LCD表面液晶污染的非接触检测新算法

New Method of Non-Contact Inspection for Liquid Crystal Residues on LCDs Surface

针对液晶器件的特点,对其表面残留液晶的非接触式检测技术进行了研究,提出了基于迭代阈值分割和数学形态学的边缘检测方法,可以从目标区域中判断出液晶残留形貌与边缘。首先将原图按坐标分成若干子图像,再对子图像进行迭代阈值分割,然后采用不同尺度的结构元素来检测图像边缘,再进行加权合成来获得边缘图像,并从理论上分析了噪声对边缘提取的影响情况。实验表明,该方法很好地抑制了噪声和保持图像边缘细节,并且能够提取出精确且封闭的残留液晶的边缘轮廓,为下一步缺陷特征量的提取与选择奠定了基础。

Regarding of the characters of liquid crystal products, the non-contact inspection techniques for surface liquid crystal residues were studied. A novel edge-extraction algorithm based on iterative threshold segmentation method and mathematical morphology was proposed, which could find out the shape and edge of the liquid crystal residue. First, the original image was divided into several sub- images according to coordinate. Then, the iterative threshold segmentation was applied to the subimages, and the multi-scale structural elements were used to detect the image edge. Finally,they were synthesized in a weighting way so as to form the edge, and the effect of noise on edge-extraction was analyzed in theory. The experimental results indicated that the method could suppress noise and have much stronger ability to preserve edge details, provide accurate and closed region contours, which lays a good foundation for defect feature extraction and selection.