DIP外观缺陷机器视觉检测系统

A Machine Vision System for Inspecting Defect of DIP

设计简单光学成像系统实现单个面阵CCD的三维机器视觉检测,利用图像顺序形态学以及基于知识的阈值选择算法对双列直插式芯片DIP(Dual Inl i ne Package)管脚图像进行预处理,检测过程引入质量控制图方法分析系统误差对系统稳定性的影响,采用时间序列建模预测系统检测误差并调整检测结果.实验结果表明系统检测精度高,实时性好,满足多种DIP在线检测的要求.

In this paper, a machine vision system for measuring leg＇s dimension of DIP （Dual lnfine Package） is presented. A simple optics system was designed for three-dimensional inspection by single CCD image sensor. Ranked-order Morphology and threshold selection knowledge-based were used for processing image of DIP. Quality Control Chart was used to analyze error for system stability, time series analycal model was adopted to forecast system errors for adjusting inspecting result. The presented results show that the proposed machine system allows high precision, real time, and robust 100% inspection for online DIP inspecting.