基于小波多尺度分析的TFT-LCD云纹自动检测技术

Automatic Mura Detection for TFT-LCD Based on Wavelet Multi-scale

本文提出了一种基于背景拟合及小波多尺度分析的TFT-LCD云纹自动检测方法,大大减少了液晶显示器背景总体不均匀给检测结果带来的影响,同时综合了小波小尺度和大尺度变换的优点,成功地将云纹区域提取出来。国际相关组织已经制定了云纹量化的标准,但是仅仅考虑到云纹的对比度、面积这两个因素,本文根据人类视觉空间性质,提出了云纹的位置参数,开展了相关的人因试验,对云纹量化标准做了适当的补充。最后,通过对13片LCD样板的检测,实验结果和人眼观测的结果完全一致。

An automatic inspection method was proposed based on polynomial surface fitting and wavelet multi-scale. The influence of non-uniform background is successfully removed and Mura area was gotten. Related international organizations have drafted a standard on quantification Mura considering on area and contrast. Parameter of location was proposed according to the visual space of human eyes. We carried out some experiments on human factors, and got the relationship between the location of the Mura area and threshold of visual contrast. At last, 13 real TFT-LCD panel samples were evaluated with the proposed method, and the result is consistent with human intuition.