一种遥感图像云影检测方法、系统、终端及存储介质
By employing wavelet transform and heterogeneous centroid offset vector techniques, the problems of misjudgment and missed detection in cloud shadow detection in remote sensing images were solved, achieving physical separation and geometric alignment of cloud shadow features and improving detection performance.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NANJING UNIV OF INFORMATION SCI & TECH
- Filing Date
- 2026-05-13
- Publication Date
- 2026-07-17
AI Technical Summary
Existing remote sensing image cloud shadow detection methods lack modeling of the spatial correspondence between cloud shadows, which makes it easy for detection results to be misjudged and missed. Furthermore, cloud shadows are easily confused with low-reflectivity ground features, affecting the detection effect.
Wavelet transform is used to decompose optical remote sensing image data to generate low-frequency and high-frequency sub-bands. Cloud shadow feature enhancement signals are generated through a cloud shadow feature spatial frequency enhancement module. Combined with heterogeneous centroid offset vector and bidirectional texture matching score, physical separation and geometric alignment of cloud shadow features are achieved. Cloud shadow segmentation is performed using adaptive enhancement factor and geometric consistency loss constraint.
It effectively captures detailed information of cloud and shadow boundaries, alleviates the problems of shadow blurring and phase reversal, achieves physical separation of cloud and shadow features and cross-class geometric alignment, and ensures the cloud and shadow detection effect in complex scenes.
Smart Images
Figure CN122223013B_ABST