A compressed sensing-based pan-sharpening using joint data fidelity and blind blurring kernel estimation

Jiang, Yiyong; Chen, Liqin; Wang, Wei; Ding, Xinghao; Huang, Yue

doi:10.1109/icip.2014.7026021

Cited by 6 publications

(2 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We denote k as a blurring kernel. An averaging kernel [19], [29], [48] is widely used for k, since its experimental results are comparable with other results using estimated kernels [20], [21]. The likelihood p(y|x) can be defined as…”

Section: Spectral Preservation Likelihood P(y|x)mentioning

confidence: 88%

“…Prior p(x): Natural image gradients contain edge-based structures and follow a heavy-tailed distribution learnt from generic image databases [24], [25]. We assume that image gradients are piecewise continuous, and the first-order gradient distribution of the HRMS image x is defined by a Laplacian distribution [21], [26] with location zero and scale s 1 , and the second-order gradient distribution of x obeys a Laplacian distribution with…”

Section: Spectral Preservation Likelihood P(y|x)mentioning

confidence: 99%

See 1 more Smart Citation

Bayesian Pan-Sharpening With Multiorder Gradient-Based Deep Network Constraints

Guo

Zhuang

Guo

2020

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

Pan-sharpening aims at acquiring a multispectral image with a high spatial resolution by fusing a low-resolution multispectral image and a panchromatic image. In order to improve spatial details and reduce spectral distortions, we develop a new pan-sharpening model based on the Bayesian theory, which involves three assumptions: 1) the low-resolution multispectral images are generally decimated from the high-resolution multispectral images by convolution with a blurring kernel; 2) different from most pan-sharpening methods that use linear manners to preserve spatial information, we attempt a nonlinear manner based on a convolutional neural network composed of the proposed multiscale recursive blocks, and we train our network parameters in multiorder gradient domains to preserve more spatial structures; and 3) we introduce an anisotropic total variation prior in multiorder gradient domains to reconstruct better image edges and details. We establish the posterior probability model based on the above assumptions and derive an efficient optimization scheme to address the proposed objective function. Final experimental results demonstrate that the proposed model can overcome the restriction of a linear model and achieve better spectral and spatial fusion, compared with several traditional and deep-learning-based pan-sharpening approaches. In addition, our model achieves more promising generalization across different satellites than other deep-learning-based methods.

show abstract