CN115830185A - Double-window simulation reconstruction method and device for short-wave infrared remote sensing image - Google Patents
Double-window simulation reconstruction method and device for short-wave infrared remote sensing image Download PDFInfo
- Publication number
- CN115830185A CN115830185A CN202211520657.1A CN202211520657A CN115830185A CN 115830185 A CN115830185 A CN 115830185A CN 202211520657 A CN202211520657 A CN 202211520657A CN 115830185 A CN115830185 A CN 115830185A
- Authority
- CN
- China
- Prior art keywords
- image
- remote sensing
- resolution
- window
- spatial
- 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.)
- Pending
Links
Images
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses a double-window simulation reconstruction method for a short-wave infrared remote sensing image, which comprises the following steps: step 1, inputting a remote sensing image; step 2, determining the size NxN of the mobile window; step 3, adjusting the spatial resolution of the remote sensing image B; step 4, superposing and recombining the images A and C1; step 5, VNIR + SWIR wave band image space scale reduction; step 6, improving the spatial scale of the high-resolution image VNIR wave band image; step 7, calculating the optimal similar pixel of the low-resolution image VNIR + SWIR band image; step 8, reconstructing a high-resolution SWIR band image spectrum; and 9, reconstructing to generate a high-resolution SWIR band image. The short wave infrared remote sensing image simulation reconstruction method based on double window analysis and spatial scale conversion provided by the invention can be used for remarkably improving the spatial resolution of the short wave infrared image, well maintaining the original spectral characteristics, has higher operation efficiency, is simpler and more concise, consumes less time in operation and has more accurate calculation results.
Description
Technical Field
The invention relates to the technical field of digital image processing in photogrammetry and remote sensing disciplines, in particular to a method and a device for double-window analog reconstruction of a short-wave infrared remote sensing image.
Background
With the development of satellite remote sensing technology, a stereoscopic space observation system with multiple spatial scales, high spectrum and multiple time resolutions is gradually improved, and the system is widely applied to various fields closely related to national life, such as environmental monitoring, national survey, trip navigation and the like.
However, due to the limitations of the remote sensors themselves, it is often difficult to have both high spectral and spatial resolution. Most high spatial resolution satellites such as IKONOS, quickbird, worldView-1, and WorldView-2 are limited to the Visible Near Infrared (VNIR) band and lack the Short Wave Infrared (SWIR) band. The short wave infrared band imaging has the advantages of small atmospheric scattering effect, strong smoke, fog or haze transmission capability, long effective detection distance and the like, and has obviously better adaptability to climatic conditions and environment than visible light imaging. The problem of missing of high-spatial-resolution short-wave infrared data is solved, and the method becomes a research hotspot to be solved urgently in the field of remote sensing application. The image simulation is a technology for obtaining a simulated image by utilizing mathematical calculation or physical modeling under a certain constraint condition according to the existing remote sensing image, remote sensing prior knowledge and a remote sensing theoretical model. The short-wave infrared image missing from the high-spatial-resolution satellite can be made up by the image simulation technology, and the method has important theoretical and practical significance.
Space-spectrum image simulation is a more mature image simulation technique that is widely used. A variety of data simulation methods have been developed for many years of use. For example, different images acquired by different imaging modes of the same sensor, or image fusion of different sensors, or image combination of images acquired by the same ground object at different times, so that the simulated image has the detailed spatial characteristics of the high spatial image and the spectral characteristics of the multispectral image. The mainstream methods mainly include high-pass filtering, brovey transform, PCA replacement, IHS transform, wavelet transform, etc., and various improved methods based on the methods are generated.
In the prior art, patent CN106384340A discloses a method for performing image fusion on remote sensing images with different spatial resolutions in different periods, which obtains a high spatial resolution image in a later period by analyzing an early low spatial resolution image and a later low spatial resolution image to obtain change data. The structural mode of the calculation model structure is developed into a linear spiral mode from a planar window mode, so that a calculation pixel set can be constructed in a targeted manner, and the upper limit of the component quantity of the mixed pixel set calculated by the downscaling fusion method is improved.
The invention patent application document CN112102218A discloses a fusion method for generating a high spatial resolution multispectral image, which is used for fusing an image with high spatial resolution and less spectral bands with an image with low spatial resolution and more spectral bands to generate an image with high spatial resolution and multispectral bands. And eliminating the plaque effect in the fusion result by using residual compensation and moving the information of the neighborhood similar pixels in the window, thereby improving the fusion precision.
However, the method is complex, the calculation efficiency is still not high, the requirement on the ground object similarity of two remote sensing images for calculation is high, and when the time phase difference of two different spatial resolution images is large or the difference of the ground surface type is large, large spectral distortion is easily generated.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: the short-wave infrared remote sensing image simulation reconstruction method and the implementation device thereof solve the existing technical problems and are simpler, less in calculation time consumption and more accurate in calculation result.
The technical scheme adopted by the invention for solving the technical problems is as follows:
a double-window simulation reconstruction method for short-wave infrared remote sensing images comprises the following steps:
step 1, inputting remote sensing images A and B with different spatial resolutions after geometric correction;
and 9, simulating and reconstructing to generate a high-spatial-resolution SWIR band image E.
Preferably, the remote sensing images a and B of step 1 are geometrically corrected high-resolution remote sensing images a containing visible light-near infrared (VNIR) m1 wave bands and high-resolution remote sensing images a containing visible light-near infrared and short-wave infrared
(VNIR + SWIR) m2 bands of low spatial resolution remote sensing image B.
Preferably, the step 2 is to calculate a ratio R of the spatial resolution of the visible-near infrared and short wave infrared (VNIR + SWIR) band remote sensing image B to the spatial resolution of the visible-near infrared band remote sensing image a, and determine the size of the moving window to be N × N according to a numerical interval of R.
Preferably, the new image C in step 3 1 The generation method specifically comprises the following steps: the total m2 visible-near infrared and short wave infrared (VNIR + SWIR) bands of the low spatial resolution image B are resampled to N times the high resolution remote sensing image a to generate new m2 images C1 containing visible-near infrared and short wave infrared (VNIR + SWIR) bands.
Preferably, the method for generating the image D in step 4 includes: taking the spatial resolution of the image A as a reference, resampling, overlapping and recombining the two groups of remote sensing images according to the sequence of the images A and C1 to generate a new image D with m1+ m2 wave bands; low spatial resolution visible-near infrared and short wave infrared (VNIR + SWIR) band images C with m2 bands 1 The image size of (a) is W1 × W2 pixels, and the image size of the m1+ m2 band picture D superposed and recombined is set to (N × N) × (W1 × W2) pixels.
Preferably, the specific method for implementing the spatial scale reduction in step 5 is as follows: low spatial resolution remote sensing image C in image D by NxN moving window 1 Performing window domain operations on the corresponding m2 visible-near infrared and short wave infrared (VNIR + SWIR) band images to realize spatial scale reductionAnd (4) originally.
Preferably, the step 6 specifically includes taking an nxn window for each pixel of the image D with the pixel as the center, calculating a MEAN and a standard deviation STD of a gray value (DN) of the high-spatial-resolution RED light (RED) band image, determining the pixel with the gray value (DN) in a (MEAN-0.5 × STD, MEAN +0.5 × STD) interval as a dominant object pixel, and identifying; calculating the average value of each VNIR wave band spectrum of the object pixels with high spatial resolution dominance in the window, and assigning to the central pixel of the window to obtain a synthetic spectrum vector Y (1, 2, \ 8230;, m 1) of N × N window 1-m 1 wave bands, thereby promoting the spatial scale of the synthetic spectrum Y to be consistent with the image C 2 The same spatial scale.
Preferably, the step 7 specifically comprises: an NxN window of the high spatial resolution image D is mapped to a 1-m 1 band synthesized spectral vector Y (1, 2, \ 8230;, m 1) of the high spatial resolution image A and to the low spatial resolution image C 2 A pixel of (1); at low spatial resolution image C 2 In addition, an NxN window is taken to obtain NxN spectral vectors X (1, 2, \ 8230;, m 2) of visible light-near infrared and short wave infrared (VNIR + SWIR) wave bands; the values X (1, 2, \ 8230;, m 1) and Y (1, 2, \ 8230;, m 1) of the first m1 wave bands of the spectral vector X (1, 2, \ 8230;, m 2) are fitted by the least square method, and the image C corresponding to the maximum value of the correlation number is fitted 2 The pixel is judged as the optimal similar pixel and identified; the optimal similar pixel corresponds to the image C 2 The visible-near infrared and short-wave infrared (VNIR + SWIR) band spectra of (A) are the optimal spectral vector Xmax (1, 2, \ 8230;, m 2).
Preferably, the step 8 specifically includes: for the optimal spectral vector obtained in the above steps
Xmax (1, 2, \ 8230;, m 2) takes the value X (1, 2, \ 8230;, m 1) of its first m1 bands and the resultant spectral vector Y
(1, 2, \ 8230;, m 1) performing spectrum fitting by using least square method to obtain values of parameters a and b of fitting function y = a × x + b, and then using optimal spectrum vector Xmax (1, 2, \ 8230;, m 2) to obtain spectrum value of Short Wave Infrared (SWIR) band
Substituting Xmax (m 1+1, m 2) into the fitting function to obtain Y (m 1+1, m 2), and determining the reconstruction spectrums of m2-m1 SWIR wave bands with high spatial resolution; the step 9 specifically comprises: and (3) executing the steps 6 to 8 for each pixel of the high-spatial-resolution image D, and simulating and reconstructing to generate m2-m1 short-wave infrared (SWIR) band images E with high spatial resolution.
A device for realizing the double-window simulation reconstruction method of the shortwave infrared remote sensing image according to claim 1 comprises remote sensing image input modules which are in communication connection with each other and are used for inputting remote sensing images A and B with different spatial resolutions after geometric correction;
the moving window size determining module is used for calculating the ratio R of the spatial resolution of the remote sensing image B to the spatial resolution of the remote sensing image A and determining the size NxN of the moving window;
a spatial resolution adjusting module for resampling the low spatial resolution image B to N times of the resolution of the high resolution remote sensing image A to generate a new image C 1
An image superposition and recombination module for superposing and recombining the images A and C based on the spatial resolution of the image A 1 Superposing and recombining to generate a new image D;
an image space scale reduction module for carrying out the low spatial resolution remote sensing image C in the image D by using the NxN moving window 1 Performing window domain operation to obtain image C 2 To achieve spatial scale reduction;
the image space scale lifting module is used for judging the dominant object pixels of the NxN window in the image D, calculating a synthetic spectral vector Y and lifting the space scale of the synthetic spectral vector Y to the image C 2 Consistency;
the optimal similar pixel calculation module is used for judging the optimal similar pixel of the NxN window in the low-resolution image to obtain an optimal spectral vector Xmax;
the spectrum simulation reconstruction module is used for reconstructing to obtain a Short Wave Infrared (SWIR) band spectrum with high spatial resolution by using the optimal spectrum vector Xmax and the fitting function;
and the high-spatial-resolution SWIR band image construction module is used for simulating and reconstructing to generate a high-spatial-resolution short-wave infrared (SWIR) band image E.
The invention has the beneficial effects that: the short wave infrared remote sensing image simulation reconstruction method based on double window analysis and spatial scale conversion provided by the invention can be used for remarkably improving the spatial resolution of the short wave infrared image, well maintaining the original spectral characteristics, has higher operation efficiency, is simpler and more concise, consumes less time in operation and has more accurate calculation results.
Drawings
Fig. 1 is a flowchart of a short-wave infrared remote sensing image simulation reconstruction method based on dual window analysis and spatial scale conversion in embodiment 11 of the present invention;
FIG. 2 is a high spatial resolution visible image A according to an embodiment of the present invention;
FIG. 3 is a low spatial resolution short wave infrared image B according to embodiment 1 of the present invention;
FIG. 4 is a schematic diagram of dual window analysis and spatial scale transformation in example 11 of the present invention;
FIG. 5 is a high spatial resolution SWIR band image E generated by simulation reconstruction in an embodiment of the present invention;
FIG. 6 is a scatter diagram of predicted and actual spectral values of SWIR band in example 1 of the present invention;
fig. 7 is a schematic structural diagram of an implementation apparatus of a short-wave infrared remote sensing image simulation reconstruction method based on dual window analysis and spatial scale conversion in embodiment 1 of the present invention.
Detailed Description
The invention is further illustrated with reference to the following figures and examples, which are not intended to limit the scope of the invention in any way.
Example 1
The invention relates to a double-window simulation reconstruction method of a short wave infrared remote sensing image, which comprises the following steps:
referring to fig. 1, step 1, remote sensing image input;
images A of 4 VNIR bands of a Quickbird satellite with 2.4 m spatial resolution and images B of 6 VNIR + SWIR bands of a Landsat-5 terrestrial satellite TM with 30 m spatial resolution are input with geometric correction.
The Quickbird satellite is a commercial satellite capable of providing a sub-meter resolution, and can provide satellite images with 0.61 meter resolution full color and 2.4 meter resolution 4 multispectral wave bands. The Thematic Mapper (TM) carried by Landsat-5 has 4 VNIR bands and 2 SWIR bands with spatial resolution of 30 meters, and a thermal infrared band with spatial resolution of 120 meters.
In the embodiment, the remote sensing data selected are a QuickBird image of 2011, 9 and 14 days and a Landsat5-TM image of 2010, 10 and 16 days (center coordinates E36 degrees 3 'and N21 degrees 9') of certain regions of african sudan. The QuickBird image is divided into 1800 × 1800 pixels, and the Landsat5-TM image in the corresponding spatial range is divided into 144 × 144 pixels, which are used as the experimental data of the embodiment of the present invention.
Specifically, fig. 2 shows a high spatial resolution QuickBird image a in the present embodiment.
the spatial resolution of a visible light-near infrared and short wave infrared band remote sensing image B of the Landsat5-TM image is 30 meters, the spatial resolution of a visible light-near infrared band remote sensing image A of the Quickbird is 2.4 meters, the ratio R of the visible light-near infrared band remote sensing image B to the Quickbird remote sensing image A is equal to 12.5, and the value range is located when the ratio R is larger than 10, so that the size of the moving window is determined to be 9 x 9.
the spatial resolution of the VNIR band remote sensing image A of the Quickbird is 2.4 meters, and the spatial resolution is 21.6 meters after the VNIR band remote sensing image A is multiplied by 9 times. By re-sampling, the spatial resolution of the TM 6 VNIR + SWIR band remote sensing images B with a spatial resolution of 30 m is adjusted to 21.6 m, and the image size of the newly generated image C1 in the area corresponding to the 1800 × 1800 pixel QuickBird image is 200 × 200 pixels. That is, the spatial extent of 1 TM image C1 pixel corresponds exactly to 81 pixels of the Quickbird image A.
Specifically, fig. 3 shows a low spatial resolution Landsat5-TM image C1 according to an embodiment of the invention.
based on 2.4 m of spatial resolution of the QuickBird image a, two groups of remote sensing images are resampled and superposed to recombine according to the sequence of the images a and C1, so as to generate a new image D with 10 (i.e. 4+ 6) wave bands.
Therefore, in step 3, the image size of the 6 VNIR + SWIR band images C1 of the low spatial resolution TM is 200 × 200 pixels, and the image size of the superimposed and recombined 10 band images D is (9 × 9) × (200 × 200) pixels.
window domain operations are performed on the 6 VNIR + SWIR band images corresponding to the low spatial resolution TM image C1 in the image D with a 9 × 9 moving window to implement spatial scale reduction.
The specific operation rule is as follows: the image D is divided into 200 × 200 windows with the size of 9 × 9, for each waveband in the TM image with 6 wavebands, the spectral average value of the pixel of each 9 × 9 window is sequentially calculated and assigned to the central pixel of the window, and a new image C2 with the image size of 200 × 200 and 6 VNIR + SWIR wavebands is generated, so that the spatial scale of the images with 6 VNIR + SWIR wavebands in the image D is restored to the scale level of the image C1. So that a strict spatial correspondence exists between the image C2 and the image D, i.e., the pixels of one image C2 correspond to the pixels of 9 × 9 images D, creating conditions for subsequent dual window analysis.
for each pixel of the image D, taking a 9 multiplied by 9 window by taking the pixel as the center, calculating the MEAN and standard deviation STD of the gray value (DN) of the high-spatial-resolution QuickBird RED light (RED) band image, judging the pixel of which the DN value is located in the (MEAN-0.5 multiplied STD, MEAN +0.5 multiplied STD) interval as the dominant pixel, and identifying. The average value of each VNIR wave band spectrum of the high spatial resolution dominant object pixels in the window is calculated and assigned to the central pixel of the window to obtain a synthetic spectrum vector Y (1, 2, \ 8230;, 4) of 1-4 wave bands of the 9 x 9 window, so that the spatial scale of the synthetic spectrum Y is promoted to the same spatial scale as the image C2.
Specifically, in the present embodiment, the synthesized spectral vector of 1-4 bands of a 9 × 9 window QuickBird image is Y (350, 584,470, 498).
Specifically, as shown in fig. 4, a schematic diagram of dual window analysis and spatial scale conversion in the embodiment of the present invention is shown.
a 9 x 9 window of the high spatial resolution image D corresponds to a 1-4 band synthesized spectral vector Y (1, 2, \ 8230;, 4) of the high spatial resolution QuickBird image, and also corresponds to a pixel of the low spatial resolution TM image C2; another 9 × 9 window is taken in the low spatial resolution TM image C2 to obtain 9 × 9 spectral vectors X (1, 2, \ 8230;, 6) of VNIR + SWIR bands; fitting values X (1, 2, \8230;, 4) of the first 4 wave bands of the spectral vector X (1, 2, \8230;, 6) and Y (1, 2, \8230;, 4) by using a least square method, judging an image C2 pixel corresponding to the maximum value of the relative number as an optimal similar pixel and identifying; the VNIR + SWIR band spectrum of the image C2 corresponding to the optimal similar pixel is the optimal spectral vector Xmax (1, 2, \ 8230;, 6).
Specifically, the embodiment has a synthetic spectral vector of a certain 9 × 9 window Quickbird 1-4 wave band as Y (584, 450,498, 470); one spectral vector of TM image 1,2,3,4,5,7 wave bands (VNIR + SWIR wave bands) in 9X 9 window is X (85, 42,64,52,120, 72); in 81 TM image pixels in the 9 × 9 window, the maximum correlation coefficient is 0.9821, the pixel corresponding to the maximum correlation coefficient is determined and the optimal similar pixel is identified, and the optimal spectral vector Xmax (87, 47,62,58,123, 69) corresponding to the 1,2,3,4,5,7 waveband of the TM image is obtained.
and (3) performing spectral fitting on the optimal spectral vector Xmax (1, 2, \ 8230;, 6) obtained in the above step by using the values X (1, 2, \ 8230;, 4) of the first 4 wave bands and the synthetic spectral vector Y (1, 2, \ 8230;, 4) by using a least square method to obtain the values of parameters a and b of a fitting function Y = a X + b, and substituting the spectral values Xmax (5, 6) of the SWIR wave bands of the optimal spectral vector Xmax (1, 2, \ 8230;, 6) into the fitting function to obtain Y (5, 6), namely the reconstructed spectrum of 2 SWIR wave bands with high spatial resolution, namely the spectrum simulation of 5 wave bands and 7 wave bands is used for reconstructing 2 spectra of the Quickbird image corresponding to the SWIR wave bands.
Specifically, in this embodiment, the optimal spectral vector of a certain 9 × 9 window TM
And taking the values of 4 wavebands of the Xmax (87, 47,62,58,123 and 69) to obtain a spectrum vector Xmax (87, 47,62 and 58), and performing spectrum fitting on the spectrum vector Y (584, 450,498 and 470) of a synthesized spectrum vector of 1-4 wavebands of a certain 9 × 9 window Quickbird by using a least square method to obtain a fitting function Y =3.4621 × x +280.66. And substituting the values of the optimal spectral vectors Xmax (123, 69) of 2 SWIR wave bands of the TM image into the fitting function to obtain the reconstructed spectral values of 2 SWIR wave bands of the Quickbird image.
and (6) performing steps 6 to 8 on each pixel of the high-spatial-resolution image D, and simulating and reconstructing m2-m1 SWIR band images E with high spatial resolution.
Specifically, in this embodiment, 2 new SWIR band images E with high spatial resolution corresponding to TM images 5 and 7 bands are obtained through reconstruction, as shown in fig. 5, and fig. 5 is a high spatial resolution SWIR band image E generated by simulation reconstruction in the embodiment of the present invention.
For comparison research, result comparison and analysis are respectively carried out by adopting the method, a Hyperspectral Super-resolution (Hy Sure) based method and a method for improving multispectral image fusion precision by residual compensation (residual compensation for short) disclosed in patent application document CN 112102218A.
The principle and calculation procedure of the method for hyperspectral super resolution (hyper Sure) is given in document 1 (Simoes M, bioucas-Dias J, almeida L B & Chanussot Jocellin. Aconvex formation for hyperspectral image super resolution video-based regulation [ J ]. IEEE Transactions on Geoscience and Remote Sensing,2015,53 (6): 3373-3388).
Specifically, as shown in fig. 6, fig. 6 is a scatter diagram of predicted spectral values and actual spectral values of SWIR band in the embodiment of the present invention
In this embodiment, the computation time of the method of the present invention, the Hy Sure method, and the residual error compensation method is 3.5 seconds, 7.8 seconds, and 4.6 seconds, respectively, which shows that the method of the present invention has a significant advantage in the aspect of computation efficiency. The ground feature shape and the spectrum value in the prediction result image of the three methods keep high consistency with the original M5 and TM7 wave band images, the space detail is well reserved, and the space resolution is obviously improved. The gray level images of the calculation results of the three methods are quantized by 8 bits of 0-255, and the correlation coefficients of the reconstructed 2 SWIR band spectrums and the TM5 and TM7 band spectrum values are counted.
Fig. 6 shows scatter plots of the actual values of the TM5 and TM7 band spectra and the predicted values of the 2 corresponding SWIR band spectra for the three methods. In fig. 6, a, b, and c are scatter diagrams corresponding to the prediction results of the method of the present invention, the Hy Sure method, and the residual error compensation method, respectively. Correlation coefficients of predicted values and actual values of three methods of SWIR1 and SWIR2 wave bands are respectively as follows: 0.9456 and 0.9502,0.9263 and 0.9189,0.9237 and 0.9205, and the correlation coefficients obtained by the method are all over 0.94, which is better than the values of about 0.92 of other 2 methods. The closer the scatter point in the scatter diagram is to 1, the better the baseline indicates that the consistency between the fusion result and the actual result is better, so the scatter point in the scatter diagram of the predicted result by the method described in this embodiment is closer to 1:1 base line and very concentrated; in contrast, the prediction results of the Hy Sure method and the residual compensation method are more dispersed and deviate from 1:1 the reference line is far. From the visual effect, the Hy Sure method and the residual compensation method have certain color distortion in the predicted image. The Hy Sure method is complex in process parameter optimization and is a great challenge for users, and the Hy Sure has low calculation efficiency and consumes a great amount of time in the face of large-scale images. When the phase difference is large in the face or the ground coverage type difference of two types of images is large, the residual error compensation method also has the condition that the spectrum fitting function cannot be constructed to cause the spectrum reconstruction of the local pixel to be invalid.
In conclusion, the correlation coefficient and the scatter diagram of the visual observation and prediction spectrum and the actual spectrum prove that the short-wave infrared remote sensing image simulation reconstruction method based on the double-window analysis and the spatial scale conversion can well keep the original spectrum characteristic while remarkably improving the spatial resolution of the short-wave infrared image and has higher operation efficiency.
On the other hand, the embodiment of the invention also provides a device for realizing the short-wave infrared remote sensing image simulation reconstruction method based on double window analysis and space scale conversion.
Fig. 7 is a schematic structural diagram of an implementation apparatus of the short-wave infrared remote sensing image simulation reconstruction method based on dual window analysis and spatial scale conversion according to the embodiment of the present invention. The realizing device is connected with each other in a communication way and comprises:
the remote sensing image input module 101 inputs the high spatial resolution visible light-near infrared image a and the low spatial resolution Short Wave Infrared (SWIR) image B which are subjected to geometric correction.
And the moving window size determining module 102 is used for calculating the ratio R of the spatial resolution of the remote sensing image B to the spatial resolution of the remote sensing image A and determining the size NxN of the moving window.
The spatial resolution adjustment module 103 resamples the low spatial resolution image B to N times the resolution of the high resolution remote sensing image a to generate a new image C1.
The image stacking and reconstructing module 104 uses the spatial resolution of the image A as the reference to stack the images A and C 1 And superposing and recombining to generate a new image D.
The image space scale reduction module 105 uses an NxN moving window to perform low spatial resolution remote sensing image C in the image D 1 Performing window domain operation to obtain image C 2 To achieve spatial scale reduction.
The image spatial scale lifting module 106 determines dominant pixels of the nxn window in the image D, calculates the synthesized spectral vector Y, and lifts the spatial scale to the image C 2 And (5) the consistency is achieved.
The optimal similar pixel calculation module 107 determines the optimal similar pixel of the nxn window in the low-resolution image to obtain the optimal spectral vector X max 。
The spectral modeling reconstruction module 108 uses the optimal spectral vector X max And fitting the function y = a × x + b, and reconstructing to obtain a SWIR band spectrum with high spatial resolution.
The high spatial resolution SWIR band image construction module 109 repeatedly executes steps 6 to 8, and generates the high spatial resolution SWIR band image E through simulation reconstruction.
The device for implementing the short-wave infrared remote sensing image simulation reconstruction method based on the dual window analysis and the spatial scale conversion is used for implementing the short-wave infrared remote sensing image simulation reconstruction method based on the dual window analysis and the spatial scale conversion in the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which are not described herein again.
The foregoing shows and describes the general principles, features and advantages of the present invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, which are described in the specification and illustrated only for the purpose of illustrating the principles of the present invention, but that various changes and modifications may be made therein without departing from the spirit and scope of the invention as defined in the appended claims.
Claims (10)
1. A double-window simulation reconstruction method for short-wave infrared remote sensing images is characterized by comprising the following steps:
step 1, inputting remote sensing images A and B with different spatial resolutions after geometric correction;
step 2, calculating a ratio R of spatial resolutions of the remote sensing image B and the remote sensing image A, and determining the size NxN of the moving window;
step 3, resampling the low spatial resolution image B to N times of the resolution of the high resolution remote sensing image A, and generating a new image C 1 ；
Step 4, taking the spatial resolution of the image A as the reference, and taking the images A and C as the reference 1 Superposing and recombining to generate a new image D;
step 5, using the NxN moving window to perform low spatial resolution remote sensing image C in the image D 1 Performing window domain operation to obtain an image C 2 To achieve spatial scale reduction;
step 6, judging the dominant object pixels of the NxN window in the image D, calculating a synthetic spectral vector Y, and promoting the spatial scale of the synthetic spectral vector Y to the image C 2 The consistency is achieved;
step 7, judging the optimal similar pixel of the NxN window in the low-resolution image to obtain an optimal spectral vector Xmax;
step 8, reconstructing by using the optimal spectrum vector Xmax and a fitting function to obtain a short wave infrared band spectrum with high spatial resolution;
and 9, simulating and reconstructing to generate a high-spatial-resolution short-wave infrared band image E.
2. The double-window simulation reconstruction method for the short-wave infrared remote sensing image according to claim 1, wherein the remote sensing images A and B in the step 1 are geometrically corrected high-resolution remote sensing images A containing visible light-near infrared m1 wave bands and low-spatial-resolution remote sensing images B containing visible light-near infrared and short-wave infrared m2 wave bands.
3. The double-window simulation reconstruction method for the shortwave infrared remote sensing image according to claim 2, wherein the step 2 specifically comprises: and calculating the ratio R of the spatial resolution of the visible light-near infrared and short wave infrared band remote sensing images B to the spatial resolution of the visible light-near infrared band remote sensing images A, and determining the size of the moving window to be NxN according to the numerical value interval of R.
4. The double-window simulation reconstruction method for short-wave infrared remote sensing images according to claim 3, wherein the new image C in the step 3 1 The generation method specifically comprises the following steps: resampling all m2 visible light-near infrared and short wave infrared bands of the low spatial resolution image B to N times of the high resolution remote sensing image A, and generating new m2 images C1 containing the visible light-near infrared and short wave infrared bands.
5. The double-window simulation reconstruction method for the short-wave infrared remote sensing image according to claim 4, wherein the image D generation method in the step 4 specifically comprises the following steps: taking the spatial resolution of the image A as a reference, resampling, overlapping and recombining the two groups of remote sensing images according to the sequence of the images A and C1 to generate a new image D with m1+ m2 wave bands; low latitude for setting m2 wave bandsIntermediate resolution visible light-near infrared + short wave infrared band image C 1 The image size of (a) is W1 × W2 pixels, and the image size of the m1+ m2 band picture D superposed and recombined is set to (N × N) × (W1 × W2) pixels.
6. The double-window simulation reconstruction method for the shortwave infrared remote sensing image according to claim 5, wherein the specific method for realizing the spatial scale reduction in the step 5 is as follows: low spatial resolution remote sensing image C in image D by NxN moving window 1 And carrying out window field operation on the corresponding m2 visible light-near infrared + short wave infrared band images to realize space scale reduction.
7. The double-window simulation reconstruction method for the short-wave infrared remote sensing image according to claim 6, wherein the step 6 is specifically to take an nxn window with the pixel as the center for each pixel of the image D, calculate the MEAN value and the standard deviation of the gray value of the high-spatial-resolution red-light band image, determine the pixel with the gray value in the (MEAN-0.5 xstd, MEAN +0.5 xstd) interval as the dominant land feature pixel, and identify the pixel; calculating the average value of each visible light-near infrared band spectrum of the high spatial resolution dominant object pixel in the window, and assigning the average value to the central pixel of the window to obtain a synthetic spectrum vector Y (1, 2, \ 8230;, m 1) of N × N window 1-m 1 bands, thereby increasing the spatial scale of the synthetic spectrum Y to the image C 2 The same spatial scale.
8. The double-window simulation reconstruction method for the shortwave infrared remote sensing image according to claim 7, wherein the step 7 specifically comprises: an NxN window of the high spatial resolution image D is mapped to a 1-m 1 band synthesized spectral vector Y (1, 2, \ 8230;, m 1) of the high spatial resolution image A and to the low spatial resolution image C 2 A pixel of (a); at low spatial resolution image C 2 In addition, an NxN window is taken to obtain NxN spectral vectors X (1, 2, \ 8230;, m 2) of visible light-near infrared + short wave infrared bands; the first m1 waves of the spectral vector X (1, 2, \8230;, m 2) are takenThe values X (1, 2, \ 8230;, m 1) and Y (1, 2, \ 8230;, m 1) of the segments are fitted by the least square method, and the image C corresponding to the maximum value of the correlation number is fitted 2 The pixel is judged as the optimal similar pixel and identified; the optimal similar pixel corresponding image C 2 The visible light-near infrared + short wave infrared band spectrum is the optimal spectrum vector Xmax (1, 2, \ 8230;, m 2).
9. The double-window simulation reconstruction method for the shortwave infrared remote sensing image according to claim 8, wherein the step 8 specifically comprises: for the optimal spectrum vector Xmax (1, 2, \ 8230;, m 2) obtained in the above step, the values X (1, 2, \ 8230;, m 1) of the previous m1 wave bands are taken, and the spectrum fitting is carried out on the optimal spectrum vector Xmax (1, 2, \ 8230;, m 1) and the synthesized spectrum vector Y (1, 2, \ 8230;, m 1) by using a least square method to obtain the values of parameters a and b of a fitting function Y = a X + b, and then the spectrum value Xmax (m 1+1, m 2) of the short wave infrared wave band of the optimal spectrum vector Xmax (1, 2, \\ 8230;, m 2) is substituted into the fitting function to obtain Y (m 1+1, m 2), and the reconstructed spectrum of the high spatial resolution m2-m1 short wave infrared wave band is determined; the step 9 specifically comprises: and (4) executing the steps 6 to 8 for each pixel of the high-spatial-resolution image D, and simulating and reconstructing to generate m2-m1 short-wave infrared band images E with high spatial resolution.
10. A device for realizing the double-window simulation reconstruction method of the shortwave infrared remote sensing image according to claim 1, characterized by comprising the shortwave infrared remote sensing image and the short-wave infrared remote sensing image which are mutually communicated and connected
The remote sensing image input module is used for inputting remote sensing images A and B with different spatial resolutions after geometric correction;
the moving window size determining module is used for calculating the ratio R of the spatial resolution of the remote sensing image B to the spatial resolution of the remote sensing image A and determining the size NxN of the moving window;
a spatial resolution adjusting module for resampling the low spatial resolution image B to N times of the resolution of the high resolution remote sensing image A to generate a new image C 1
An image superposition and recombination module for superposing and recombining the images A and C based on the spatial resolution of the image A 1 Superposing and recombining to generate a new image D;
image space scale reduction modelBlock for low spatial resolution remote sensing image C in image D with NxN moving window 1 Performing window domain operation to obtain an image C 2 To achieve spatial scale reduction;
an image spatial scale promotion module for determining dominant ground object pixels of NXN window in image D, calculating synthetic spectral vector Y, and promoting spatial scale to image C 2 The consistency is achieved;
the optimal similar pixel calculation module is used for judging the optimal similar pixel of the NxN window in the low-resolution image to obtain an optimal spectral vector Xmax;
the spectrum simulation reconstruction module is used for reconstructing by using the optimal spectrum vector Xmax and a fitting function to obtain a short wave infrared band spectrum with high spatial resolution;
and the high-spatial-resolution short-wave infrared band image construction module is used for simulating and reconstructing to generate a high-spatial-resolution short-wave infrared band image E.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211520657.1A CN115830185A (en) | 2022-11-29 | 2022-11-29 | Double-window simulation reconstruction method and device for short-wave infrared remote sensing image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211520657.1A CN115830185A (en) | 2022-11-29 | 2022-11-29 | Double-window simulation reconstruction method and device for short-wave infrared remote sensing image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115830185A true CN115830185A (en) | 2023-03-21 |
Family
ID=85533108
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211520657.1A Pending CN115830185A (en) | 2022-11-29 | 2022-11-29 | Double-window simulation reconstruction method and device for short-wave infrared remote sensing image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115830185A (en) |
-
2022
- 2022-11-29 CN CN202211520657.1A patent/CN115830185A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Weng et al. | Generating daily land surface temperature at Landsat resolution by fusing Landsat and MODIS data | |
| Song et al. | Spatiotemporal satellite image fusion through one-pair image learning | |
| Cao et al. | Thick cloud removal in Landsat images based on autoregression of Landsat time-series data | |
| Zhu et al. | An enhanced spatial and temporal adaptive reflectance fusion model for complex heterogeneous regions | |
| CN112991288B (en) | Hyperspectral remote sensing image fusion method based on abundance image sharpening reconstruction | |
| CN107966210B (en) | Hyperspectral image-based thermal infrared fusion reconstruction method | |
| CN112733596A (en) | Forest resource change monitoring method based on medium and high spatial resolution remote sensing image fusion and application | |
| CN114565843A (en) | Time series remote sensing image fusion method | |
| Zhu et al. | Robust registration of aerial images and LiDAR data using spatial constraints and Gabor structural features | |
| CN116245757B (en) | Multi-scene universal remote sensing image cloud restoration method and system for multi-mode data | |
| KR101132272B1 (en) | A method for pan-sharpening of high-spatial resolution satellite image by using parameter reflecting spectral and spatial characteristics of image | |
| CN114780904B (en) | End member self-adaptive mountain vegetation coverage remote sensing inversion method | |
| CN109359264B (en) | Chlorophyll product downscaling method and device based on MODIS | |
| CN114387531A (en) | Ground surface temperature downscaling method based on improved geographic weighted regression model | |
| CN109840539B (en) | Remote sensing space-time data fusion method based on ground block pattern spots | |
| Kong et al. | Super resolution of historic Landsat imagery using a dual generative adversarial network (GAN) model with CubeSat constellation imagery for spatially enhanced long-term vegetation monitoring | |
| CN114140357A (en) | Multi-temporal remote sensing image cloud region reconstruction method based on cooperative attention mechanism | |
| CN112285808B (en) | Method for reducing scale of APHRODITE precipitation data | |
| CN113705538A (en) | High-resolution remote sensing image road change detection device and method based on deep learning | |
| Quan et al. | Generating 60–100 m, hourly, all-weather land surface temperatures based on the Landsat, ECOSTRESS, and reanalysis temperature combination (LERC) | |
| CN115830185A (en) | Double-window simulation reconstruction method and device for short-wave infrared remote sensing image | |
| CN111523451A (en) | Method for constructing high-space-time resolution NDVI data | |
| Kim et al. | Digital elevation modeling via curvature interpolation for LiDAR data | |
| CN115272144A (en) | Spatio-temporal spectrum fusion method for hyperspectral image and multispectral image | |
| Jin et al. | Multisensor Fusion of Landsat Images for High‐Resolution Thermal Infrared Images Using Sparse Representations |
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 |