摘要
针对零件在生产检测中识别与定位难的问题,基于LabVIEW开发环境下利用视觉工具包,采用灰度值金字塔算法进行模式匹配,设计亚像素精度的零件识别系统。通过抽取单个颜色平面分量实现原始图像的灰度化,经过高斯滤波预处理,对目标进行模板图像的创建与学习,根据设置的匹配参数和灰度值金字塔算法,设计人机交互界面和程序框图实现系统功能,对识别的目标进行矩形框标识并输出像素坐标。实验结果表明,当目标发生旋转、光照变化、部分遮挡时,能准确识别目标,系统准确性高、实用性强,对自动化生产中识别、进一步定位测量有参考价值。
As it was difficult in identifying sand locating parts in the production,part identification system based on sub-pixel precision was designed,using Visual Toolkit in the LabVIEW development environment and gray value pyramid algorithm for pattern match.Extracting the single color plane components could realize the original image gray process and smoothing the image was done through Gauss filtering,learning and creating template image of targets.Human-computer interaction interface and program block diagram based on parameters set and gray value pyramid,were designed to realize system function,identify and output pixel coordinates for the identified target.The experimental results showed that the targets,despite retated,changed illumination and shielded part,could be accurately identified and positioned.Being high accuracy and strong practicability,the system boasted reference value on identification and location measurement in automatic production.
作者
任楷飞
孟令军
顾泽凌
REN Kaifei;MENG Lingjun;GU Zeling(National Key Laboratory for Electronic Measurement Technology,North University of China,Taiyuan 030051,China)
出处
《中国测试》
CAS
北大核心
2018年第7期83-87,共5页
China Measurement & Test
基金
国家自然科学基金资助项目(61774138)
作者简介
任楷飞(1990-),男,山西晋中市人,硕士研究生,专业方向为动态测试存储技术、机器视觉与模式识别。
