一种并行化边界感知云或阴影快速分割方法
By using a parallelized boundary-aware segmentation model and a fast bilateral convolution method based on Gaussian KD trees, edge segmentation of clouds/shadows in remote sensing images is optimized, solving the problems of pixel-level accuracy and computational efficiency in existing technologies, and achieving efficient edge-aware segmentation results.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA UNIV OF MINING & TECH (BEIJING)
- Filing Date
- 2025-09-29
- Publication Date
- 2026-07-17
AI Technical Summary
Existing boundary-aware semantic segmentation methods struggle to achieve pixel-level accurate cloud/shadow marking in remote sensing images, failing to generate precise edge marking results. Furthermore, existing methods require extensive manual design, making it difficult to meet the demands of massive amounts of remote sensing data.
A parallelized boundary-aware fast cloud or shadow segmentation method is designed. By constructing a boundary-aware segmentation model, utilizing domestically produced image-level weak labeling and multispectral features, and combining a parallelized fast bilateral convolution method of Gaussian KD trees, the segmentation effect of target edges is optimized to achieve edge-aware segmentation.
It achieves pixel-level region labeling accuracy and tight, sharp edge segmentation, improves the computational efficiency of the segmentation model, generates high-quality preprocessing results, and is suitable for ultra-high resolution boundary-aware image segmentation.
Smart Images
Figure CN121329997B_ABST