基于剪切波变换的光学元件表面缺陷检测方法  

Surface defect detection method of optical components based on shearlet transformation

作　　者：张传博 李林福[1] 陈建军[2] 徐艳丽[1] ZHANG Chuanbo;LI Linfu;CHEN Jianjun;XU Yanli(School of Physics and Mechatronics Engineering,Guizhou Minzu University,Guiyang 550025,China;School of Medical Engineering and Technology,Xinjiang Medical University,Urumqi 830011,China)

机构地区：[1]贵州民族大学物理与机电工程学院,贵州贵阳550025 [2]新疆医科大学医学工程技术学院,新疆乌鲁木齐830011

出　　处：《传感器与微系统》2023年第6期116-119,共4页Transducer and Microsystem Technologies

基　　金：贵州省省级科技计划资助项目(黔科合基础—ZK[2022]211);贵州省大学生创新创业培训计划资助项目(202110672055)。

摘　　要：针对精密光学元件表面划痕缺陷形态大小随机分布,难于检测的问题,提出了一种基于非下采样剪切波变换(NSST)和L0梯度最小化(LGM)的精密光学元件表面缺陷检测的方法。首先,采用NSST将被测元件表面数据分解为不同尺度、不同方向的子块;然后,在子块上基于自适应阈值使用LGM分离被测表面的纹理;最后,重构变换域子块,实现元件表面缺陷的检测。实验结果表明:对包含256×256数据点的表面,特征分离仅需要0.7862 s,满足元件的自动化处理的时间要求。与小波等方法相比,该方法能更准确地分离器件表面的纹理与缺陷,在光学元件表面缺陷检测方面效果更优。Aiming at the problem that the shape and size of scratch defects on the surface of precision optical elements are distributed randomly and difficult to detect,a new method for surface defect detection of precision optical elements based on non-subsampled shearlet transform(NSST)and L0 gradient minimization(LGM)is proposed.NSST is used to decompose the surface data of the tested component into sub blocks with different scales and directions,and LGM is used on the sub blocks to separate the texture of the tested surface based on the adaptive threshold.Finally,the processed sub blocks are reconstructed to realize the detection of component surface defect.The experimental results show that feature separation only takes 0.7862 s for the surface containing 256×256 data points,which meets the time requirements of automatic processing of optical elements.Compared with wavelet and other methods,this method can more accurately separate the texture and defects on the surface of the device,and has better effect on the surface defect detection of optical elements.

关 键 词：缺陷检测 特征提取 非下采样剪切波变换 L0梯度最小化 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TG84[自动化与计算机技术—计算机科学与技术]