一种复杂图像背景中残留液晶识别方法

发布时间：2018-10-12 09:57

【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。
[Abstract]:LCD screen is a component with strict requirements for surface cleanliness. The residual liquid crystal recognition method based on digital image processing can quantitatively and objectively evaluate the surface cleanliness. Aiming at the problem that it is difficult to identify the residual liquid crystals in complex background images, an effective image processing flow and algorithm is proposed and implemented in this paper, including segmentation, edge connection, image filling, removal of redundant edges and noise, and so on. In this paper, an edge join operator is proposed. After segmenting the image, the discontinuous image boundary is connected to the closed image, which lays the foundation for the image filling. A pattern classifier based on binary classification method is proposed and implemented, which is used to recognize the geometric shape of pollutants automatically. The experiments show that the residual liquid crystal edge can be accurately obtained by the method of identification and analysis for the residual liquid crystal with a single background, and the noise interference can be controlled. The methods of residual liquid crystal recognition and analysis for complex background can deal with specific types of images with obvious differences in background.
【作者单位】：河北工业大学微电子研究所;河北工业大学网络管理中心;
【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)
【分类号】：TN873.93;TP391.41

【相似文献】