| 引用本文: | 胡跃明,黄丹,于永兴,罗家祥.面向高密度柔性集成电路封装基板的精密视觉检测算法[J].控制理论与应用,2024,41(7):1255~1263.[点击复制] |
| HU Yue-ming,HUANG Dan,YU Yong-xin,LUO Jia-xiang.Precision visual inspection algorithms for high-density flexible integrated circuit packaging substrates[J].Control Theory and Technology,2024,41(7):1255~1263.[点击复制] |
|
| 面向高密度柔性集成电路封装基板的精密视觉检测算法 |
| Precision visual inspection algorithms for high-density flexible integrated circuit packaging substrates |
| 摘要点击 96 全文点击 37 投稿时间：2023-09-08 修订日期：2024-06-24 |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| DOI编号 DOI: 10.7641/CTA.2024.30610 |
| 2024,41(7):1255-1263 |
| 中文关键词 柔性IC封装基板 品质控制 外观视觉检测 自动对焦检测算法 微分几何 |
| 英文关键词 flexible IC packaging substrate quality control visual inspection of appearance autofocus detection algorithm differential geometric |
| 基金项目 广州市基础与应用基础研究项目(2023A04J1691), 广西汽车零部件与整车技术重点实验室项目(2022GKLACVTKF04), 国家自然基金面上项目 (62271214), 中央高校基本科研业务费专项资金项目(2023ZYGXZR009), 广州市科技计划项目(2023B01J0037)资助.037). |
|
| 中文摘要 |
| 高密度柔性集成电路(IC)封装基板是电子组件的核心载体, 其制造过程中的刻蚀、显影等工艺需要严格的品质控制. 本文开发了一种融合金相显微镜与工业相机的成像视觉检测系统, 用于柔性IC封装基板的图像采集与品质控制. 针对显微成像视野小、易离焦等问题, 提出了一种基于中值的三元模式局部纹理快速自动对焦算法. 此外,针对高倍镜下柔性IC封装氧化缺陷分布不均匀问题, 引入微分几何工具, 利用曲率等几何特征提出了一种基于动态曲线演化的高精度表面氧化分布特征检测模型. 通过定性与定量的实验研究, 验证了两种算法在快速性和高精度方面的优越性能. |
| 英文摘要 |
| Flexible integrated circuit (IC) packaging substrates with a high density are essential for holding electronic components. Strict quality control is required during their manufacturing operations, which include developing and etching. For the purpose of acquiring images and ensuring the quality control of flexible IC packaging substrates, an industrial camera and metallographic microscope are integrated into an imaging vision inspection system. To address the challenges of narrow field of vision and defocus vulnerability in microscopic imaging, a fast autofocus algorithm for local textures in ternary mode based on median value is proposed. Furthermore, to tackle the uneven distribution of oxidation defects on flexible IC packaging under high magnification, a high-precision surface oxidation distribution feature detection model based on dynamic curve evolution is developed, incorporating differential geometry and utilizing geometric features like curvature. Experimental research, both qualitative and quantitative, confirms the superior performance of both algorithms in terms of speed and accuracy. |
|
|
|
|
|