基于显著性模型和区域生长法的药卷缺陷检测  

A Defect Detection of Cartridged Explosive Based on Saliency Model and Region Growing Method

作　　者：刘学福[1] 何小敏[1] 许亮[1] 徐海波[1] 

机构地区：[1]广东工业大学自动化学院,广州510006

出　　处：《科学技术与工程》2015年第4期125-130,共6页Science Technology and Engineering

基　　金：国家科技支撑计划(2012BAK13B02);国家自然科学基金(21176089;21376091)资助

摘　　要：针对工业炸药生产过程中药卷表面裂痕的包装缺陷问题,运用机器视觉技术,提出一种基于显著性模型和局部方差区域生长法的药卷缺陷检测方法。该方法经过图像预处理,对药卷图像进行背景估计与差分;利用显著性模型,提取缺陷特征;通过局部方差区域生长法,分割目标区域,完成对缺陷药卷的检测。实验结果表明,该算法可快速有效地提取缺陷区域,平均检测时间为55.72 ms,缺陷检测率高达96.36%。Aimming to the problems which there are surface defections of cartridged explosives in the packing prcoesses. A detecting method for surface flaws of cartridged explosives was proposed which are based on saliency model and local variance region growing method by using machine vision technology. The proposed method firstly performs image pre-processing,then the images are treated by the method of background estimation and differential processing. And the defect features are extracted by using saliency model. The detected objects are separated by the local variance-based region growing method. The experiment results show that the proposed method can be quickly and efficiently extract the defect area,and the average detecting time are 55. 72 ms and the accurate rate of detection is up to 96. 36%.

关 键 词：药卷 机器视觉 显著性模型 区域生长 缺陷检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]