Nonlinear simulation method and system for high-resolution multispectral remote sensing image

Abstract

The invention provides a nonlinear simulation method and system for a high-resolution multispectral remote sensing image. On the basis of the scale invariance principle of images, the non-linear simulation method for the remote sensing images is deduced and obtained by utilizing the space detail difference and the spectrum detail difference of the images with different space scales. The method is not limited by the number of wavebands and can predict a remote sensing image simulation result. On the basis of a moderate balance target of a multispectral correlation coefficient and a panchromatic correlation coefficient, an optimal feature extraction coefficient kI is provided, a feature fusion coefficient is designed as a reciprocal of the feature extraction coefficient, namely # imgabs0 #, and meanwhile, two image simulation modes of injecting panchromatic bands into multispectral bands and injecting multispectral bands into panchromatic bands are provided; and high-fidelity simulation of remote sensing image spectrum and spatial information can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing image fusion enhancement, and more specifically, to a nonlinear simulation method and system for high-resolution multispectral remote sensing images. Background Art

[0002] Satellite remote sensing image data is diverse, and different remote sensing image data have different spatial resolutions, temporal resolutions, and spectral resolutions. Remote sensing image fusion refers to integrating the basic features of different remote sensing images into a new remote sensing image map. Commonly used methods are Brovey fusion method and CN fusion method.

[0003] For example, the patent CN112464834A adopts the Brovey fusion method, which is also called the color normalization transformation fusion method. It multiplies three multispectral bands with a high spatial resolution full-color image, simplifies the image fusion operation coefficients, and retains the spectral information of the source image to a large extent. However, it also has the defect of only being able to fuse three multispectral band images. In addition, the existing fusion method has the defect of unpredictable results in nonlinear simulation. Summary of the invention

[0004] In view of the above problems, the purpose of the present invention is to provide a nonlinear simulation method and system for high-resolution multispectral remote sensing images, which overcomes the unpredictable defects of traditional nonlinear simulation results of remote sensing images, and proposes a scientific selection method for image feature fusion parameters, which can maintain the image simulation results highly inherited from the spectral information of the original multispectral image and the spatial detail information of the original panchromatic image.

[0005] A first aspect of the present invention provides a nonlinear simulation method for high-resolution multispectral remote sensing images, the method comprising:

[0006] S1: Obtain satellite remote sensing images, which include panchromatic band images P with high spatial resolution and multispectral band images with low spatial resolution. The multispectral band images include images of n bands, denoted as M1, M2, ..., M n ;

[0007] S2: spatially register all images to make the geometrical spatial positions of the same ground objects in the panchromatic band image P and the multispectral band image consistent, and resample the multispectral band images involved in the simulation as high spatial resolution images, and the multispectral band images involved in the simulation are recorded as M;

[0008] S3: Constructing intermediate band images using low-resolution multispectral band images I:

[0009]

[0010] Wherein, l is the preset number of bands of the multispectral band image involved in constructing the intermediate band image I, l≤n;

[0011] S4: Select the panchromatic band to be injected into the multispectral band image simulation or the multispectral band to be injected into the panchromatic band image simulation, and construct the corresponding image simulation operator, specifically:

[0012] When the panchromatic band is injected into the multispectral band image simulation, a pair of product simulation operators P is constructed. M and I M for:

[0013]

[0014] Where: μ P is the mean value of the panchromatic band image P, μ I The mean value of the intermediate band image I;

[0015] When multispectral bands are injected into the panchromatic band image simulation, a pair of product simulation operators M is constructed. P and I P for:

[0016]

[0017] Among them, μ M is the mean value of the multispectral band image M, μ I The mean value of the intermediate band image I;

[0018] S5: Select the image simulation method and determine the feature extraction coefficient k based on the moderate balance between the multispectral correlation coefficient and the panchromatic correlation coefficient. I And feature fusion coefficient k E ,make According to remote sensing images, image simulation methods and characteristic parameters, high-resolution multispectral band image simulation calculations are performed, specifically:

[0019] When the panchromatic band is injected into the multispectral band image simulation:

[0020]

[0021] Among them, M f is the high-resolution multispectral band corresponding to the simulated low spatial resolution multispectral band M, μ M , σ Mf M and P respectively M ,I M , mean square error of Mf, μ M , μ Mf M and P respectively M ,IM , the mean of Mf;

[0022] When multispectral bands are injected into the panchromatic band image simulation:

[0023]

[0024] Where Mf is the high-resolution multispectral band corresponding to the simulated low spatial resolution multispectral band M, μ M , μ P , μ Mf They are M, P, and M respectively. P ,I P , mean square error of Mf, μ M , μ P , μ Mf They are M, P, and M respectively. P ,I P , the mean of Mf.

[0025] A second aspect of the present invention provides a nonlinear simulation system for high-resolution multispectral remote sensing images, comprising a memory and a processor, wherein the memory comprises a nonlinear simulation method program for high-resolution multispectral remote sensing images, and when the nonlinear simulation method program for high-resolution multispectral remote sensing images is executed by the processor, the steps of the nonlinear simulation method for high-resolution multispectral remote sensing images are implemented.

[0026] The third aspect of the present invention provides a computer-readable storage medium, which includes a nonlinear simulation method program for high-resolution multispectral remote sensing images. When the nonlinear simulation method program for high-resolution multispectral remote sensing images is executed by a processor, the steps of the nonlinear simulation method for high-resolution multispectral remote sensing images are implemented.

[0027] Compared with the prior art, the technical solution of the present invention has the following beneficial effects:

[0028] The present invention provides a nonlinear simulation method and system for high-resolution multispectral remote sensing images and a readable storage medium. Based on the scale invariance principle of images, the present invention utilizes the spatial detail differences and spectral detail differences of images of different spatial scales to derive a nonlinear simulation method for remote sensing images. The method has no limit on the number of bands and can predict the simulation results of remote sensing images. Based on the moderate balance goal of multispectral correlation coefficient and panchromatic correlation coefficient, the present invention provides an optimal feature extraction coefficient k I , the design feature fusion coefficient is the inverse of the feature extraction coefficient, that is, At the same time, this method provides two image simulation modes: injecting panchromatic band into multispectral band and injecting multispectral band into panchromatic band, which can realize high-fidelity simulation of spectral and spatial information of remote sensing images.

[0029] The visual discrimination of the simulated high-resolution multispectral image results verifies the effectiveness of the scheme. Compared with the original multispectral image, the spatial information of the geometric details, texture, edge, layer, etc. of the objects in the simulated high-resolution multispectral image is greatly enriched, and compared with the original multispectral image, the spectral characteristics and color display of the objects such as water bodies, vegetation, exposed surfaces, and buildings in the simulated multispectral image remain basically stable. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a flow chart of the nonlinear simulation method of high-resolution multispectral remote sensing images in Example 1.

[0031] Figure 2 This is a full-color band map (0.5-meter resolution).

[0032] Figure 3 This is a multispectral NRG composite image (2-meter resolution).

[0033] Figure 4 It is a multispectral RGB composite image (2-meter resolution).

[0034] Figure 5 This is an intermediate band I image (2-meter resolution).

[0035] Figure 6 This is a standard false color image synthesized by the red, green and blue channels of the near infrared band (Nf), red band (Rf) and green band (Gf) simulated in Scheme 1.

[0036] Figure 7 This is a true color image synthesized by the red band (Nf), green band (Gf) and blue band (Bf) according to the red, green and blue channels after simulation of Scheme 1.

[0037] Figure 8 This is a standard false color image synthesized by the red, green and blue channels of the near infrared band (Nf), red band (Rf) and green band (Gf) simulated in Scheme 2.

[0038] Fig. 9 This is a true color image synthesized by the red band (Nf), green band (Gf) and blue band (Bf) according to the red, green and blue channels after simulation of Scheme 2. DETAILED DESCRIPTION

[0039] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0040] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0041] Example 1

[0042] like Figure 1 As shown, this embodiment provides a nonlinear simulation method for high-resolution multispectral remote sensing images, the method comprising:

[0043] S1: Obtain satellite remote sensing images, which include panchromatic band images P with high spatial resolution and multispectral band images with low spatial resolution. The multispectral band images include images of n bands, denoted as M1, M2, ..., M n ;

[0044] S2: spatially register all images to make the geometrical spatial positions of the same ground objects in the panchromatic band image P and the multispectral band image consistent, and resample the multispectral band images involved in the simulation as high spatial resolution images, and the multispectral band images involved in the simulation are recorded as M;

[0045] Among them, all images are spatially registered, including:

[0046] Preprocessing: Correct the original image, including radiation correction and atmospheric correction, to eliminate the influence of sensor characteristics and atmospheric conditions on the image;

[0047] Select control points: determine a set of control points from the panchromatic band image P;

[0048] Feature point extraction: Extract feature points from multispectral images and match them with control points;

[0049] Image transformation: Determine a transformation model and transform all images by the transformation model to achieve spatial alignment.

[0050] Among them, the multispectral band images involved in the simulation are resampled as high spatial resolution images, specifically:

[0051] S201: Assume that the resolution of the high spatial resolution panchromatic image is p meters × p meters, and the resolution of the low spatial resolution multispectral band image is m meters × m meters, and determine the upsampling factor s:

[0052]

[0053] S202: insert pixels into the multispectral band image so that the horizontal pixels and vertical pixels of the multispectral band image are increased to s times.

[0054] In step S202, the nearest neighbor interpolation method can be used to interpolate pixels in the multispectral band image, that is, the nearest pixel value is assigned to the new pixel. This method is simple to calculate and fast, but may cause discontinuous brightness of the image, and is suitable for scenes with low accuracy requirements.

[0055] In step S202, the bicubic convolution interpolation method can also be used to interpolate pixels in the multispectral band image, that is, 16 pixel values ​​around the interpolation point are used to interpolate using a cubic convolution function. This method has little high-frequency information loss and enhances the edge, and is suitable for scenes with high precision requirements.

[0056] S3: Constructing intermediate band images using low-resolution multispectral band images I:

[0057]

[0058] Wherein, l is the preset number of bands of the multispectral band image involved in constructing the intermediate band image I, l≤n;

[0059] S4: Select the panchromatic band to be injected into the multispectral band image simulation or the multispectral band to be injected into the panchromatic band image simulation, and construct the corresponding image simulation operator, specifically:

[0060] When the panchromatic band is injected into the multispectral band image simulation, a pair of product simulation operators P is constructed. M and I M for:

[0061]

[0062] Where: μ P is the mean value of the panchromatic band image P, μ I The mean value of the intermediate band image I;

[0063] When multispectral bands are injected into the panchromatic band image simulation, a pair of product simulation operators M is constructed. P and I P for:

[0064]

[0065] Among them, μ M is the mean value of the multispectral band image M, μ I The mean value of the intermediate band image I;

[0066] S5: Select the image simulation method and determine the feature extraction coefficient k based on the moderate balance between the multispectral correlation coefficient and the panchromatic correlation coefficient. I And feature fusion coefficient k E , high-resolution multispectral band image simulation calculations are performed based on remote sensing images, image simulation methods and characteristic parameters, specifically:

[0067] When the panchromatic band is injected into the multispectral band image simulation:

[0068]

[0069] Where:

[0070]

[0071] Where r() represents the correlation coefficient, Mf is the high-resolution multispectral band corresponding to the simulated low spatial resolution multispectral band M, and σ M , σ Mf M and P respectively M ,I M 、M f The mean square error, μ M , μ Mf M and P respectively M ,I M , the mean of Mf;

[0072] When multispectral bands are injected into the panchromatic band image simulation:

[0073]

[0074] Where:

[0075]

[0076] Where r( ) represents the correlation coefficient, Mf is the high-resolution multispectral band corresponding to the simulated low spatial resolution multispectral band M, σ M , σ P , σ Mf They are M, P, and M respectively. P ,I P , mean square error of Mf, μ M , μ P , μ Mf They are M, P, and M respectively. P ,I P , the mean of Mf.

[0077] The principle of the nonlinear simulation method of high-resolution multispectral remote sensing images of the present invention is explained below:

[0078] 1. Scale invariant transformation of the product image of the mean filtered image and the ratio image

[0079] Generally speaking, satellite remote sensing images have multiple low-resolution multispectral bands M1, M2, ..., M n And a high-resolution panchromatic band P. It is assumed that corresponding to the high-resolution panchromatic band P there is a low-resolution panchromatic band I (intermediate band) corresponding to the spectral band, and corresponding to the low-resolution multispectral band M there is a high-resolution multispectral band Mf corresponding to the spectral band.

[0080] Let μ M , μ P , μ I , μ Mf are the average values ​​of low-resolution multispectral band, high-resolution panchromatic band, low-resolution panchromatic band, and high-resolution multispectral image, respectively. M , σ P , σ I , σ Mf are the mean square error of low-resolution multispectral band, high-resolution panchromatic band, low-resolution panchromatic band, and high-resolution multispectral image, respectively. r(P,I) is the correlation coefficient between P and I, and r(Mf,M) is the correlation coefficient between Mf and M.

[0081] Solution 1: Panchromatic band injection into multispectral band nonlinear simulation mode

[0082] (1) The full-color product image is proportional to k1

[0083] make:

[0084]

[0085] Inferred from the scale invariance of remote sensing images: (P-μ P )=k1(I-μ I )

[0086] We can get:

[0087] P M =k1I M

[0088] That is, under scale invariance, the product image of the panchromatic mean filtered image and the image of the same dimension is also proportional, and the correlation coefficients of the two product images and any image of the same dimension are equal.

[0089] make:

[0090] E1=P M -k1I M =0

[0091] For any non-zero vector X1 (X1 has the same dimension as P), we have:

[0092] r(E1,X1)=0

[0093] And then:

[0094]

[0095] (2) The multi-spectral mean filter image is proportional to k2

[0096] make:

[0097] E2=(Mf-μ Mf )-k2(M-μ M )=0

[0098] E2 is independent of any non-zero vector X2 (X2 has the same dimension as P), then:

[0099] r(E2,X2)=0

[0100]

[0101] (3) Scale invariance inference: the relationship between multispectral images and panchromatic product images

[0102] Depend on:

[0103]

[0104] We can get:

[0105]

[0106] Depend on:

[0107]

[0108] We can get:

[0109]

[0110] Therefore:

[0111]

[0112] m1, m2 are any real numbers greater than 0. X1, X2 are any remote sensing images with the same dimensions as M, P, I.

[0113] (4) Nonlinear simulation method of high-resolution multispectral images

[0114] From the above formula, we can get:

[0115]

[0116] Then we can get:

[0117]

[0118] make:

[0119]

[0120] but:

[0121]

[0122] Matching the above results to the multispectral band M, the general scheme of nonlinear simulation of full-color injection multispectral mode is:

[0123]

[0124] in,

[0125]

[0126] They are image simulation operators P M ,I M The mean of They are image simulation operators P M ,I M The mean square error of k I is the feature extraction coefficient, k E is the feature fusion coefficient, k I , k E is a constant greater than zero.

[0127] In this scheme, let X1∈{M,P,I}, aiming at a moderate balance between the multispectral correlation coefficient and the panchromatic correlation coefficient, and generally the feature fusion coefficient should be the inverse of the feature extraction coefficient. At this time, the feature extraction coefficient is k I , the feature fusion coefficient is Right now:

[0128]

[0129] Solution 2: Multispectral band injection into panchromatic band nonlinear simulation mode (1) Panchromatic product image is proportional to k1

[0130] Inferred from the scale invariance of remote sensing images: (P-μ P )=k1(I-μ I )

[0131] We can get:

[0132] P=k1I+μ P -k1μ I

[0133] To make P and I proportional, that is:

[0134] P=k1I

[0135]

[0136] Must have:

[0137]

[0138] Therefore:

[0139]

[0140] (2) The multi-spectral mean filter image is proportional to k2

[0141] When the mean filter image is proportional, we generally have:

[0142]

[0143] (3) Scale invariance inference: The relationship between the high-resolution image and the low-resolution product image has been deduced before:

[0144]

[0145] make:

[0146]

[0147] m1, m2 are any real numbers greater than 0. X1, X2 are any remote sensing images with the same dimensions as M, P, I.

[0148] (4) Nonlinear simulation method of high-resolution multispectral images

[0149] From the above formula, we can get:

[0150]

[0151] make

[0152]

[0153] but:

[0154]

[0155] make:

[0156]

[0157] Matching the above results to the multispectral band M, the general scheme of nonlinear simulation of multispectral injection into full-color mode is:

[0158]

[0159] Among them, X1 and X2 are images of the same dimension as M, P, and I, and m1 and m2 are arbitrary real numbers greater than zero.

[0160] They are image simulation operators M P ,I P The mean of They are image simulation operators M P ,I P The mean square error of k I is the feature extraction coefficient, k E is the feature fusion coefficient.

[0161] make:

[0162]

[0163] Let E spe It has nothing to do with P, M, and I, that is:

[0164] r(E spe ,P)=0

[0165] r(E spe ,M)=0

[0166] r(E spe ,I)=0

[0167] Can be obtained:

[0168]

[0169] Likewise, in this scheme, the goal is to achieve a moderate balance between the multispectral correlation coefficient and the panchromatic correlation coefficient, and generally the feature fusion coefficient should be the inverse of the feature extraction coefficient.

[0170]

[0171] This method is suitable for simulating high-resolution multispectral images using high-resolution panchromatic bands and low-resolution multispectral bands of remote sensing images to enhance the spatial resolution and geometric texture information of multispectral images while truly maintaining the original spectral information of the objects.

[0172] Example 2

[0173] In order to verify the effect of nonlinear simulation of high-resolution remote sensing images, this embodiment is mainly implemented using ENVI remote sensing image processing software, and is further described with a satellite remote sensing image with a full-color band (P), a blue band (B), a green band (G), a red band (R), and a near-infrared (N).

[0174] 1. Input remote sensing image.

[0175] Open a WV-02 remote sensing image with panchromatic band (P), blue band (B), green band (G), red band (R), and near infrared (N). Figure 2 , Figure 3 , Figure 4 They are respectively full-color band image (0.5-meter resolution, that is, the ground area represented by one pixel is 0.5 meters × 0.5 meters), multispectral NRG synthetic image (2-meter resolution, that is, the ground area represented by one pixel is 2 meters × 2 meters), and multispectral RGB synthetic image (2-meter resolution) (the effect image of 0.1% stretching according to the default setting of envi).

[0176] 2. Select the multispectral blue band (B), green band (G), red band (R), and near infrared (N) to construct the intermediate band I. The calculation formula is: b=fix((1.0*b1+b2+b3+b4) / 4), b1, b2, b3, and b4 are the B, G, R, and N band images respectively. Figure 5 This is an intermediate band I image (2-meter resolution).

[0177] Solution 1: Panchromatic injection multi-spectral band nonlinear simulation

[0178] 3. Constructing image simulation operators

[0179] Taking the nonlinear simulation mode of injecting panchromatic band into multispectral band as an example, the image simulation operator P is constructed. M and I M Specifically, the simulation operator P M The calculation method is: long((b1-373.051946)*1.0*b3 / b2); simulation operator I M The calculation method is: long((b2-383.914786)*1.0*b3 / b2); where b1, b2, and b3 are the panchromatic band P, intermediate band I, and multispectral band M, respectively.

[0180] 4. Use ENVI software to convert the panchromatic band P, multispectral band M, and fusion operator P M and I M Resample the high spatial resolution image to synthesize an image file, and then calculate the image characteristic statistical parameters such as the mean μ and standard deviation σ of each band image. The basic characteristic statistical parameters of each band image are shown in Table 1 and Table 2.

[0181] Table 1 Statistics of mean and mean square error of each band of image

[0182]

[0183]

[0184] Table 2 Statistics of correlation coefficients of each band of images

[0185]

[0186] 5. Multi-spectral band nonlinear simulation calculation

[0187] In this case, the characteristic simulation parameters Take 4 bands of multi-spectral nonlinear simulation as an example, and the specific simulation algorithm is as follows:

[0188] (1) The simulation operation expression corresponding to the blue band is:

[0189] fix(96.636771 / 100.309513*((b3-301.598439)+1.008369*(0.675532*(b1-(-18.950312))-0.9917*0.687211*(b2-(-20.698493))))+301.598439), where b1 is the simulation factor P B , b2 is the simulation factor I B , b3 is the blue band B, and the simulated blue band (Bf) image is obtained by calculation.

[0190] (2) The simulation operation expression corresponding to the green band is:

[0191] fix(163.499142 / 170.241165*((b3-409.707032)+1.018724*(0.839874*(b1-(-11.842505))-0.98162*0.86091*(b2-(-13.92145))))+409.707032), where b1 is the simulation factor P G , b2 is the simulation factor I G , b3 is the green band G, and the simulated green band (Gf) image is obtained by calculation.

[0192] (3) The simulation operation expression corresponding to the red band is:

[0193] fix(199.762851 / 209.089349*((b3-361.486579)+1.032462*(1.13971*(b1-(12.199673))-0.968559*1.186003*(b2-(10.243729))))+361.486579), where b1 is the simulation factor P R , b2 is the simulation factor I R, b3 is the red band R, and the simulated red band (Rf) image is obtained by calculation.

[0194] (4) The simulation operation expression corresponding to the near-infrared band is fix(263.263849 / 285.950104*((b3-464.362264)+1.181088*(1.386487*(b1-(18.904132))-0.846677*1.490177*(b2-(24.5701))))+464.362264), where b1 is the simulation factor P N , b2 is the simulation factor I N , b3 is the near-infrared band N, and the simulated near-infrared band (Nf) image is obtained by calculation.

[0195] The simulated near infrared band (Nf), red band (Rf) and green band (Gf) are synthesized into a standard false color image according to the red, green and blue channels. Figure 6 The simulated red band (Nf), green band (Gf) and blue band (Bf) are synthesized into a true color image according to the red, green and blue channels. Figure 7 .

[0196] Solution 2: Multi-spectral injection panchromatic band nonlinear simulation

[0197] 3. Constructing image simulation operators

[0198] Taking the panchromatic band injection multispectral band image simulation mode as an example, the image simulation operator M is constructed. P and I P Specifically, the simulation operator B P The calculation formula is: long((b3-301.598439)*(1.0*b1 / b2)); G P The calculation formula is: long((b3-409.707032)*(1.0*b1 / b2)); R P The calculation formula is: long((b3-361.486579)*(1.0*b1 / b2)); N P The calculation formula is: long((b3-464.362264)*(1.0*b1 / b2)); simulation operator I P The calculation formula is: long((b2-383.914786)*(1.0*b1 / b2)); where b1, b2, and b3 are the panchromatic band P, intermediate band I, and multispectral band M, respectively.

[0199] 4. Use ENVI software to convert the panchromatic band P, multispectral band M, and fusion operator M P and I PResample the high spatial resolution image to synthesize an image file, and then calculate the image characteristic statistical parameters such as the mean μ and standard deviation σ of each band image. The basic characteristic statistical parameters of each band image are shown in Table 3 and Table 4.

[0200] Table 3 Statistics of mean and mean square error of each band of image

[0201] Image bands Mean / μ Mean square error / σ Panchromatic Band P 373.05195 173.10819 Intermediate Band I 383.91479 166.55612 Blue Band B 301.59844 96.636771 Green Band G 409.70703 163.49914 Red Band R 361.48658 199.76285 Near infrared 464.36226 263.26385 <![CDATA[Fusion factor B P > 1.354793 95.155261 <![CDATA[Fusion factor G P > 1.913086 161.48699 <![CDATA[Fusion factor R P > 2.422845 196.83802 <![CDATA[Fusion factor N P > -4.914095 257.04172 <![CDATA[Fusion factor I P > 0.19573 165.4889

[0202] Table 4 Statistics of correlation coefficients of each band of images

[0203]

[0204] 5. Multi-spectral band simulation calculation

[0205] In this case, the characteristic simulation parameters Take 4 bands of multi-spectral nonlinear simulation as an example, and the specific simulation algorithm is as follows:

[0206] (1) The simulation operation expression corresponding to the blue band is:

[0207] fix(96.636771 / 182.819211*((b1-373.051946)+0.971622*(1.819218*(b3-(1.354793))-0.964747*1.046041*(b2-(0.19573))))+301.598439), where b1 is the panchromatic band P and b2 is the simulation factor I P , b3 is the simulation factor B P , calculate and obtain the simulated blue band (Bf) image.

[0208] (2) The simulation operation expression corresponding to the green band is:

[0209] fix(163.499142 / 178.762468*((b1-373.051946)+0.946118*(1.071964*(b3-(1.913086))-0.985545*1.046041*(b2-(0.19573))))+409.707032), where b1 is the panchromatic band P and b2 is the simulation factor I P , b3 is the simulation factor G P , calculate and obtain the simulated green band (Gf) image.

[0210] (3) The simulation operation expression corresponding to the red band is:

[0211] fix(199.762851 / 178.353002*((b1-373.051946)+0.944052*(0.879445*(b3-(2.422845))-0.990107*1.046041*(b2-(0.19573))))+361.486579), where b1 is the panchromatic band P and b2 is the simulation factor I P , b3 is the simulation factor R P , calculate and obtain the simulated red band (Rf) image.

[0212] (4) The simulation operation expression corresponding to the near-infrared band is fix(263.263849 / 192.152052*((b1-373.051946)+1.02921*(0.673463*(b3-(-4.914095))-0.840993*1.046041*(b2-(0.19573))))+464.362264), where b1 is the panchromatic band P and b2 is the simulation factor I P , b3 is the simulation factor N P , calculate and obtain the simulated near-infrared band (Nf) image.

[0213] The simulated near infrared band (Nf), red band (Rf) and green band (Gf) are synthesized into a standard false color image according to the red, green and blue channels. Figure 8 The simulated red band (Nf), green band (Gf) and blue band (Bf) are synthesized into a true color image according to the red, green and blue channels. Fig. 9 .

[0214] In this case, four remote sensing images, namely, the original WV-02 multispectral image, the panchromatic injection multispectral simulated image by the method of the present invention, the multispectral injection panchromatic simulated image by the method of the present invention, and the Gram-Schmidit method fused image, were subjected to band data statistical analysis. The comparison of the image band statistical characteristic parameters is shown in Table 5. As can be seen from the data in the table, compared with the original multispectral image, the spatial resolution of the simulated image by the method of the present invention is increased from the original 2 meters to 0.5 meters, and the image spatial accuracy is greatly improved. The simulated image has both the original multispectral band and the panchromatic band. Compared with the GS fused image, the panchromatic injection multispectral simulated image (Scheme 1) of this method has better information consistency and slightly reduced information entropy, but the gradient information of the band is enhanced, and the spatial information of the ground objects in the fused multispectral image is richer. Compared with GS fused images, the multispectral bands of the multispectral injected into the panchromatic fused image (Scheme 2) in this method have better information consistency with the original image. While the information entropy is slightly improved, the image gradient information is basically consistent with the GS fused image. This method simulates the spectral information of the objects in the multispectral image more realistically, which is conducive to the development of various image applications such as object recognition, interpretation, and analysis.

[0215] The high-resolution multispectral image nonlinear simulation method, system and readable storage medium described in the present invention are mainly aimed at remote sensing images with panchromatic bands and multispectral bands such as near-infrared, red, green and blue bands. Firstly, based on the scale invariance assumption of images, the method uses the spatial detail differences and spectral detail differences of images of different spatial scales to construct a nonlinear simulation method for high-resolution multispectral images; secondly, in view of the defect that traditional nonlinear image simulation methods are difficult to predict the fusion results, a pair of image simulation operators are constructed to achieve predictability of image simulation results. Based on the moderate balance goal of multispectral correlation coefficient and panchromatic correlation coefficient, the present invention provides an optimal feature extraction coefficient k I , the design feature fusion coefficient is the inverse of the feature extraction coefficient, that is, Finally, two nonlinear simulation modes, full-color injection into multi-spectrum and multi-spectrum injection into full-color, are provided.

[0216] This method improves the spectral preservation ability of nonlinear image simulation while maintaining the spatial details of the full-color image. The fused image has bright colors, rich information, stable spectral information, and is easy to visually and automatically classify. Especially in the context of the rapid development of high-resolution satellite remote sensing, it has a huge role in promoting the promotion and application of domestic high-resolution images in various industries at home and abroad.

[0217] Table 5 Comparison of band statistical characteristic parameters of original multispectral images, simulated images of this method, and GS fusion images

[0218]

[0219] Example 3

[0220] The present embodiment provides a nonlinear simulation system for high-resolution multispectral remote sensing images, including a memory and a processor. The memory includes a nonlinear simulation method program for high-resolution multispectral remote sensing images. When the nonlinear simulation method program for high-resolution multispectral remote sensing images is executed by the processor, the steps of the nonlinear simulation method for high-resolution multispectral remote sensing images of the embodiment are implemented.

[0221] Example 4

[0222] The present embodiment provides a computer-readable storage medium, which includes a nonlinear simulation method program for high-resolution multispectral remote sensing images. When the nonlinear simulation method program for high-resolution multispectral remote sensing images is executed by a processor, the steps of the nonlinear simulation method for high-resolution multispectral remote sensing images described in Example 1 are implemented.

[0223] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0224] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0225] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0226] Those skilled in the art can understand that: all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiments; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), disks or optical disks, and other media that can store program codes.

[0227] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention can be essentially or partly reflected in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

Claims

1. A nonlinear simulation method for high-resolution multispectral remote sensing images, characterized in that: The method comprises: S1: Obtain satellite remote sensing images, which include panchromatic band images P with high spatial resolution and multispectral band images with low spatial resolution. The multispectral band images include images of n bands, denoted as M1, M2, ..., M n ; S2: spatially register all images to make the geometrical spatial positions of the same ground objects in the panchromatic band image P and the multispectral band image consistent, and resample the multispectral band images involved in the simulation as high spatial resolution images, and the multispectral band images involved in the simulation are recorded as M; S3: Constructing intermediate band images using low-resolution multispectral band images I: Wherein, l is the preset number of bands of the multispectral band image involved in constructing the intermediate band image I, l≤n; S4: Select the panchromatic band to be injected into the multispectral band image simulation or the multispectral band to be injected into the panchromatic band image simulation, and construct the corresponding image simulation operator, specifically: When the panchromatic band is injected into the multispectral band image simulation, a pair of product simulation operators P is constructed. M and IM 为 : Where: μ P is the mean value of the panchromatic band image P, μ I The mean value of the intermediate band image I; When multispectral bands are injected into the panchromatic band image simulation, a pair of product simulation operators M is constructed. P and I P for: Among them, μ M is the mean value of the multispectral band image M, μ I The mean value of the intermediate band image I; S5: Select the image simulation method and determine the feature extraction coefficient k based on the moderate balance between the multispectral correlation coefficient and the panchromatic correlation coefficient. I And feature fusion coefficient k E ,make According to remote sensing images, image simulation methods and characteristic parameters, high-resolution multispectral band image simulation calculations are performed, specifically: When the panchromatic band is injected into the multispectral band image simulation: Among them, M f is the high-resolution multispectral band corresponding to the simulated low spatial resolution multispectral band M, σ M , σ Mf M and P respectively M ,I M 、M f The mean square error, μ M , μ Mf M and P respectively M ,I M 、M f The mean of When multispectral bands are injected into the panchromatic band image simulation: Where Mf is the high-resolution multispectral band corresponding to the spectral band of the simulated low spatial resolution multispectral band M, σ M , σ P , σ Mf They are M, P, and M respectively. P ,I P 、M f The mean square error, μ M , μ P , μ Mf They are M, P, and M respectively. P ,I P 、M f The mean of .

2. The nonlinear simulation method of high-resolution multispectral remote sensing images according to claim 1, characterized in that: In step S2, all images are spatially registered, specifically including: Preprocessing: Correct the original image, including radiation correction and atmospheric correction, to eliminate the influence of sensor characteristics and atmospheric conditions on the image; Select control points: determine a set of control points from the high spatial resolution panchromatic band image P; Feature point extraction: Extract feature points from each image of low spatial resolution multispectral images and match them with control points; Image transformation: Determine a transformation model and transform all images by the transformation model to achieve spatial alignment.

3. The nonlinear simulation method of high-resolution multispectral remote sensing images according to claim 1, characterized in that: In step S2, the multispectral band images involved in the simulation are resampled as high spatial resolution images, specifically: S201: Assume that the resolution of the high spatial resolution image is p meters × p meters, and the resolution of the low resolution multispectral band image is m meters × m meters, and determine the upsampling factor s: S202: insert pixels into the multispectral band image so that the horizontal pixels and vertical pixels of the multispectral band image are increased to s times.

4. The nonlinear simulation method of high-resolution multispectral remote sensing images according to claim 3, characterized in that: In step S202, the nearest neighbor interpolation method is used to interpolate pixels in the multispectral band image, that is, the nearest pixel value is assigned to the new pixel.

5. The nonlinear simulation method of high-resolution multispectral remote sensing images according to claim 3, characterized in that: In step S202, bicubic convolution interpolation is used to interpolate pixels in the multispectral band image, that is, 16 pixel values ​​around the interpolation point are used to interpolate using a cubic convolution function.

6. The nonlinear simulation method of high-resolution multispectral remote sensing images according to claim 1, characterized in that: In step S3, the number of bands of the multispectral band image involved in constructing the intermediate band image I is l≤n, where n is the total number of bands of the multispectral image.

7. The nonlinear simulation method of high-resolution multispectral remote sensing images according to claim 1, characterized in that: In step S4, when the panchromatic band is injected into the multispectral band image simulation, k I The value of is: Where r() represents the correlation coefficient.

8. The nonlinear simulation method of high-resolution multispectral remote sensing images according to claim 1, characterized in that: In step S4, when the multispectral band is injected into the panchromatic band image simulation, k I The value of is: Where r() represents the correlation coefficient.

9. A nonlinear simulation system for high-resolution multispectral remote sensing images, characterized in that: It comprises a memory and a processor, wherein the memory comprises a nonlinear simulation method program for high-resolution multispectral remote sensing images, and when the nonlinear simulation method program for high-resolution multispectral remote sensing images is executed by the processor, the steps of the nonlinear simulation method for high-resolution multispectral remote sensing images as described in any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a nonlinear simulation method program for high-resolution multispectral remote sensing images. When the nonlinear simulation method program for high-resolution multispectral remote sensing images is executed by a processor, the steps of the nonlinear simulation method for high-resolution multispectral remote sensing images as described in any one of claims 1 to 8 are implemented.

Images