Pan-sharpening method based on multimodal texture correction and adaptive edge detail fusion

A pan-sharpening method based on multimodal texture correction and adaptive edge detail fusion, comprising following steps:

• obtaining a low-resolution multispectral (LRMS) image and a panchromatic image, fusing an upsampled LRMS image with the panchromatic image to obtain a fused image, respectively extracting intensity components of the LRMS image and the fused image, inputting the intensity components and the panchromatic image into a multimodal texture correction model, and performing optimization solution on the multimodal texture correction model through an optimization method to obtain a texture-corrected image, wherein the multimodal texture correction model is constructed based on a variational optimization model;
• and
• performing detail extraction and edge protection on the texture-corrected image to obtain first image details;
• performing detail extraction and edge protection on the upsampled LRMS image to obtain second image details;
• performing adaptive fusion on the first image details and the second image details to obtain detail information, and adding the detail information to the upsampled LRMS image to obtain a final high-resolution multispectral (HRMS) image;
• wherein the multimodal texture correction model is:
• T
• C
• =
• arg
• ⁢
• min
• T
• C
• ⁢
• 1
• 2
• ⁢
• 
• DHT
• C
• -
• I
• 0
• 
• F
• 2
• +
• α
• 2
• ⁢
• 
• ∇
• 2
• T
• C
• -
• ∇
• 2
• P
• 
• F
• 2
• +
• β
• 2
• ⁢
• 
• ∇
• 2
• (
• DHT
• C
• )
• -
• ∇
• 2
• I
• 0
• 
• F
• 2
• +
• γ
• 2
• ⁢
• 
• T
• C
• -
• I
• net
• 
• F
• 2
• +
• δ
• 2
• ⁢
• 
• ∇
• 2
• T
• C
• -
• ∇
• 2
• I
• net
• 
• F
• 2
• +
• θ
• ⁢
• 
• ∇
• 2
• T
• C
• 
• 1
• wherein TC is the texture-corrected image, D represents a downsampling matrix, H represents a degradation filter, I0 represents the intensity component of the LRMS image, α, β, γ, δ, θ represent penalty parameters corresponding to different terms, ∇2 is a Laplacian operator, P represents the panchromatic image, Inet represents the intensity component of the fused image, | |F represents an Frobenius norm, and ∥·∥1 represents a 1-norm;
• wherein the degradation filter H is obtained through an adaptive degradation filter algorithm, wherein the degradation filter H adopts a Gaussian filter HA, and the adaptive degradation filter algorithm is:
• H
• A
• =
• arg
• ⁢
• min
• H
• A
• ⁢
• 1
• 2
• ⁢
• 
• DH
• A
• ⁢
• T
• C
• -
• I
• 0
• 
• F
• 2
• wherein DHATC=DF−1(HA (u, v) F(TC));
• F(·) represents a fast Fourier transform (FFT) operation, and F−1(·) represents an inverse fast Fourier transform (IFFT) operation;
• and
• a frequency domain expression HA (u, v) of the Gaussian filter HA is:
• H
• A
• (
• u
• ,
• v
• )
• =
• e
• -
• D
• C
• 2
• (
• u
• ,
• v
• )
• 2
• ⁢
• σ
• 2
• wherein DC (u, v) represents distance from a point (u, v) to a center of the frequency domain, σ represents standard deviation, and σ obtains an optimal value according to correlation and similarity indexes, and the optimal value of σ is σbest:
• σ
• best
• =
• arg
• ⁢
• max
• σ
• ⁢
• ρ
• ⁡
• (
• DH
• A
• ⁢
• T
• C
• ,
• I
• 0
• )
• +
• S
• ⁡
• (
• DH
• A
• ⁢
• T
• C
• ,
• I
• 0
• )
• 2
• wherein ρ (DHATC, I0) is a correlation coefficient (CC) index between DHATC and I0, and S (DHATC, I0) is a structural similarity index measure (SSIM) index between the DHATC and the I0.