首页|基于网格的多目标模板匹配晶粒位置检测方法

基于网格的多目标模板匹配晶粒位置检测方法

Grid-based Multi-targets Template Matching Die Position Detection Method

扫码查看
针对传统方法中晶粒位置检测的耗时长、精度低等局限性,提出一种基于网格的多目标模板匹配晶粒位置检测方法.通过改进传统的模板匹配方法,结合非极大值抑制算法,将芯片晶粒的检测速度和精度提高.实验结果表明:该算法在单一同种晶粒的算法识别率能够达到97%以上,单张图像耗时<200 ms,能够克服明暗不同造成的检测困难,达到技术指标要求.
To address the limitations of traditional methods in terms of long detection time and low accuracy,a multi-objective template matching algorithm detecting position of die based on grid is proposed.By improving the traditional template matching method and combining with non-maximum suppression algorithm,the detection speed and accuracy of chip dice are improved.Experimental results show that the proposed algorithm has a recognition rate of over 97%for a single same die,and the single image processing time does not exceed 200 ms.It can overcome the detection difficulties caused by different brightness and meet the technical specifications.

photoelectric detectiondie position detectionmachine visionnon-maximum suppressiongridmulti-target template matching

周书辰、陈晓荣、王子旋

展开 >

上海理工大学光电信息与计算机工程学院,上海 200093

光电检测 晶粒位置检测 机器视觉 非极大值抑制 网格 多目标模板匹配

国家自然科学基金

52175513

2024

计量学报
中国计量测试学会

计量学报

CSTPCD北大核心
影响因子:0.303
ISSN:1000-1158
年,卷(期):2024.45(5)
  • 23