融合多尺度多方向模糊特征的自适应门控去模糊系统及方法
By designing an adaptive gating deblurring system that integrates multi-scale and multi-directional blurring features, the problems of large computational resources, inaccurate detail generation, and loss of edge information in existing image deblurring methods are solved, achieving efficient and accurate image deblurring results.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- XIAN INST OF OPTICS & PRECISION MECHANICS CHINESE ACAD OF SCI
- Filing Date
- 2026-04-15
- Publication Date
- 2026-07-17
AI Technical Summary
Existing image deblurring methods based on GAN networks have lengthy training processes, long inference times, large computational resources, and inaccurate detail generation. Methods based on multi-scale blur information combination are prone to losing image edge information, while image block processing methods suffer from insufficient contextual information interaction.
Design an adaptive gated deblurring system that integrates multi-scale and multi-directional fuzzy features, including a multi-scale and multi-directional fuzzy feature extraction module, a local fuzzy description and confidence gating map acquisition module, a residual extraction module, and an adaptive gated fusion module. Through multi-directional and multi-scale feature enhancement, local fuzzy description estimation, and confidence gating map acquisition, combined with residual extraction, adaptive gated fusion is performed to generate a deblurred image.
It achieves simple computation, strong orientation awareness, and accurate details in the generated deblurred image, reduces computational load, avoids sharpening and artifacts, and improves the deblurring effect.
Smart Images
Figure CN122048719B_ABST