稳健性(进化)
光学
相干衍射成像
衍射
RGB颜色模型
相位恢复
计算机科学
残余物
迭代重建
物理
解耦(概率)
人工智能
激光器
计算机视觉
迭代法
波长
振幅
探测器
重建算法
衍射光栅
数据采集
算法
作者
Zhipeng Li,Suhas P. Veetil,Aihui Sun,Xiaoliang He,Yan Kong,Cheng Liu,Jianqiang Zhu,Zhilong Jiang
标识
DOI:10.1088/2040-8986/ae9f70
摘要
Abstract Coherent diffractive imaging (CDI) reconstructs object complex amplitudes from diffraction patterns using iterative phase retrieval algorithms, while multi-wavelength CDI (MW-CDI) improves reconstruction robustness through wavelength diversity. However, practical MW-CDI systems remain affected by probe-object coupling, non-uniform illumination, and sequential acquisition instability, leading to probe-related reconstruction artifacts. To address these limitations, a single-shot RGB MW-CDI framework combined with a gradient-based probe-decoupling reconstruction strategy is proposed. An RGB laser source and a color CCD camera are employed to simultaneously acquire diffraction patterns at three wavelengths, thereby reducing temporal instability associated with sequential acquisition. An illumination-weighted gradient update strategy is further incorporated into the iterative framework to suppress residual probe-related artifacts caused by non-uniform illumination. Numerical simulations and experimental results demonstrate improved reconstruction fidelity, faster convergence, and enhanced robustness compared with conventional MW-CDI reconstruction. Experimental imaging of a USAF1951 target and weakly absorbing biological samples further verifies the effectiveness of the proposed method for stable and high-precision multi-wavelength lensless imaging.
科研通智能强力驱动
Strongly Powered by AbleSci AI