液晶顯示器(LCD)具備輕、薄、省電、零輻射、省空間及攜帶方便之特性,廣為應用於消費性電子產品上。目前TFT-LCD的瑕疵檢測仍然仰賴大量人工進行,使用人工除檢測效率較差外,其主觀的判斷亦是造成檢測上的問題,故本研究導入機器視覺於TFT-LCD面板之表面瑕疵檢測,以避免人工檢測誤判造成之損失,有效降低生產成本與提昇良率。本研究瑕疵偵測對象設定為TFT-LCD面板中人眼不易觀察之微觀瑕疵,包括指觸、粉塵、配向膜孔洞與刮痕瑕疵,透過對TFT-LCD面板之瑕疵偵測,於製程中有效地偵測出瑕疵位置。 近年來TFT-LCD有解晰度愈高、尺寸愈大趨勢,故取像方式需採線性掃描之模式以取得範圍較廣、解晰度較高之影像。本研究即是針對線性掃描之取像模式,設計一有效之自動檢測方法,針對TFT-LCD面板之一維掃描影像進行微觀瑕疵的偵測(Detection),由於TFT面板表面是由垂直信號線與水平閘線之規律紋路所構成,因此本研究直接針對TFT-LCD面板所掃描之一維灰階訊號,以一維傅立葉影像轉換(1D Fourier transform)與反傅立葉影像還原的技術,配合濾除頻譜中TFT面板規律性紋路之頻率元素與保留瑕疵能量在頻譜中所貢獻最多的區域之策略,將TFT面板規律紋路加以濾除並且保留瑕疵特徵,再經過統計管制界限法突顯瑕疵達到瑕疵偵測目的,實驗結果顯示本研究方法對TFT-LCD面板之微觀瑕疵的偵測具良好效果。
The surface defects in TFT-LCD array panels generally can be categorized as macro- and micro-defects. Sizes of micro-defects are generally very small and cannot be easily detected by naked eyes. In this study, we propose an automatic 1D visual system for defect detection, and especially focus on micro-defects in TFT-LCD array panels in the manufacturing process. Micro-defects of TFT-LCD array panels investigated in this study include pinholes, particles, scratches and fingerprints. The main trends of TFT-LCD are toward the “Large Size” and the “High Resolution”. Therefore, a visual system may require the line scan cameras to get TFT-LCD panel images for wider range and higher resolution. In this research, a one-dimensional defect inspection scheme is proposed for the purpose of cooperating with the line scan mode. Defects in a one-dimension gray signal of TFT-LCDs are detected directly by the proposed method. Since the surface of TFT panel is formed with equally-spaced vertical data lines and horizontal gate lines, the profile of its one-dimension gray signal is composed of a periodical-peak structural pattern. The distance between two neighboring peaks in the 1D signal represents the spacing between two adjacent data lines. The periodical pattern of one-dimension gray signal is analyzed by the 1D Fourier transform. By eliminating the frequency components in the power spectrum that represent the periodical-peak structural pattern of the one-dimension gray signal of TFT-LCDs, and back transforming the signal using the inverse Fourier transform, we can effectively remove the repetitive, periodical pattern and enhance local defects in the reconstructed signal. Experimental results from a variety of TFT-LCD array samples have shown the efficacy of the proposed method for micro-defect detection.