自动金丝球引线键合机图像识别系统  

Image Recognition System of Automatic Gold Wire Bonder

作　　者：陈良锋[1] 张鸿海[1] 马豪[1] 

机构地区：[1]华中科技大学微系统研究中心,武汉430074

出　　处：《计算机与数字工程》2007年第8期98-99,156,共3页Computer & Digital Engineering

摘　　要：自动金丝球焊机是半导体制造后端工序的关键设备之一,图像识别系统是其核心技术。通过直方图均衡、中值滤波实现图像的增强、去噪,采用MPSA算法减少计算量,同时结合计算机并行处理技术,利用MMX指令对图像处理部分进行代码优化。该图像识别系统已成功应用于手动金丝球焊机中,将其改造为自动金丝球焊机。The automatic wire bonder is one of the primary equipments in semiconductor back-end production.Image recognition system is a crucial issue.Median filtering algorithm for suppressing noise,histogram algorithm for image enhancement,and multi-resolution pagoda structure algorithm(MPSA) for reducing the matching time are proposed in this paper.At the same time,integrating with parallel processing technology,the image processing system is optimized by means of MMX instructions.By adding this Image recognition system,a manual wire bonder has been successfully ameliorated to be an automatic one.

关 键 词：金丝球焊机 图像识别 机器视觉 MMX 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]