一种基于光流修复的半监督航拍视频去厚云方法
By employing semi-supervised segmentation and optical flow inpainting, the problem of insufficient utilization of spatial and temporal correlations in aerial video declouding was solved, enabling automated declouding processing of large-scale aerial images. The output declouding results are stable, smooth, and reliable.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINESE PEOPLES LIBERATION ARMY ARMY ARTILLERY & AIR DEFENSE ACAD
- Filing Date
- 2022-08-23
- Publication Date
- 2026-07-17
AI Technical Summary
Existing aerial video declouding methods fail to adequately utilize the spatial and temporal correlations of the video, resulting in unstable, unsmooth, and unreliable declouding results, making them particularly impractical for large-scale aerial image processing.
A semi-supervised method for removing thick clouds from aerial videos based on optical flow inpainting is adopted. By semi-supervised cloud region segmentation, optical flow estimation and optical flow completion, combined with single-frame image inpainting technology, the automatic cloud removal processing of video frames is achieved.
It achieves automated cloud removal processing of large-scale aerial images, saving training time, making full use of the spatial and temporal information of video frames, and outputting stable, smooth and highly reliable cloud-removed images.
Smart Images
Figure CN115908770B_ABST