technology was an imaging spectrometer development project, which was first developed by Jet Propulsion Lab (JPL). The first generation of high resolution aerial imaging spectrometers in the world, AIS-1, was born in the United states in 1983. The second generation hyperspectral imager, AVIRIS, was also successfully developed in the United states four years later. From the eighties of the last century to the present, the development of imaging spectrometers has been emphasized at home and abroad, and the development of imaging spectrometers shows a flourishing trend. In foreign countries, CASI, AISI, Hymap and other series of spectrometers are developed in canada, finland, usa and other countries, and are in the leading position in the field of spectrometers. In China, excellent devices such as thermal infrared spectrum, multiband spectrum, multispectral scanner and the like are developed at present, the gap with the advanced level of the world is reduced, and the cross-over development of the spectrometer in China is realized.
Automatic acquisition of end members becomes very difficult due to the susceptibility of the remote sensing data to environmental interference and the limited spectral library. Therefore, a blind decomposition method is generally adopted, that is, the remote sensing image is decomposed under the condition that other information is known, so as to obtain the end-member spectrum and the abundance value of the remote sensing image.
The unsupervised mixed pixel decomposition method roughly comprises three types: independent component analysis, nonnegative matrix factorization, and complexity analysis. The non-negative matrix factorization method is mainly described herein.
NMF was first reported in an article by Paatero and Tapper, 1994. In the 1999 Lee and Seung article of Nature, a non-Negative Matrix Factorization (NMF) algorithm was proposed, which is a non-negative linear factorization of the original matrix in the case where all its elements are non-negative.
The NMF algorithm is put forward formally and developed rapidly for decades, and reaches a mature stage, and the application range of the NMF algorithm is expanded to the fields of image processing, data mining, voice processing and the like. In recent years, many scholars introduce NMF into the problem of unsupervised remote sensing image unmixing, and achieve certain results. However, later researchers found that in practical applications, the non-negative constraint of only non-negative matrix factorization is far from sufficient, and in order to obtain an ideal factorization value, Pauca and Piper et al proposed a non-negative matrix factorization algorithm (CNMF) with a smooth constraint; pixels in the Zymnis and Kim and other combination fields have similar characteristics, and a non-negative matrix factorization algorithm (APS-NMF) of interactive projection gradient is provided; miao and Qi et al propose an algorithm (MVC-NMF) that combines minimum volume constraints and NMF; wu Bo and Zhao Ying et al propose a non-negative matrix factorization algorithm with the spectral difference of end members as a limiting condition based on the characteristics of mixed pixels; liuxue pine and Wanbin et al propose non-negative matrix factorization algorithms for abundance separability and smoothness constraints; the non-negative matrix factorization algorithm under the constraint of weighted end members is provided by the people of the Fukuai and Liu enter army and the like by taking the weighted distance sum of each vertex of the simplex to the data center as the limit. The optimization algorithms improve some defects existing in remote sensing image unmixing and improve the accuracy of unmixing.
However, the optimization algorithms only consider the spectral information and the spatial information of the remote sensing data, and mostly introduce L1、L2The algorithm is optimized by the limitation of the equal norm, and the physical information of the remote sensing data is not considered.
Disclosure of Invention
The invention aims to solve the technical problem that the physical information of remote sensing data is not considered in the current optimization algorithm, and provides a CHNMF remote sensing image unmixing method based on information entropy to solve the technical defects.
The CHNMF remote sensing image unmixing method based on the information entropy comprises the following steps:
s1, establishing a remote sensing image unmixing algorithm based on NMF for the current remote sensing image;
s2, establishing a CNMF remote sensing image unmixing algorithm based on smooth constraint for the current remote sensing image according to the NMF-based remote sensing image unmixing algorithm to obtain L of the CNMF remote sensing image unmixing algorithm based on smooth constraint2A norm;
s3, acquiring the information entropy of the current remote sensing image;
s4, for the remote sensing image unmixing algorithm based on NMF, using the L obtained in S22And (3) the norm is used as a regularization function to constrain an end-member spectrum matrix M, the information entropy obtained in S3 is used as a regularization function to constrain an abundance matrix S, and the CHNMF remote sensing image unmixing method based on the information entropy is established.
Further, step S2 specifically includes:
according to the theory of sparse representation, with L2And the norm restrains an end member spectrum matrix and an abundance matrix of the remote sensing image unmixing algorithm based on the NMF to obtain the CNMF remote sensing image unmixing algorithm based on the smooth constraint.
Further, step S3 specifically includes:
assume that it is currentThe remote sensing image information source has n values: u shape1...Ui...UnThe corresponding probability is: p is a radical of1...pi...pnAnd the occurrence of each symbol is independent, the information entropy of the current remote sensing image information source is as follows:
Figure GDA0003235971830000021
further, step S4 specifically includes:
s41, establishing an objective function of the CHNMF remote sensing image unmixing method based on the information entropy:
Figure GDA0003235971830000031
Figure GDA0003235971830000032
wherein M is an end member spectrum matrix, S is an abundance matrix, a first term represents reconstruction errors, a second term represents sparse of the abundance matrix by using information entropy, a third term introduces smoothness limitation into the end member spectrum matrix, and lambda and phi are regularization parameters; sijRepresenting the proportion of corresponding end members in the pixels for each element, wherein L is the number of wave bands of the remote sensing image, R is the remote sensing image with L wave bands, P is the number of end members of the remote sensing image to be detected, and N is the number of pixel points of the remote sensing image to be detected;
s42, solving the end-member spectrum matrix M and the abundance matrix S by adopting a multiplicative iteration rule, and solving partial derivatives of the M and the S according to the properties of the matrixes to obtain:
Figure GDA0003235971830000033
Figure GDA0003235971830000034
then, iteration is carried out by using a gradient descent method to obtain a final multiplicative iteration rule of M and S:
M←M*RST/(MSST+ε)
Figure GDA0003235971830000035
and (3) the fraction is constant to be a positive number by using a small positive number epsilon, and when iteration is carried out for a certain number of times, the change value of f (M, S) is smaller than a preset value, so that the final optimized CHNMF remote sensing image unmixing method based on the information entropy is obtained.
Compared with the prior art, the invention has the beneficial effects that: aiming at the characteristic that end members of a remote sensing image are unevenly distributed, physical information of remote sensing data is mined, norm and information entropy are used for respectively restraining an end member spectrum matrix M and an abundance matrix S, the CHNMF remote sensing image unmixing method based on the information entropy is provided, and compared with the traditional NMF and CNMF algorithms, the CHNMF remote sensing image unmixing method based on the information entropy has a better unmixing effect.
Drawings
The invention will be further described with reference to the accompanying drawings and examples, in which:
FIG. 1 is a flow chart of a CHNMF remote sensing image unmixing method based on information entropy;
FIG. 2 is a graph of abundance of three end members in a first embodiment of the present invention;
FIG. 3 is a graph of abundance of three end members in example two of the present invention.
Detailed Description
For a more clear understanding of the technical features, objects and effects of the present invention, embodiments of the present invention will now be described in detail with reference to the accompanying drawings.
The CHNMF remote sensing image unmixing method based on the information entropy is shown in figure 1 and comprises the following steps:
s1, establishing a remote sensing image unmixing algorithm based on NMF for the current remote sensing image;
establishing a remote sensing image unmixing algorithm based on NMF, wherein the optimization problem of the algorithm can be regarded as minimizing the following objective function:
Figure GDA0003235971830000041
Figure GDA0003235971830000042
wherein M is an end-member spectral matrix, S is an abundance matrix, f (M, S) ═ 1/2| | | R-MS | | purple2The second term and the third term are respectively regularization functions of the parameter M, S, and are used for adding characteristics of smoothness, low rank and the like to the end-member spectrum matrix or the abundance matrix, and alpha and beta are regularization parameters.
In the optimization model, only one parameter can be restricted, or two parameters can be restricted at the same time, and the selection of the regularization function directly influences the quality of the final result. L ispNorm, kernel norm, trace norm, and the like are all regularly used regularization functions that can all achieve sparseness to varying degrees.
S2, establishing a CNMF remote sensing image unmixing algorithm based on smooth constraint for the current remote sensing image according to the NMF-based remote sensing image unmixing algorithm, and obtaining L2A norm;
according to the theory of sparse representation, L0Norm is good at describing sparsity, but L0The regularization function is an NP problem, is difficult to solve and is not beneficial to practical application, and later Tao and Cande et al prove that under the condition of RIP (corrected isometric property), L is1Norm is L0Optimal convex approximation of norm and simple solution, so the scholars use L1Norm instead of L0And solving the norm. Except for L1Norm, L2The norm solution is also very easy and it improves the over-fitting problem well. When a model is constructed, the model is not only suitable for training data, but also does not have the problem of overfitting. So L2The norm is very suitable for the optimization problem and is widely applied. One constrained non-negative matrix factorization (CNMF) algorithm proposed by Pauca and Piper et al is that using L2And (3) constraining M and S by the norm to obtain a target function of the CNMF remote sensing image unmixing algorithm based on smooth constraint:
Figure GDA0003235971830000043
Figure GDA0003235971830000044
wherein M is an end member spectrum matrix, S is an abundance matrix, the first item of the objective function represents an image reconstruction error, the second two items represent the smooth limitation of the end member spectrum matrix and the abundance matrix, and alpha and beta are regularization parameters for balancing the strength relation between constraint and the error.
The multiplicative iteration rule for CNMF is as follows:
M←M*(RST-αM)/(MSST+ε)
S←S*(MTR-βS)/(MTMS+ε)
in the formula, a small positive number epsilon is used for keeping the fraction constant as a positive number, and when iteration is carried out for a certain number of times, the value of f (M, S) tends to be stable, and the final optimized L-based2And (3) a norm CNMF remote sensing image unmixing algorithm.
S3, acquiring the information entropy of the current remote sensing image;
assuming that the current remote sensing image information source has n values: u shape1...Ui...UnThe corresponding probability is: p is a radical of1...pi...pnAnd the occurrence of each symbol is independent, the information entropy of the current remote sensing image information source is as follows:
Figure GDA0003235971830000051
as can be seen from the above, the degree of misordering of the probability distribution of the source determines the magnitude of the information entropy, and the two show negative correlation. The more uneven the probability distribution, the smaller the information entropy. Mixed image elements are very common in remote sensing images, in other words, the end members tend to be unevenly distributed.
S4, for the remote sensing image unmixing algorithm based on NMF, using the L obtained in S22And (3) the norm is used as a regularization function to constrain an end-member spectrum matrix M, the information entropy obtained in S3 is used as a regularization function to constrain an abundance matrix S, and the CHNMF remote sensing image unmixing method based on the information entropy is established.
S41, establishing an objective function of the CHNMF remote sensing image unmixing method based on the information entropy:
Figure GDA0003235971830000052
Figure GDA0003235971830000053
wherein M is an end member spectrum matrix, S is an abundance matrix, a first term represents reconstruction errors, a second term represents sparse of the abundance matrix by using information entropy, a third term introduces smoothness limitation into the end member spectrum matrix, and lambda and phi are regularization parameters; sijAnd representing the proportion of corresponding end members in the pixels for each element, wherein L is the number of wave bands of the remote sensing image, R is the remote sensing image with L wave bands, P is the number of end members of the remote sensing image to be detected, and N is the number of pixel points of the remote sensing image to be detected.
S42, solving the end-member spectrum matrix M and the abundance matrix S by adopting a multiplicative iteration rule, and solving partial derivatives of the M and the S according to the properties of the matrixes to obtain:
Figure GDA0003235971830000061
Figure GDA0003235971830000062
then, iteration is carried out by using a gradient descent method to obtain a final multiplicative iteration rule of M and S:
M←M*RST/(MSST+ε)
Figure GDA0003235971830000063
and (3) the fraction is constant to be a positive number by using a small positive number epsilon, when iteration is carried out for a certain number of times, the variation value of f (M, S) is smaller than a preset value, namely the value of f (M, S) tends to be stable, and the finally optimized CHNMF remote sensing image unmixing method based on the information entropy is obtained.
The following data of the embodiment illustrates the beneficial effects of the CHNMF remote sensing image unmixing method based on the information entropy compared with the traditional NMF algorithm and CNMF algorithm:
first, in this embodiment, the validity of the CHNMF algorithm is verified by using the synthesized data, and compared with the conventional NMF algorithm and CNMF algorithm. And acquiring an initial value by adopting a random method. And the regularization parameter of CHNMF is also an empirical value with reference to CNMF. Three types of ground substances, Carnallite (Carnallite), flint (Chert) and grenadite (Andradite), were still selected as experimental results, which are the average results of 25 experiments.
As shown in fig. 2, the three end-member abundance maps obtained by unmixing through the three algorithms NMF, CNMF and CHNMF are compared with the real end-member abundance map, and the area enclosed by the white rectangle in the table is closer to the reference result from top to bottom, that is, the CHNMF unmixed end-member abundance map and the real end-member abundance map are the closest.
Table 1 shows the SAD and RMSE values for the four algorithm unmixing results. From the experimental results, the CHNMF algorithm is superior to the CNMF algorithm in terms of both the spectral angular distance SAD and the root mean square error RMSE, and certainly superior to the traditional NMF algorithm.
TABLE 1 Performance indices of three NMF unmixing algorithms
Figure GDA0003235971830000064
Example two, in this example, the data used is partial data of Cuprite in nevada, and the unmixing situation of three algorithms, NMF, CNMF and hcmf, is further compared and analyzed through real data experiments, wherein the abundance map is referred to the result of the tetracord operation, for the spectrum curve of the end member, the spectrum curve of the end member extracted from the raw data at the ENVI4.8 platform is referred to, and three typical features, namely hematite (hematite), chalcedony (chalcedony) and pyroxene (pyroxene), are selected as the reference result, as shown in fig. 3 and table 2, and the abundance map, SAD value and RMSE value of the three end members after unmixing of the three algorithms are given in sequence.
From the abundance map, for hematite, the area enclosed by the white frame in the second column of the table, from top to bottom, is closer to black in color, more similar to the reference result; for the chalcedony, the third column in the table is the area enclosed by a white frame, the abundance maps of the CNMF and CHNMF algorithms are superior to those of the traditional NMF, the reference result is more approximate, but the CNMF and the CHNMF are similar; for pyroxene, the area enclosed by the white frame in the fourth column of the table, from top to bottom, is closer to black in color, more closely approximating the reference result. That is, the CHNMF algorithm has better unmixing effect than the CNMF and NMF algorithms in terms of visual effect integration.
From table 2, it can be seen that from the SAD values, CHNMF has the best unmixing effect for hematite, chalcedony and pyroxene, and CNMF has the second highest and NMF is the worst. From the RMSE value, for hematite and pyroxene two end members, the demixing effect of the CHNMF algorithm is the best, the CNMF is the second, and the NMF is the worst; for chalcedony, the unmixing effect of the CHNMF algorithm is the best, and CNMF is the second and the NMF is the worst.
TABLE 2 SAD and RMSE values for each algorithm unmixing
Figure GDA0003235971830000071
By combining the two embodiments, the CHNMF remote sensing image unmixing method based on the information entropy obtains the best unmixing effect.
While the present invention has been described with reference to the embodiments shown in the drawings, the present invention is not limited to the embodiments, which are illustrative and not restrictive, and it will be apparent to those skilled in the art that various changes and modifications can be made therein without departing from the spirit and scope of the invention as defined in the appended claims.

Claims (3)

1. The CHNMF remote sensing image unmixing method based on the information entropy is characterized by comprising the following steps:
s1, establishing a remote sensing image unmixing algorithm based on NMF for the current remote sensing image;
s2, establishing a CNMF remote sensing image unmixing algorithm based on smooth constraint for the current remote sensing image according to the NMF-based remote sensing image unmixing algorithm to obtain L of the CNMF remote sensing image unmixing algorithm based on smooth constraint2A norm;
s3, acquiring the information entropy of the current remote sensing image;
s4, for the remote sensing image unmixing algorithm based on NMF, using the L obtained in S22The norm is used as a regularization function constraint end member spectrum matrix M, the information entropy obtained in S3 is used as a regularization function constraint abundance matrix S, and a CHNMF remote sensing image unmixing method based on the information entropy is established;
step S4 specifically includes:
s41, establishing an objective function of the CHNMF remote sensing image unmixing method based on the information entropy:
Figure FDA0003235971820000011
Figure FDA0003235971820000012
wherein M is an end member spectrum matrix, S is an abundance matrix, a first term represents reconstruction errors, a second term represents sparse of the abundance matrix by using information entropy, a third term introduces smoothness limitation into the end member spectrum matrix, and lambda and phi are regularization parameters; sijFor each element representing a corresponding end-member in the picture elementThe method comprises the following steps of (1) accounting for the ratio, wherein L is the number of wave bands of the remote sensing image, R is the remote sensing image with L wave bands, P is the number of end members of the remote sensing image to be detected, and N is the number of pixel points of the remote sensing image to be detected;
s42, solving the end-member spectrum matrix M and the abundance matrix S by adopting a multiplicative iteration rule, and solving partial derivatives of the M and the S according to the properties of the matrixes to obtain:
Figure FDA0003235971820000013
Figure FDA0003235971820000014
then, iteration is carried out by using a gradient descent method to obtain a final multiplicative iteration rule of M and S:
M←M*RST/(MSST+ε)
Figure FDA0003235971820000015
and (3) the fraction is constant to be a positive number by using a small positive number epsilon, and when iteration is carried out for a certain number of times, the change value of f (M, S) is smaller than a preset value, so that the final optimized CHNMF remote sensing image unmixing method based on the information entropy is obtained.
2. The CHNMF remote sensing image unmixing method based on information entropy as claimed in claim 1, wherein step S2 specifically includes:
according to the theory of sparse representation, with L2And the norm restrains an end member spectrum matrix and an abundance matrix of the remote sensing image unmixing algorithm based on the NMF to obtain the CNMF remote sensing image unmixing algorithm based on the smooth constraint.
3. The CHNMF remote sensing image unmixing method based on information entropy as claimed in claim 1, wherein step S3 specifically includes:
assume that it is currentThe remote sensing image information source has n values: u shape1...Ui...UnThe corresponding probability is: p is a radical of1...pi...pnAnd the occurrence of each symbol is independent, the information entropy of the current remote sensing image information source is as follows:
Figure FDA0003235971820000021
CN201910538105.5A 2019-06-20 2019-06-20 CHNMF remote sensing image unmixing method based on information entropy Expired - Fee Related CN110428369B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910538105.5A CN110428369B (en) 2019-06-20 2019-06-20 CHNMF remote sensing image unmixing method based on information entropy

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910538105.5A CN110428369B (en) 2019-06-20 2019-06-20 CHNMF remote sensing image unmixing method based on information entropy

Publications (2)

Publication Number Publication Date
CN110428369A CN110428369A (en) 2019-11-08
CN110428369B true CN110428369B (en) 2021-10-08

Family

ID=68408794

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910538105.5A Expired - Fee Related CN110428369B (en) 2019-06-20 2019-06-20 CHNMF remote sensing image unmixing method based on information entropy

Country Status (1)

Country Link
CN (1) CN110428369B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111914893B (en) * 2020-06-24 2023-09-12 西安交通大学 Hyperspectral unmixing method and hyperspectral unmixing system based on entropy regular non-negative matrix factorization model
CN113836483A (en) * 2021-08-10 2021-12-24 中国地质大学（武汉） Deep non-negative matrix unmixing method based on information entropy sparseness and storage medium

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102789639A (en) * 2012-07-16 2012-11-21 中国科学院自动化研究所 Method for fusing high-spectrum image and visible light image based on nonnegative matrix decomposition
CN104240212A (en) * 2014-09-03 2014-12-24 西安电子科技大学 ISAR image fusion method based on target characteristics
CN107169946A (en) * 2017-04-26 2017-09-15 西北工业大学 Image interfusion method based on non-negative sparse matrix Yu hypersphere color transformation
CN107992802A (en) * 2017-11-10 2018-05-04 桂林电子科技大学 A kind of microseism weak signal recognition methods based on NMF
CN108388863A (en) * 2018-02-27 2018-08-10 南昌工程学院 A kind of hyperspectral remote sensing image mixed pixel decomposition method
CN109241843A (en) * 2018-08-02 2019-01-18 南京理工大学 Sky spectrum joint multiconstraint optimization nonnegative matrix solution mixing method
CN109345495A (en) * 2018-09-11 2019-02-15 中国科学院长春光学精密机械与物理研究所 Image interfusion method and device based on energy minimum and gradient regularisation

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108009182B (en) * 2016-10-28 2020-03-10 京东方科技集团股份有限公司 Information extraction method and device

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102789639A (en) * 2012-07-16 2012-11-21 中国科学院自动化研究所 Method for fusing high-spectrum image and visible light image based on nonnegative matrix decomposition
CN104240212A (en) * 2014-09-03 2014-12-24 西安电子科技大学 ISAR image fusion method based on target characteristics
CN107169946A (en) * 2017-04-26 2017-09-15 西北工业大学 Image interfusion method based on non-negative sparse matrix Yu hypersphere color transformation
CN107992802A (en) * 2017-11-10 2018-05-04 桂林电子科技大学 A kind of microseism weak signal recognition methods based on NMF
CN108388863A (en) * 2018-02-27 2018-08-10 南昌工程学院 A kind of hyperspectral remote sensing image mixed pixel decomposition method
CN109241843A (en) * 2018-08-02 2019-01-18 南京理工大学 Sky spectrum joint multiconstraint optimization nonnegative matrix solution mixing method
CN109345495A (en) * 2018-09-11 2019-02-15 中国科学院长春光学精密机械与物理研究所 Image interfusion method and device based on energy minimum and gradient regularisation

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
Nonnegative Matrix Factorization with Rank Regularization and Hard Constraint;Ronghua Shang 等;《neural computation》;20170930;第2553-2579页 *
基于信息熵属性约简的矩阵方法;蒙韧 等;《广西计算机学会年会》;20090919;第1-5页 *
改进的MVG—NMF算法在高光谱图像解混中的应用;李二森 等;《海洋测绘》;20100930;第30卷(第5期);第17-20页 *
融入结构信息的稀疏低秩丰度估计算法研究;袁静 等;《红外与毫米波学报》;20180430;第37卷(第2期);第144-153页 *

Also Published As

Publication number Publication date
CN110428369A (en) 2019-11-08

Similar Documents

Publication Publication Date Title
He et al. Total-variation-regularized low-rank matrix factorization for hyperspectral image restoration
Dundar et al. Sparse representation-based hyperspectral image classification using multiscale superpixels and guided filter
Palsson et al. Multispectral and hyperspectral image fusion using a 3-D-convolutional neural network
Correa et al. Spatiotemporal blue noise coded aperture design for multi-shot compressive spectral imaging
Erichson et al. Randomized nonnegative matrix factorization
Huang et al. Spatial and spectral image fusion using sparse matrix factorization
US20150262549A1 (en) Color Palette Generation
Dev et al. Categorization of cloud image patches using an improved texton-based approach
Heikkinen et al. Evaluation and unification of some methods for estimating reflectance spectra from RGB images
Dev et al. Systematic study of color spaces and components for the segmentation of sky/cloud images
Wang et al. Tensor decompositions for hyperspectral data processing in remote sensing: A comprehensive review
CN104182978B (en) A kind of high spectrum image object detection method based on empty spectrum nuclear sparse expression
CN110428369B (en) CHNMF remote sensing image unmixing method based on information entropy
CN105184302B (en) A kind of high optical spectrum image end member extraction method
CN107316309B (en) Hyperspectral image saliency target detection method based on matrix decomposition
Yu et al. Locality sensitive discriminant analysis for group sparse representation-based hyperspectral imagery classification
CN110458760B (en) HNMF remote sensing image unmixing method based on information entropy
Cao et al. A robust PCA approach with noise structure learning and spatial–spectral low-rank modeling for hyperspectral image restoration
Charles et al. Spectral Superresolution of Hyperspectral Imagery Using Reweighted $\ell_ {1} $ Spatial Filtering
Li et al. Locally linear embedded sparse coding for spectral reconstruction from RGB images
Chen et al. Dimensionality reduction of hyperspectral imagery using sparse graph learning
CN107274360A (en) A kind of high spectrum image denoising method based on Fisher dictionary learnings, low-rank representation
Nie et al. Graph-regularized tensor robust principal component analysis for hyperspectral image denoising
Bakken et al. The effect of dimensionality reduction on signature-based target detection for hyperspectral remote sensing
CN115984155A (en) Hyperspectral, multispectral and panchromatic image fusion method based on spectrum unmixing

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
CF01 Termination of patent right due to non-payment of annual fee
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20211008