作　　者： ; (林绪虹）; (刘桂雄）;

机构地区： 华南理工大学机械与汽车工程学院

出　　处： 《光学精密工程》 2005年第z1期206-211,共6页

摘　　要： 介绍空瓶检测的图像处理关键技术,对酒瓶破损模式、空瓶检测方法的特点进行分析。以瓶身检测为例,用三水平三因素的正交设计对空瓶检测的图像处理参数,即Sobel、Prewitt、Roberts等边缘检测算子、处理步长以及决策算法等进行组合优化设计。通过对试验结果和极差值的分析,得出了各参数对空瓶检测精度和实时性影响的主次因素,为优化参数、平衡空瓶检测的准确率与工作时间提供了依据。在与论文实验条件相类似的情况下,推荐采用Prewitt、处理步长为1、加权求和决策算法的算法组合。本方法可推广到不同检测系统要求下所确定的最优参数组合中。 The paper simply introduces the key technology,beer bottle fault modes and the characteristics of bottle detection system based on image processing methods.According to the layout of 3 levels with 3 factors(3~3),3 different parameters are examined such as Sobel,Prewitt and Roberts for the imaging processing methods,by optimizing parameter combinations of steps,decision-making algorithms and edge detection techniques.An analysis of the margin of experimental data,shows that the relative importance of different parameters affecting the accuracy and real-time properties of the system.The orthogonal experiment provides a good idea for optimizing parameters and adjusting the equilibrium constant between accuracy and real-time properties.Under the similar experiment conditions as described in this paper,the optimized parameter combination for the imaging processing methods is recommended,that is one element for one step,weighting addition for decision-making algorithm and Prewitt for edge detection technique.This method can also be generalized to optimize parameters for the requirements of different inspect systems.

关 键 词： 空瓶检测 图像处理 正交设计

领　　域： [自动化与计算机技术—计算机应用技术] [自动化与计算机技术—计算机科学与技术] [机械工程—仪器科学与技术] [机械工程—精密仪器及机械]