一种复杂图像背景中残留液晶识别方法  

Identification method of rudimental liquid crystal in complex background image

作　　者：黄妍妍[1,2] 高宝红[1] 张海涛[2] 康军广[2] 

机构地区：[1]河北工业大学微电子研究所,天津300130 [2]河北工业大学网络管理中心,天津300130

出　　处：《液晶与显示》2014年第6期1024-1031,共8页Chinese Journal of Liquid Crystals and Displays

基　　金：02国家重大专项项目(No.2009zx02308);河北省高等学校科学技术研究项目(No.Z2011293);河北省自然科学基金项目(No.E2014202147)

摘　　要：LCD液晶屏是对表面洁净度有严格要求的部件,基于数字图像处理的残留液晶识别方法能够定量、客观地评价表面的洁净度。针对复杂背景图像中残留液晶难以准确识别的问题,本文提出并实现了有效的图像处理流程和算法,包括分割、边缘连接、图像填充、去除多余边界以及噪声等操作。提出了一种边缘连接算子,在对图像进行分割后,将不连续的图像边界连接为闭合,为图像填充奠定基础。提出并实现了一个基于二叉分类方法的模式分类器,用于对污染物几何形状进行自动识别。实验表明,针对单一背景残留液晶的识别与分析方法能够精确获得残留液晶边缘,并且控制了噪声的干扰;针对复杂背景残留液晶识别与分析方法能够对背景存在明显差异的特定类型图像进行准确的处理。LCD screen requires strict surface cleanliness.The inspection method based on digital image processing focusing on the residual liquid crystal can give quantitative and objective assessment on the components surface cleanliness.According to the difficulty to identify the rudimental liquid accurately in a complex background image,this paper proposes an effective image processing flow and algorithm,including the operation of segmentation,edge connection,image filling,redundant boundary and noise removing,etc.An edge connecting operator is designed to connect the discontinuous boundary in the image after the segmentation.A classifier is designed and implemented based on binary classification method for automatic identification of geometry types of contaminants.Experiments show that the residual liquid crystal in the simple background image can accurately inspected,and the interference of the noise is excluded.The identification and analysis method for the complex background image can accurately process particular types of images that their backgrounds have significant differences.

关 键 词：图像处理 污染物识别 模式识别 液晶屏 

分 类 号：TP141.9[自动化与计算机技术—控制理论与控制工程] TP751.1[自动化与计算机技术—控制科学与工程]