快照(计算机存储)
帧速率
显微镜
融合
人工智能
纳米-
计算机视觉
高分辨率
材料科学
光学
计算机科学
物理
遥感
地质学
哲学
操作系统
复合材料
语言学
作者
Xiangyu Zhao,Mingyu Li,Jian Wang,Shiyuan Liu,Zhengqiong Dong,Lei Nie,Hui Deng,Jinlong Zhu
出处
期刊:Optica
[Optica Publishing Group]
日期:2025-08-07
卷期号:12 (9): 1469-1469
标识
DOI:10.1364/optica.573166
摘要
High-speed and high-resolution optical imaging is critical in diverse fields such as defect detection, biological sample observation, and laser processing monitoring. However, when capturing rapidly evolving dynamic structures, conventional cameras face limitations in frame rates, leading to motion blur and degradation in resolution. To address these challenges, we propose a deep feature-fusion snapshot compression microscopy (DFF-SCM) for high-speed, high-resolution sub-pixel imaging. The system utilizes snapshot compression to reconstruct multi-frame images from a single-exposure frame, substantially mitigating motion blur. Structural complementary masks are incorporated to enhance demodulation stability and image fidelity. Furthermore, a deep feature-fusion network is employed to reduce noise and to improve feature contrast. Experimental results demonstrated that DFF-SCM could achieve sub-pixel accuracy at 1000 fps for the goal of surface inspection, such as silicon wafer scratch detection and spot identification on a mirror.
科研通智能强力驱动
Strongly Powered by AbleSci AI