基于特征提取和SVM分类的LED芯片缺陷快速检测与实现  

LED Chip Defect Sorting System Based on Feature Extraction and SVM

作　　者：吴乾生 高健[1] 张揽宇 郑卓鋆 WU Qian-sheng;GAO Jian;ZHANG Lan-yu;ZHENG Zhuo-yun(Provincial and Ministry Jointly Build the State Key Laboratory of Precision Electronics Manufacturing Technology and Equipment,Guangdong University of Technology,Guangdong Guangzhou 510006,China)

机构地区：[1]省部共建精密电子制造技术与装备国家重点实验室广东工业大学,广东广州510006

出　　处：《机械设计与制造》2024年第6期250-255,共6页Machinery Design & Manufacture

基　　金：国家自然科学基金NSFC面上项目(52075106)。

摘　　要：针对LED芯片工业化生产中人工目检缺陷检测效率低,速度慢,易疲劳,受主观影响等问题,这里研究设计了一套LED芯片缺陷的快速检测系统,提出了一种分角度多方向快速卷积、分区统计特征量、支持向量机分类的LED芯片缺陷识别算法,基于QT和Opencv开发了一套LED芯片缺陷快速检测与分类系统,搭建LED芯片检测系统的硬件平台,实现LED芯片实时在线缺陷识别和自动分拣。基于研发的LED芯片缺陷快速检测与分拣系统样机,对LED芯片进行了缺陷分拣的测试。Aiming at the problems of low efficiency,slow speed,fatigue,and subjective influence of manual visual inspection of defects in the industrial production of LED chips,this paper studies and designs a set of rapid detection system for LED chip de-fects,and propose an LED chip defect recognition algorithm based on multi-angle and multi-direction fast convolution,regional statistical feature quantity,and support vector machine classification.Based on QT and Opencv,a rapid detection and classifica-tion system for LED chip defects was developed,and the hardware platform for the LED chip detection system was built to realize real-time online defect recognition and automatic sorting of LED chips.Based on the developed LED chip defect sorting system prototype,the LED chip was tested for defect sorting.

关 键 词：LED 特征量提取 PCA 支持向量机 多分类 

分 类 号：TH16[机械工程—机械制造及自动化] TB478[一般工业技术—工业设计] TP29[自动化与计算机技术—检测技术与自动化装置]