摘要
自动光学检测(automated optical inspection,AOI)相机运动速度过高,拍摄印制电路板(printed circuit board,PCB)图像会产生严重模糊.如能正确恢复因此而产生的退化,可提高相机的运动速度,进而提高AOI的检测效率.受代数迭代重建算法(algebraic reconstruction technique,ART)启发,提出一种基于光线重排的图像非盲去模糊算法.该方法在模糊核已知的情况下,基于快速迭代收缩阈值算法(fast iterative shrinkage thresholding algorithm,FISTA)迭代重建算法,运用Nestrov加速和光线重排提高了迭代收敛速度,运用数据正则抑制图像的各类加性噪声和伪影,较好地重建出原本的清晰图像.结果表明,该方法对噪声具有较好的抑制能力,对AOI相机运动曝光引起的严重模糊,较常用的传统逆滤波方法具有更好的图像恢复效果.
Abstract
The image of the printed circuit board(PCB)captured by the automated optical inspection(AOI)while in fast motion is significantly blurred.To enhance scanning and improve efficiency,restoring the image's clarity is essential.This article draws inspiration from the algebra reconstruction technique(ART)and presents a non-blind deblurring method based on ray rearrangement.This approach is founded on the iterative reconstruction algorithm of fast iterative shrinkage thresholding algorithm(FISTA).It employs Nestrov acceleration and ray rearrangement to enhance convergence speed and regularization to diminish the additive noise and artifacts in the image.The findings suggest that this technique demonstrates enhanced proficiency in noise suppression and image restoration in contrast to the conventional inverse filtering approach.
基金项目
国家自然科学基金(61871126)
江苏省重点研发计划(BE2022828)
江苏省前沿引领技术基础研究专项(BK20222002)
NETL
NSTL
万方数据