CN107146212A - A kind of remote sensing image fusion method based on Steerable filter - Google Patents

A kind of remote sensing image fusion method based on Steerable filter Download PDF

Info

Publication number
CN107146212A
CN107146212A CN201710242117.4A CN201710242117A CN107146212A CN 107146212 A CN107146212 A CN 107146212A CN 201710242117 A CN201710242117 A CN 201710242117A CN 107146212 A CN107146212 A CN 107146212A
Authority
CN
China
Prior art keywords
image
window
multispectral
output
pan
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201710242117.4A
Other languages
Chinese (zh)
Other versions
CN107146212B (en
Inventor
李旭
张艺鸣
高雅楠
李立欣
高昂
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.)
Northwestern Polytechnical University
Original Assignee
Northwestern Polytechnical 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 Northwestern Polytechnical University filed Critical Northwestern Polytechnical University
Priority to CN201710242117.4A priority Critical patent/CN107146212B/en
Publication of CN107146212A publication Critical patent/CN107146212A/en
Application granted granted Critical
Publication of CN107146212B publication Critical patent/CN107146212B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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 by the use of more than one image, e.g. averaging, subtraction
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10032—Satellite or aerial image; Remote sensing
    • G06T2207/10041—Panchromatic image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20212—Image combination
    • G06T2207/20221—Image fusion; Image merging

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Image Processing (AREA)
  • Color Television Image Signal Generators (AREA)

Abstract

The invention provides a kind of remote sensing image fusion method based on Steerable filter, the space structure details of panchromatic light image is imported in the multispectral image of amplification by using the coefficient matrix of Steerable filter device, improve the spatial resolution of multispectral image；The spectral information of IHS conversion separation multispectral images is recycled, spectral information is injected into multispectral image using IHS inverse transformations, original spectral information is helped to maintain；Finally by the extraction modulation and injection of details, spatial resolution is further improved on the basis of multispectral image spectral information is kept, high-quality fused images are obtained.The present invention is a kind of suitable for the multispectral effective integration method merged with panchromatic light image of high-resolution spaceborne.

Description

A kind of remote sensing image fusion method based on Steerable filter
Technical field
The present invention relates to the visual enhancement treatment technology of remote sensing images, in particular for satellite-borne multispectral image and panchromatic The image interfusion method of light image.
Background technology
Physical limitation and the barrier of data transmission technology due to sensor itself so that current satellite borne sensor is only Spectral resolution height can be provided but the low multiband multispectral image of spatial resolution and spatial resolution are high but spectrally resolved The low panchromatic light image of single wave band of rate.In-orbit satellite borne sensor, such as QuickBird, GeoEye-1, WorldView-2 Etc. the panchromatic light image of multiband multi-spectral remote sensing image and single band high spatial resolution can be provided simultaneously.It is many by merging Spectrum picture and panchromatic light image, can improve its spatial resolution while multispectral image spectral information is kept, reach Strengthen visual effect.
Panchromatic light image spatial resolution is higher than multispectral image, therefore is first had to during fusion by multispectral image resampling Be amplified to the same size of panchromatic light image, then the multispectral image after amplification and original panchromatic light image are inputted to fusion again Model, exports the multispectral image merged.It is general both at home and abroad at present to do in the resampling processing to multispectral image Method is that each wave band of multispectral image is independently entered from arest neighbors interpolation method, bilinear interpolation or bicubic interpolation method Row interpolation amplifies (referring to document IEEE Transactions on Geoscience and Remote Sensing, 45 (10): 3012-3021,2007).In fact, the multispectral image obtained by above-mentioned interpolation method occur different degrees of spectrum distortion with Spatial information distortion, so that further the information of influence fused images is kept (referring to document IEEE Geoscience and Remote Sensing Letters,4(1):27-31,2007；IEEE Journal of Selected Topics in Signal Processing,5(3):446-453,2011)。
The content of the invention
In order to overcome the deficiencies in the prior art, the present invention provides a kind of image interfusion method based on Steerable filter, passed through Steerable filter is combined resampling of the panchromatic light image completion to multispectral image and amplified, and recycles Intensity-Hue- Saturation (IHS) conversion separation spectral informations, most the structural information of panchromatic light image is extracted and is injected into multispectral at last In image, the spatial resolution of multispectral image is effectively improved.
The technical solution adopted for the present invention to solve the technical problems comprises the following steps：
The first step, carries out LPF, low pass filter is designated as LPF, then defeated by convolution algorithm to panchromatic light image PAN Go out image PL=PAN*LPF；Make lower 2 sample process to output image PL, output result is designated as P2；
Second step, sets guide image I, any j-th of multi light spectrum hands M that image P2 is Steerable filterjTo be oriented to filter The input picture P of ripple, Steerable filter output image is designated as Q；Pixel i is located at the window ω centered on pixel kkIn, should Windows radius r >=1, window size is (2r+1) × (2r+1) pixels；Filter filtering output values of the output image Q in pixel i QiBy guide image I the point pixel value IiAnd the window coefficient a of local window where i pointskAnd bkTogether decide on,Matrix form is that Q=A × I+B, wherein A and B are window coefficient matrixes, Window coefficientWindow coefficient| ω | it is window ωkThe number of middle all pixels, piFor pixel values of the input picture P at i, μkAnd σkRefer respectively to lead image in window ωkIn Average and variance,It is input picture P in window ωkIn average, ε be regularisation parameter and ε ∈ (0, 1)；
3rd step, is amplified using arest neighbors interpolation method to window coefficient matrix A and B resamplings, is respectively obtained corresponding new Window coefficientWith
4th step, utilizes original panchromatic light image PAN and new window coefficientWithCalculate large-sized output imageThe output imageIt is identical with original panchromatic light image size；
5th step, the optional three wave bands multispectral image M from n band multispectral imageT1、MT2And MT3；Repeat the Two steps calculate corresponding three width large scale output image to the 4th stepWithCalculate strength component
6th step, to three selected wave band multispectral image MT1、MT2、MT3IHS conversion is carried out, is obtainedUtilize bicubic interpolation method pair U, V carry out resampling interpolation, be amplified to the original panchromatic same sizes of light image PAN, be designated asWith
7th step, to strength component INT andWithIHS inverse transformations are carried out, the wave band figure of original color space three is obtained Picture；With three wave band multispectral image MT1、MT2、MT3Corresponding IHS inverse transformations output
8th step, calculates spatial detail D=PAN-INT；Own using arest neighbors interpolation method to original multispectral image Multi light spectrum hands M1..., MNCarry out resampling interpolation, be amplified to the panchromatic same sizes of light image PAN, be designated asCalculate the spatial detail index of modulation of each wave bandThe utilization space details index of modulation Three wave band multispectral image S are injected after spatial detail D is modulatedT1、ST2、ST3In, output image FT1=ST1+gT1×D、FT2= ST2+gT2× D and FT3=ST3+gT3× D is fusion results；
9th step, returns to the 5th step and selects three different wave bands to be merged, finally travel through all N number of wave bands, export N number of The fusion results of wave band.
The beneficial effects of the invention are as follows：Easily draw during resampling multispectral image for Remote sensing image fusion The problem of entering spatially and spectrally information distortion, by using the coefficient matrix of Steerable filter device by the space structure of panchromatic light image Details is imported in the multispectral image of amplification, improves the spatial resolution of multispectral image；Recycle IHS conversion separation light more Spectral information, is injected into multispectral image by the spectral information of spectrogram picture using IHS inverse transformations, helps to maintain original light Spectrum information；Finally by the extraction modulation and injection of details, further carried on the basis of multispectral image spectral information is kept Spatial resolution has been risen, high-quality fused images are obtained, has been that a kind of to be applied to high-resolution spaceborne multispectral with panchromatic light figure As the effective integration method of fusion.
Brief description of the drawings
Fig. 1 is the principle schematic of the present invention；
Fig. 2 is the flow chart of the present invention.
Embodiment
The present invention is further described with reference to the accompanying drawings and examples, and the present invention includes but are not limited to following implementations Example.
The present invention comprises the following steps：
Assuming that original multispectral image includes N number of wave band (M1, M2..., MN), each band image size is R1 rows × C1 Row, original panchromatic light image size is R2 rows × C2 row.
The panchromatic light image of the first step, down-sampling：
LPF is carried out to panchromatic light image (PAN) by convolution algorithm, low pass filter is designated as LPF, output image PL is designated as, * is convolution operator
PL=PAN*LPF (1)
Make lower 2 sample process to PL, i.e., pixel interlacing is sampled and sampled every row, output result is designated as P2, its size It is decreased to the half (R2/2 rows × C2/2 row) of original panchromatic light image size
P2=(PL) ↓2 (2)
Second step, Steerable filter
Guide image I, any j-th of multi light spectrum hands (M that image P2 is Steerable filter are setj) it is the defeated of Steerable filter Enter image P, Steerable filter output image is designated as Q.According to the definition of Steerable filter device, (Steerable filter device principle is referring to IEEE Transactions on Pattern Analysis and Machine Intelligence,35(6):1397-1409, 2013.), it is assumed that there is local linear relation between filtering output image Q and guide image I, i.e.,
Wherein pixel i is located at neighborhood (or window) ω centered on pixel kkIn, the windows radius is r, window Size is (2r+1) × (2r+1) pixels (generally taking radius r >=1).akAnd bkFor window coefficient, it is assumed that in window ωkIt is inside Constant.It follows that filtering output value Qs of the filtering output image Q in pixel iiBy guide image I the point pixel value Ii And the window coefficient a of local window where i pointskAnd bkTogether decide on.Window ωkInterior coefficient akAnd bkRespectively by following formula meter Calculate：
Wherein | ω | it is window ωkThe number of middle all pixels, piFor pixel values of the input picture P at i, μkAnd σkPoint It is not guide image in window ωkIn average and variance,It is input picture P in window ωkIn it is equal Value, ε is regularisation parameter and ε ∈ (0,1).
Due to the neighborhood ω comprising pixel ikIt is not unique, at the same exist it is multiple, therefore can be defeated using average policy calculation Go out pixel value Qs of the image Q in pixel ii, i.e.,
Because the symmetry of square window, thereforeFormula (6) can be rewritten as
Wherein
Formula (7) is expressed as matrix form, can be obtained
Q=A × I+B (10)
Wherein A and B are window coefficient matrixes.
The resampling amplification of 3rd step, window coefficient
The window coefficient matrix A of formula (10) and B resamplings are amplified using arest neighbors interpolation method, respectively obtained corresponding New window coefficientWith
4th step, Steerable filter output
Utilize original panchromatic light image PAN and new window coefficientWithCalculate large-sized output image
The output imageIdentical with original panchromatic light image size, size is R2 rows × C2 row.
5th step, calculating strength component INT：
Optional three wave band is (assuming that be designated as M from n band multispectral imageT1、MT2、MT3) colored synthesis is carried out, by three ripple Section is respectively as the input picture of Steerable filter, and it is defeated using second step to the 4th step to calculate corresponding three width large scale respectively Go out image, be designated as respectivelyWithIt is averaged and calculates strength component INT
Intensity-Hue-Saturation (IHS) conversion of 6th step, multispectral image
To three selected wave band multispectral image MT1、MT2、MT3IHS conversion is carried out, it is specific as follows
Using bicubic interpolation method to U, V carry out resampling interpolation, be amplified to the original panchromatic same sizes of light image PAN, It is designated asWith
7th step, IHS inverse transformations
To strength component INT andWithIHS inverse transformations are carried out, the band image of original color space three is obtained.With three ripples Section multispectral image MT1、MT2、MT3Corresponding IHS inverse transformations output is designated as S respectivelyT1、ST2、ST3
8th step, the extraction of spatial detail, modulation and injection
Spatial detail D is directly by panchromatic light image PAN and strength component INT mathematic interpolation
D=PAN-INT (18)
Using arest neighbors interpolation method to all multi light spectrum hands (M1..., MN) carry out resampling interpolation, be amplified to it is panchromatic The same sizes of light image PAN, are designated as
Calculate the spatial detail index of modulation g of each wave bandj, j=1 ..., N is specific as follows
Three wave band multispectral image S of the 7th step output are injected into after spatial detail D is modulated using the index of modulationT1、 ST2、ST3In, output image is designated as F respectivelyT1、FT2、FT3, i.e.,
FT1=ST1+gT1×D (20)
FT2=ST2+gT2×D (21)
FT3=ST3+gT3×D (22)
Image FT1、FT2、FT3The fusion results as exported.For the multispectral image (M comprising N number of wave band1, M2..., MN), it can repeatedly select three wave band therein to be merged, finally travel through all N number of wave bands, export the fusion knot of N number of wave band Really.
Embodiment of the method：
Using true WorldView-2 space remote sensings multispectral image and panchromatic light image, multispectral image includes eight Wave band (M1,M2,…,M8), panchromatic light image (PAN) is single band.The spatial resolution of multispectral image is 2.0m, and size is 200 rows × 200 are arranged.Panchromatic light image spatial resolution is 0.5m, and size is that 800 rows × 800 are arranged.Implement the present invention including following Step：
The panchromatic light image of the first step, down-sampling
Because panchromatic light image spatial resolution is four times of multispectral image spatial resolution, therefore panchromatic light image Size is four times of multispectral image, therefore is needed to panchromatic light image down-sampling twice.Here low pass filter selection is used Low pass filter CDF9, i.e. LPF=in the bi-orthogonal filter groups of Cohen-Daubechies-Fauveau (CDF) 9/7 [0.026748,-0.016864,-0.078223,0.266864,0.602949,0.266846,-0.078223,-0.016864, 0.026748] complete first time down-sampling using formula (1) and (2) to handle, obtain P2, P2 size reduction to 400 row × 400 Row.Formula (1) and (2) is recycled to complete second of down-sampling to P2, i.e.,
PL2=P2*LPF
P4=(PL2) ↓2
Obtained P4 size reductions to 200 rows × 200 are exported to arrange, it is consistent with multispectral image size.
Second step, Steerable filter
Three wave band (M are selected from multispectral imagej, j=2,3,5), the guidance diagram that image P4 is Steerable filter is set As I, MjFor input picture, respectively these three wave bands are carried out with Steerable filter using formula (3)-(10), output is designated as Qj respectively, Corresponding window coefficient matrix is designated as Aj and Bj, i.e.,
Qj=Aj × I+Bj, (j=2,3,5)
Here regularisation parameter ε=0.01, windows radius r=2, window size is 5 × 5.
The resampling amplification of 3rd step, window coefficient
Using arest neighbors interpolation method to window coefficient matrix A j and Bj resampling quadruplication, corresponding new window is respectively obtained Mouth coefficientWith
4th step, Steerable filter output
Utilize panchromatic light image PAN and new window coefficientWithCalculate output imageI.e.
The output imageIt is identical with original panchromatic light image PAN sizes, it is that 800 rows × 800 are arranged.
5th step, calculating strength component INT：
By formula (12) to output imageWithIt is averaged, calculates strength component INT, i.e.,
The IHS conversion of 6th step, multispectral image
By formula (13) and (14) to three selected wave band multispectral image M2、M3、M5IHS conversion is carried out, it is specific as follows
Resampling interpolation, quadruplication and the panchromatic same sizes of light image PAN, note are carried out to U, V using bicubic interpolation method ForWith
7th step, IHS inverse transformations
According to formula (15)-(17) to strength component INT andWithIHS inverse transformations are carried out, original color space is obtained Three band images, correspondence three wave bands output is designated as S respectively2、S3、S5, i.e.,
8th step, spatial detail are extracted, modulation is with injecting
Spatial detail D is extracted using formula (18), i.e.,
D=PAN-INT
Using arest neighbors interpolation method to all multi light spectrum hands (M1..., M8) carry out resampling interpolation, quadruplication to The panchromatic same sizes of light image PAN, are designated as Size is that 800 rows × 800 are arranged.
The details index of modulation g of each wave band is calculated using formula (19)j, j=1 ..., 8, i.e.,
Output image F is calculated using formula (20-22)2、F3And F5I.e.
F2=S2+g2×D
F3=S3+g3×D
F5=S5+g5×D
F2、F3And F5The fusion results as exported.

Claims (1)

1. a kind of remote sensing image fusion method based on Steerable filter, it is characterised in that comprise the steps：
The first step, carries out LPF, low pass filter is designated as LPF, then output figure by convolution algorithm to panchromatic light image PAN As PL=PAN*LPF；Make lower 2 sample process to output image PL, output result is designated as P2；
Second step, sets guide image I, any j-th of multi light spectrum hands M that image P2 is Steerable filterjFor the defeated of Steerable filter Enter image P, Steerable filter output image is designated as Q；Pixel i is located at the window ω centered on pixel kkIn, the window half Footpath r >=1, window size is (2r+1) × (2r+1) pixels；Filter filtering output value Qs of the output image Q in pixel iiBy referring to Lead pixel value Is of the image I in the pointiAnd the window coefficient a of local window where i pointskAnd bkTogether decide on,Matrix form is that Q=A × I+B, wherein A and B are window coefficient matrixes, Window coefficientWindow coefficient| ω | it is window ωkThe number of middle all pixels, piFor pixel values of the input picture P at i, μkAnd σkRefer respectively to lead image in window ωkIn Average and variance,It is input picture P in window ωkIn average, ε be regularisation parameter and ε ∈ (0, 1)；
3rd step, is amplified to window coefficient matrix A and B resamplings using arest neighbors interpolation method, respectively obtains corresponding new window CoefficientWith
4th step, utilizes original panchromatic light image PAN and new window coefficientWithCalculate large-sized output imageThe output imageIt is identical with original panchromatic light image size；
5th step, the optional three wave bands multispectral image M from n band multispectral imageT1、MT2And MT3；Repeat second step To the 4th step, corresponding three width large scale output image is calculatedWithCalculate strength component
6th step, to three selected wave band multispectral image MT1、MT2、MT3IHS conversion is carried out, is obtainedUtilize bicubic interpolation method pair U, V carry out resampling interpolation, be amplified to the original panchromatic same sizes of light image PAN, be designated asWith
7th step, to strength component INT andWithIHS inverse transformations are carried out, the band image of original color space three is obtained；With Three wave band multispectral image MT1、MT2、MT3Corresponding IHS inverse transformations output
8th step, calculates spatial detail D=PAN-INT；Utilize all more light of the arest neighbors interpolation method to original multispectral image Compose wave band M1..., MNCarry out resampling interpolation, be amplified to the panchromatic same sizes of light image PAN, be designated asJ=1 ..., N；Meter Calculate the spatial detail index of modulation of each wave bandThe utilization space details index of modulation modulates spatial detail D After inject three wave band multispectral image ST1、ST2、ST3In, output image FT1=ST1+gT1×D、FT2=ST2+gT2× D and FT3= ST3+gT3× D is fusion results；
9th step, returns to the 5th step and selects three different wave bands to be merged, finally travel through all N number of wave bands, export N number of wave band Fusion results.
CN201710242117.4A 2017-04-14 2017-04-14 Remote sensing image fusion method based on guided filtering Active CN107146212B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710242117.4A CN107146212B (en) 2017-04-14 2017-04-14 Remote sensing image fusion method based on guided filtering

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710242117.4A CN107146212B (en) 2017-04-14 2017-04-14 Remote sensing image fusion method based on guided filtering

Publications (2)

Publication Number Publication Date
CN107146212A true CN107146212A (en) 2017-09-08
CN107146212B CN107146212B (en) 2020-05-01

Family

ID=59774586

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710242117.4A Active CN107146212B (en) 2017-04-14 2017-04-14 Remote sensing image fusion method based on guided filtering

Country Status (1)

Country Link
CN (1) CN107146212B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107958450A (en) * 2017-12-15 2018-04-24 武汉大学 Panchromatic multispectral image fusion method and system based on adaptive Gaussian mixture model
CN108537765A (en) * 2018-03-12 2018-09-14 西北工业大学 A kind of spaceborne PAN and multi-spectral image interfusion method

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102446351A (en) * 2010-10-15 2012-05-09 江南大学 Multispectral and high-resolution full-color image fusion method study
CN103218796A (en) * 2013-05-14 2013-07-24 中国科学院自动化研究所 Fusion method of full color-multispectral remote sensing images
CN105761214A (en) * 2016-01-14 2016-07-13 西安电子科技大学 Remote sensing image fusion method based on contourlet transform and guided filter

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102446351A (en) * 2010-10-15 2012-05-09 江南大学 Multispectral and high-resolution full-color image fusion method study
CN103218796A (en) * 2013-05-14 2013-07-24 中国科学院自动化研究所 Fusion method of full color-multispectral remote sensing images
CN105761214A (en) * 2016-01-14 2016-07-13 西安电子科技大学 Remote sensing image fusion method based on contourlet transform and guided filter

Non-Patent Citations (6)

* Cited by examiner, † Cited by third party
Title
AMINA JAMEEL等: "Guided Filter and IHS based Pan-Sharpening", 《IEEE SENSORS JOURNAL》 *
WENFEI ZHAO等: "Remote Sensing Fusion Based on Guided Image Filtering", 《MIPPR 2015: REMOTE SENSING IMAGE PROCESSING, GEOGRAPHIC INFORMATION SYSTEMS》 *
WENZHI LIAO等: "TWO-STAGE FUSION OF THERMAL HYPERSPECTRAL AND VISIBLE RGB IMAGE BY PCA AND GUIDED FILTER", 《2015 7TH WORKSHOP ON HYPERSPECTRAL IMAGE AND SIGNAL PROCESSING: EVOLUTION IN REMOTE SENSING (WHISPERS)》 *
YONG YANG等: "Remote Sensing Image Fusion Based on Adaptive IHS and Multiscale Guided Filter", 《IEEE ACCESS》 *
李甜甜: "基于导向滤波和多尺度分析的图像融合算法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑（月刊）计算机软件及计算机应用》 *
那彦，任梦乔: "基于引导滤波的遥感图像融合算法", 《电子科技》 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107958450A (en) * 2017-12-15 2018-04-24 武汉大学 Panchromatic multispectral image fusion method and system based on adaptive Gaussian mixture model
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
CN108537765A (en) * 2018-03-12 2018-09-14 西北工业大学 A kind of spaceborne PAN and multi-spectral image interfusion method

Also Published As

Publication number Publication date
CN107146212B (en) 2020-05-01

Similar Documents

Publication Publication Date Title
Mei et al. Spatial and spectral joint super-resolution using convolutional neural network
Wang et al. Fusion of Landsat 8 OLI and Sentinel-2 MSI data
Luo et al. Pansharpening via unsupervised convolutional neural networks
Jiang et al. A differential information residual convolutional neural network for pansharpening
Zhang et al. GTP-PNet: A residual learning network based on gradient transformation prior for pansharpening
CN104867124B (en) Multispectral and panchromatic image fusion method based on the sparse Non-negative Matrix Factorization of antithesis
CN104851077B (en) A kind of panchromatic sharpening method of adaptive remote sensing images
CN105844653B (en) A kind of multilayer convolutional neural networks optimization system and method
CN109064396A (en) A kind of single image super resolution ratio reconstruction method based on depth ingredient learning network
CN105469360A (en) Non local joint sparse representation based hyperspectral image super-resolution reconstruction method
CN109102469A (en) A kind of panchromatic sharpening method of remote sensing images based on convolutional neural networks
CN107220957B (en) It is a kind of to utilize the remote sensing image fusion method for rolling Steerable filter
CN102542549A (en) Multi-spectral and panchromatic image super-resolution fusion method based on compressive sensing
CN101930604B (en) Infusion method of full-color image and multi-spectral image based on low-frequency correlation analysis
Zhang et al. Image fusion employing adaptive spectral-spatial gradient sparse regularization in UAV remote sensing
CN107545554A (en) A kind of two-stage remote sensing image fusion method based on Steerable filter
CN107146212A (en) A kind of remote sensing image fusion method based on Steerable filter
CN102831581A (en) Method for reconstructing super-resolution image
CN108537765A (en) A kind of spaceborne PAN and multi-spectral image interfusion method
Zhang et al. SC-PNN: Saliency cascade convolutional neural network for pansharpening
Li et al. Progressive split-merge super resolution for hyperspectral imagery with group attention and gradient guidance
Zhong et al. Attention_FPNet: Two-branch remote sensing image pansharpening network based on attention feature fusion
CN115984155A (en) Hyperspectral, multispectral and panchromatic image fusion method based on spectrum unmixing
Zhou et al. PAN-guided band-aware multi-spectral feature enhancement for pan-sharpening
CN107451954A (en) Iterated pixel interpolation method based on image low-rank property

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