A Hyperspectral Image Restoration Method and System Based on Dual-Enhanced Low-Rank Tensor Constraints
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHONGQING UNIV
- Filing Date
- 2024-06-25
- Publication Date
- 2026-05-26
AI Technical Summary
Existing low-rank tensor constraint methods fail to effectively combine global and non-local priors in the restoration of noisy and missing hyperspectral remote sensing images, resulting in insufficient generalization ability of the restoration model, loss of complementary information between spatial and spectral modes, and easy generation of residual noise.
A dual-enhanced low-rank tensor constraint method is adopted. Through a global-guided nonlocal learning framework, combined with global and nonlocal priors, and utilizing enhanced global and nonlocal low-rank tensor approximations, a stable tensor quantum space and paired modal complementary fusion are constructed to suppress residual noise and improve image restoration accuracy.
It improves the restoration accuracy and generalization ability of hyperspectral remote sensing images, stably characterizes and propagates the inherent structure of images, and reduces the loss of deep information and residual noise, which is superior to existing methods.
Smart Images

Figure CN118710551B_ABST