一种基于非线性风格插值的双时相语义变化检测方法
The dual-temporal semantic change detection method using nonlinear style interpolation solves the problems of temporal information sparsity, style-structure coupling, and pseudo-change noise interference in existing technologies, and achieves high-precision detection and robustness improvement of the continuous evolution process of the Earth's surface.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA UNIV OF PETROLEUM (EAST CHINA)
- Filing Date
- 2026-05-18
- Publication Date
- 2026-07-17
AI Technical Summary
Existing semantic change detection methods have shortcomings in terms of temporal information sparsity, style and structure coupling, and spurious change noise interference. They cannot effectively model the continuous evolution process of the land surface, resulting in insufficient detection accuracy and robustness.
A bi-temporal semantic change detection method based on nonlinear style interpolation is adopted. Through latent style statistical modeling and decoupling, nonlinear time mapping, bidirectional pseudo-temporal sequence synthesis, temporal offset module and temporal gating fusion unit, the bi-temporal phase is reconstructed to be a relatively continuous pseudo-temporal evolution model, explicitly decoupling radiation inconsistency, and enhancing the perception of temporal context information and noise suppression capabilities.
It significantly improves the accuracy and robustness of change detection, effectively distinguishes between real and false changes, reduces false alarm and false negative rates, and adapts to complex environmental changes.
Smart Images
Figure CN122200658B_ABST