| Summary: | 碩士 === 國立臺灣科技大學 === 電機工程系 === 97 === The thin film transistor liquid crystal display(TFT-LCD) is one of the most popular flat panel displays that can be found in many products in our daily life such as digital camera, cellular phone, notebook computer, MP3, monitor and TV, etc.
A big challenge in the manufacturing of TFT-LCD is to detect the defects such as Mura, SIMI, pinhole, scratch, grain, SIRO-NUKE, particle, etc. Due to the equipment limitation, we test only some defects that includes Mura, SIMI, pinhole, scratch, grain, SIRO-NUKE.
In this paper, we have developed a machine vision algorithm for automatic TFT-LCD defect detection based on Haar wavelet transform and spatial filtering. Experiments show that the proposed algorithm can effectively eliminate perpendicular periodical patterns, inhomogeneous illumination on TFT-LCD panels and has better performance for most of defects.
|
|---|
