CN106204508B - WorldView-2 remote sensing PAN and multi-spectral image interfusion method based on non-negative sparse matrix - Google Patents
WorldView-2 remote sensing PAN and multi-spectral image interfusion method based on non-negative sparse matrix Download PDFInfo
- Publication number
- CN106204508B CN106204508B CN201610503045.XA CN201610503045A CN106204508B CN 106204508 B CN106204508 B CN 106204508B CN 201610503045 A CN201610503045 A CN 201610503045A CN 106204508 B CN106204508 B CN 106204508B
- Authority
- CN
- China
- Prior art keywords
- component
- image
- pan
- matrix
- sinφ
- 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
Links
- 239000011159 matrix material Substances 0.000 title claims abstract description 33
- 238000000034 method Methods 0.000 title claims abstract description 33
- 230000004927 fusion Effects 0.000 claims abstract description 21
- 230000009466 transformation Effects 0.000 claims abstract description 17
- 238000002156 mixing Methods 0.000 claims abstract description 9
- 239000000284 extract Substances 0.000 claims description 9
- 238000000354 decomposition reaction Methods 0.000 claims description 4
- 238000001228 spectrum Methods 0.000 abstract description 12
- 238000000605 extraction Methods 0.000 abstract description 10
- 238000004458 analytical method Methods 0.000 abstract description 3
- 238000002347 injection Methods 0.000 abstract description 2
- 239000007924 injection Substances 0.000 abstract description 2
- 230000000007 visual effect Effects 0.000 abstract 1
- 230000003595 spectral effect Effects 0.000 description 9
- 238000007500 overflow downdraw method Methods 0.000 description 7
- 238000007796 conventional method Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 239000000155 melt Substances 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 238000012935 Averaging Methods 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000003795 chemical substances by application Substances 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000004321 preservation Methods 0.000 description 1
- 239000000243 solution Substances 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides a kind of WorldView-2 remote sensing PAN and multi-spectral image interfusion method based on non-negative sparse matrix, it is related to image co-registration field, luminance component extraction is carried out to multispectral image by using a kind of Algorithms of Non-Negative Matrix Factorization, then image is merged using HCS transformation, obtain blending image, certain promotion has all been obtained in terms of details injection and spectrum holding, the fusion results of high quality are finally obtained, since the extraction to I component uses NMF method, improve the extraction accuracy of luminance component, it is more reasonable compared with comparison algorithm, so that WV-2 satellite blending image total quality is higher, it all improves in terms of detailed information involvement and spectrum holding, subjective assessment can reach consistent with objective analysis results, obtained blending image is visual more preferable, picture is more Clearly.More traditional remote sensing PAN and multi-spectral image interfusion method is advantageously.
Description
Technical field
The present invention relates to a kind of methods of image co-registration field, especially remote sensing PAN and multi-spectral image co-registration.
Background technique
In recent years, the wave band number of New Satellite remote sensing multispectral image is being continuously increased, and the resolution ratio of image is also quick
It improves on ground.By taking WorldView-2 (WV-2) satellite as an example, WV-2 satellite launches in 2009, is capable of providing 1.84 meters of 8 wave band
The full-colour image of 0.46 meter of resolution ratio of multispectral image and single band of resolution ratio.There is following spy compared with conventional satellite image
Point: wave band increases, and spectrum divides thinner;The spectral coverage of full-colour image narrows, and is allowed to the spectrum with multi light spectrum hands
With having greatly changed.In remote sensing application, the image simultaneously with high spatial and high spectral resolution is generally required.Image
Integration technology is exactly the spatial resolution for removing to improve multispectral image using the full-colour image of high spatial resolution, while being protected as far as possible
The spectral characteristic for holding multispectral image is constant.WV-2 satellite image represents the development trend of ultra high resolution remote sensing images, together
When be also that more stringent requirements are proposed for the fusions of remote sensing images.Just because of these variations, make existing fusion method effect not
It is good.
Many Remote sensing image fusion algorithms have the process of extract light intensity level at present, for example, Brovey transformation, HCS become
Fusion results will be directly affected, have a great impact to the performance of algorithm with MSFIM method, the quality that luminance component extracts by changing.
Wherein, MSFIM algorithm is a kind of innovatory algorithm based on brightness smothing filtering modulation (SFIM) algorithm, although this innovatory algorithm
The involvement of detailed information is improved, is but produced compared with the worse spectrum distortion of SFIM algorithm.To reduce the spectrum in MSFIM method
Distortion, needs to change the ratio of luminance component and full-colour image, makes ratio closer to 1, i.e., should make luminance component and full-colour picture
The spectral response characteristic of picture is more like.Other have averaging method, weighted mean approach and each wave band to calculate the extracting method of luminance component
The art method of average.These methods are all the differentiation of conventional method in fact, and the luminance component of extraction is inaccurate.
Summary of the invention
Existing fusion method is directed to conventional satellite remote sensing multispectral image, for ultra high resolution remote sensing images
Speech is not optimal image interfusion method, and existing luminance component extracting method can not preferably solve details involvement
With spectrum distortion problem so that extract luminance component it is inaccurate.
For overcome the deficiencies in the prior art, by the present invention in that with a kind of Non-negative Matrix Factorization (Non-negative
Matrix Factorization, NMF) algorithm to multispectral image carry out luminance component extraction, then using HCS transformation pair
Image is merged, and blending image is obtained, and blending image of the invention has all obtained one in terms of details injection and spectrum holding
Fixed promotion has finally obtained the fusion results of high quality.
The technical solution adopted by the present invention to solve the technical problems includes the following steps:
Step 1. extracts I component using Non-negative Matrix Factorization method
Firstly, by the multispectral image X of full-colour image Pan and eight wave bands1,X2,…,X8It is straightened by row, obtains P, M1,
M2,…,M8Vector, then form matrix V to be decomposed by formula (1), i.e.,
V=[P, M1,M2,...,M8] (1)
Wherein, P, M1,M2,…,M8Respectively full-colour image Pan and eight wave band multispectral image X1,X2,…,X8When operation
Image array is pulled into corresponding column vector;
Secondly, enabling again
[P,M1,M2,...,M8]=WH (2)
Wherein W is n*r matrix, and n is the line number of matrix W, and r is the columns of matrix W, and H is r*9 matrix, and W is one after decomposition
W is reverted to the i.e. obtained I component of image array by column vector;
Step 2. matches I component using Pan component
It enables
P'2=(Pan)2 (3)
Wherein Pan is full-colour image, i.e., replaces Pan variable with P ' variable, and then, following formula uses P'2Component
With the resulting I of step 12Component:
Wherein μ0、σ0Respectively I2Mean value and standard variance, μ1、σ1Respectively P'2Mean value and standard variance, P "2For
I after matching2Component;
Use IadjComponent replaces P " component to indicate the I component after matching;
Step 3. finally obtains eight new wave band components using the fusion HCS transformation of hypersphere color space resolution ratio
Firstly, to the multispectral image X of eight wave bands1,X2,...,X8It carries out HCS direct transform and obtains corresponding angle component
φ1,φ2,...,φ7, HCS direct transform is as follows:
Secondly, the I component I after the matching acquired to step 2adjWith angle component φ1,φ2,...,φ7Do HCS inverse transformation
Obtain eight new wave band component X'1,X'2,…,X'8, HCS inverse transformation is as follows:
X1'=Iadjcosφ1
X2'=Iadjsinφ1cosφ2
X7'=Iadjsinφ1sinφ2...sinφ6cosφ7
X8'=Iadjsinφ1sinφ2...sinφ6sinφ7 (7)
Each Band fusion of step 4.
X' in selecting step 31,X'2,…,X'8In any three wave bands merged, i.e., three band images are directly placed into
In RGB triple channel, blending image can be obtained.
The beneficial effects of the invention are as follows NMF method is used due to the extraction to I component, the extraction of luminance component is improved
Precision, the extraction algorithm for relatively comparing I component in the innovatory algorithm MSFIM algorithm of algorithm SFIM is more reasonable, so that WV-2 satellite melts
It is higher to close overall picture quality, detailed information incorporate and spectrum holding in terms of all improve, subjective assessment and objective
Analysis result can reach unanimously, and in addition the method for the present invention is compared with comparing the obtained blending image visuality of algorithm MSFIM algorithm more
Good, picture is apparent.More traditional remote sensing PAN and multi-spectral image interfusion method is advantageously.
Detailed description of the invention
Fig. 1 is technology path block diagram of the invention.
Fig. 2 is multispectral image X of the present invention2,X3,X5Fusion results.
Specific embodiment
Present invention will be further explained below with reference to the attached drawings and examples.
Step 1. extracts I component using Non-negative Matrix Factorization (NMF) method
Means of numerical analysis of the Non-negative Matrix Factorization as a relative maturity is dug in image analysis, text cluster, data
Pick, speech processes etc. are widely applied.In view of the I component that extracts from multispectral image will as far as possible and entirely
The spectral response of chromatic graph picture is consistent, therefore the multispectral figure of full-colour image and eight wave bands is used in the method that NMF extracts I component
As forming matrix V to be decomposed.
Firstly, by the multispectral image X of full-colour image Pan and eight wave bands1,X2,…,X8It is straightened by row, obtains P, M1,
M2,…,M8Vector, then form matrix V to be decomposed by formula (1), i.e.,
V=[P, M1,M2,...,M8] (1)
Wherein, P, M1,M2,…,M8Respectively full-colour image Pan and eight wave band multispectral image X1,X2,…,X8When operation
Image array is pulled into corresponding column vector;
Secondly, enabling again
[P,M1,M2,...,M8]=WH (2)
Wherein W is n*r matrix, and n is the line number of matrix W, and r is the columns of matrix W, and H is r*9 matrix.W after decomposition is one
W is reverted to the i.e. obtained I component of image array by a column vector.
Wherein, W is n*r matrix, and n is the line number of matrix W, and r is the columns of matrix W, and H is r*9 matrix, it is contemplated that extraction
I component will match with full-colour image, should extract the global characteristics W of full-colour image and multispectral image at this time, by the W after decomposition
Revert to the i.e. obtained I component of image array;
Step 2. matches I component using Pan component
It enables
P'2=(Pan)2 (3)
Wherein Pan is full-colour image, i.e., replaces Pan variable with P ' variable, and then, following formula uses P'2Component
With the resulting I of step 12Component:
Wherein μ0、σ0Respectively I2Mean value and standard variance, μ1、σ1Respectively P'2Mean value and standard variance, P "2For
I after matching2Component;
Use IadJ component replaces P " component to indicate the I component after matching;
Step 3. finally obtains eight new wave band components using the fusion HCS transformation of hypersphere color space resolution ratio
Fusion method HCS (Hyperspherical Color Space towards WorldView-2 satellite image
Resolution Merge) to wave band number, there is no limit for transformation, therefore is suitble to Multi-Band Remote Sensing Images fusion.In HCS transformation,
Angle variables determine the spectral information of image, and I determines the luminance information of image, the spectral information of the variation of I variable to image
It does not influence, this is the key point of HCS transformation.
Firstly, to the multispectral image X of eight wave bands1,X2,...,X8It carries out HCS direct transform and obtains corresponding angle component
φ1,φ2,...,φ7, HCS direct transform is as follows:
Secondly, the I component I after the matching acquired to step 2adjWith angle component φ1,φ2,...,φ7Do HCS inverse transformation
Obtain eight new wave band component X'1,X'2,…,X'8, HCS inverse transformation is as follows:
X1'=Iadjcosφ1
X2'=Iadjsinφ1cosφ2
X7'=Iadjsinφ1sinφ2...sinφ6cosφ7
X8'=Iadjsinφ1sinφ2...sinφ6sinφ7 (7)
Each Band fusion of step 4.
X' in selecting step 31,X'2,…,X'8In any three wave bands merged, i.e., three band images are directly placed into
In RGB triple channel, blending image can be obtained.
Just for WorldView-2 remote sensing PAN and multi-spectral New Image Fusion of the verifying based on non-negative sparse matrix
True property and Optimality carry out following experiment.Fusion method used in experiment is HCS transformation.The I component of two kinds of fusion methods mentions
Take method that MSFIM method and NMF method is respectively adopted.Experimental data is one group of true WV-2 image, is shot on April 3rd, 2011
Sydney Australia, can show for verification algorithm and clearly image for subjective assessment, experimental data is using image
A part, size 300*300pixels, gray level 256.To obtain reference picture when evaluation result, first mostly light
Spectrogram picture is down sampled to original a quarter, then is upsampled to original size, and multispectral image original in this way can be used as ginseng
Examine image.
(a) is the image that original full-colour image is down sampled to after a quarter of original image in Fig. 2;Fig. 2 (b) is former
The multispectral image of beginning is compared for experimental result.Fig. 2 (c) is original extraction I component method in HCS transformation fusion method
Fusion results;Fig. 2 (d) is the fusion results for extracting I component with NMF method herein.Observe Fig. 2, the spectral signature of Fig. 2 (d) with
Fig. 2 (b) is more close, i.e., context of methods fusion results are got well than original method in spectrum holding, it can be seen that, institute of the present invention
The obtained fusion results situation constant in spectral preservation characteristic, the fusion method that the involvement of detailed information is more traditional will be got well, and
Fusion results of the invention also improve in spectrum holding.
The present invention chooses space correlation coefficient (spatial correlate coefficient, sCC), related coefficient
(correlate coeffcient, CC), average gradient (average gradient, AG), comentropy (information
Entropy, IE) etc. the common index that objectively evaluates fusion results are objectively evaluated, these index values show more greatly fusion knot
Fruit is better.
Table 1 objectively evaluates result
Eight wave bands of Band1 ... Band8 expression multispectral image.It can be seen that the items of the method for the present invention are commented in table 1
Valence index is larger, this illustrates that the present invention is shown in remote sensing image fusion compared with conventional method better performance.
Claims (1)
1. a kind of WorldView-2 remote sensing PAN and multi-spectral image interfusion method based on non-negative sparse matrix, feature exist
In including the following steps:
Step 1. extracts I component using Non-negative Matrix Factorization method
Firstly, by the multispectral image X of full-colour image Pan and eight wave bands1,X2,…,X8It is straightened by row, obtains P, M1,M2,…,M8
Vector, then form matrix V to be decomposed by formula (1), i.e.,
V=[P, M1,M2,...,M8] (1)
Wherein, P, M1,M2,…,M8Respectively full-colour image Pan and eight wave band multispectral image X1,X2,…,X8It will figure when operation
As matrix pulls into corresponding column vector;
Secondly, enabling again
[P,M1,M2,...,M8]=WH (2)
Wherein W be n*r matrix, n be matrix W line number, r be matrix W columns, H be r*9 matrix, after decomposition W be one arrange to
Amount, reverts to the i.e. obtained I component of image array for W;
Step 2. matches I component using Pan component
It enables
P'2=(Pan)2 (3)
Wherein Pan is full-colour image, i.e., replaces Pan variable with P ' variable, and then, following formula uses P'2Component matching step
1 resulting I2Component:
Wherein μ0、σ0Respectively I2Mean value and standard variance, μ1、σ1Respectively P'2Mean value and standard variance, P "2After matching
I2Component;
Use IadjComponent replaces P " component to indicate the I component after matching;
Step 3. finally obtains eight new wave band components using the fusion HCS transformation of hypersphere color space resolution ratio
Firstly, to the multispectral image X of eight wave bands1,X2,...,X8It carries out HCS direct transform and obtains corresponding angle component φ1,
φ2,...,φ7, HCS direct transform is as follows:
Secondly, the I component I after the matching acquired to step 2adjWith angle component φ1,φ2,...,φ7HCS inverse transformation is done to obtain
Eight new wave band component X'1,X'2,…,X'8, HCS inverse transformation is as follows:
X1'=Iadjcosφ1
X2'=Iadjsinφ1cosφ2
X7'=Iadjsinφ1sinφ2...sinφ6cosφ7
X8'=Iadjsinφ1sinφ2...sinφ6sinφ7 (7)
Each Band fusion of step 4.
X' in selecting step 31,X'2,…,X'8In any three wave bands merged, i.e., three band images are directly placed into RGB
In triple channel, blending image can be obtained.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610503045.XA CN106204508B (en) | 2016-06-30 | 2016-06-30 | WorldView-2 remote sensing PAN and multi-spectral image interfusion method based on non-negative sparse matrix |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610503045.XA CN106204508B (en) | 2016-06-30 | 2016-06-30 | WorldView-2 remote sensing PAN and multi-spectral image interfusion method based on non-negative sparse matrix |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106204508A CN106204508A (en) | 2016-12-07 |
CN106204508B true CN106204508B (en) | 2018-12-14 |
Family
ID=57462579
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610503045.XA Active CN106204508B (en) | 2016-06-30 | 2016-06-30 | WorldView-2 remote sensing PAN and multi-spectral image interfusion method based on non-negative sparse matrix |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106204508B (en) |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107169946B (en) * | 2017-04-26 | 2020-04-03 | 西北工业大学 | Image fusion method based on nonnegative sparse matrix and hypersphere color transformation |
CN107958450B (en) * | 2017-12-15 | 2021-05-04 | 武汉大学 | Panchromatic multispectral image fusion method and system based on self-adaptive Gaussian filtering |
CN110427961B (en) * | 2019-06-19 | 2022-04-05 | 华南农业大学 | Building information extraction method and system based on rule and sample fusion |
CN110458760B (en) * | 2019-06-20 | 2021-11-05 | 中国地质大学(武汉) | HNMF remote sensing image unmixing method based on information entropy |
CN113781339B (en) * | 2021-09-02 | 2023-06-23 | 中科联芯(广州)科技有限公司 | Silicon-based multispectral signal processing method and device and mobile terminal |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104346790A (en) * | 2014-10-30 | 2015-02-11 | 中山大学 | Remote sensing image fusion method through combining HCS with wavelet transform |
CN104851091A (en) * | 2015-04-28 | 2015-08-19 | 中山大学 | Remote sensing image fusion method based on convolution enhancement and HCS transform |
CN104867124A (en) * | 2015-06-02 | 2015-08-26 | 西安电子科技大学 | Multispectral image and full-color image fusion method based on dual sparse non-negative matrix factorization |
-
2016
- 2016-06-30 CN CN201610503045.XA patent/CN106204508B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104346790A (en) * | 2014-10-30 | 2015-02-11 | 中山大学 | Remote sensing image fusion method through combining HCS with wavelet transform |
CN104851091A (en) * | 2015-04-28 | 2015-08-19 | 中山大学 | Remote sensing image fusion method based on convolution enhancement and HCS transform |
CN104867124A (en) * | 2015-06-02 | 2015-08-26 | 西安电子科技大学 | Multispectral image and full-color image fusion method based on dual sparse non-negative matrix factorization |
Non-Patent Citations (2)
Title |
---|
Multi-source remote sensing data fusion: status and trends;Jixian Zhang;《International Journal of Image and Data Fusion》;20100331;第1卷(第1期);全文 * |
基于改进梯度投影NMF和复Contourlet变换的遥感图像融合;吴一全 等;《光子学报》;20140531;第43卷(第5期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN106204508A (en) | 2016-12-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106204508B (en) | WorldView-2 remote sensing PAN and multi-spectral image interfusion method based on non-negative sparse matrix | |
Berman et al. | ICE: A statistical approach to identifying endmembers in hyperspectral images | |
CN106384332B (en) | Unmanned plane image and multispectral image fusion method based on Gram-Schmidt | |
Wang et al. | Comparison of different color spaces for image segmentation using graph-cut | |
CN112819737B (en) | Remote sensing image fusion method of multi-scale attention depth convolution network based on 3D convolution | |
CN109993717A (en) | A kind of remote sensing image fusion method of combination guiding filtering and IHS transformation | |
CN106023130B (en) | Unmanned plane image and multispectral image fusion method based on gradient filtering and PCA | |
CN103927741A (en) | SAR image synthesis method for enhancing target characteristics | |
CN105825217B (en) | Hyperspectral imaging area-of-interest extraction method based on movable contour model | |
CN107169946B (en) | Image fusion method based on nonnegative sparse matrix and hypersphere color transformation | |
CN113139902A (en) | Hyperspectral image super-resolution reconstruction method and device and electronic equipment | |
JP2014534699A (en) | System and method for digital image signal compression using unique images | |
Amiri et al. | A step by step recovery of spectral data from colorimetric information | |
CN105869114B (en) | Multispectral image and panchromatic image fusion method based on multilayer interband structural model | |
CN110991463B (en) | Multi-scale guide filtering feature extraction method under guidance of super-pixel map | |
CN109636768B (en) | Remote sensing image fusion method and device and electronic equipment | |
CN106530360B (en) | Complementary color wavelet color image processing method | |
CN107492085B (en) | Stereo image quality evaluation method based on dual-tree complex wavelet transform | |
CN105023261B (en) | Remote sensing image fusion method based on AGIHS and low pass filter | |
CN102156872B (en) | Multispectral-data-based object identification method and device | |
Wang et al. | A novel color calibration method of multi-spectral camera based on normalized RGB color model | |
CN113191993B (en) | Panchromatic and multispectral image fusion method based on deep learning | |
Jayadevan et al. | A review on recent pseudo-coloring techniques | |
CN108364284A (en) | A kind of opium poppy extracting method and system based on remote sensing | |
CN108229487B (en) | Significance detection method combining space domain and frequency domain |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |