Use this URL to cite or link to this record in EThOS:
Title: An image quality model based on the influence of Mura defects in TFT-LCDs
Author: Wei, Guo-Feng
ISNI:       0000 0004 2746 1181
Awarding Body: University of Leeds
Current Institution: University of Leeds
Date of Award: 2012
Availability of Full Text:
Access from EThOS:
Mura is a type of defect on LCD's that affects image quality. Due to its subtle nature - gradually and non-uniformly changes in lightness within a specific area, how Mura defects can be effectively inspected and objectively graded has been an issue in the LCD manufacturing industry for many years. Although many studies using different approaches to the Mura detection issue have been carried out, there is still a lack of reliable standard and methods that can be fully adopted in the industry and replace the human eye in the inspection work. Based on demand from the LCD industry, a uniform Mura detection model is needed to provide a stable and reliable inspection system for image quality judgment. For many years, researchers have studied this topic by using uniform colour patches. Few of them extended investigations to conditions where Mura defects are viewed against complex images in background. This might be the first time that study of Mura detection has been extended to patterned and complex image backgrounds. There are four experiments in this study. Influences of colour, Mura size and the masking effect caused by background patterns on Mura detection were investigated. Among them, masking effect is the dominant factor that significantly affects the detection thresholds. Our analyses show that the influence of Mura size was mild for noise backgrounds. Orientations of the Mura patterns had no effect for uniform and noisy backgrounds, though some other studies of visual acuity and contrast sensitivity in humans have shown an unequal sensitivity across orientation. Although colours showed little effect in this study, some seemed to produce more stable results. Particularly for uniform backgrounds, colour and size caused problems for some observers who were unable to maintain consistent thresholds for Mura detection. Mura detection against still and moving pictorial backgrounds were also studied. It seems that the human visual system uses different strategies with different mechanisms to detect Mura patterns on uniform, patterned and pictorial backgrounds respectively. On uniform backgrounds, the human visual system uses colour difference, lightness difference and boundary detection theorem to discern a Mura pattern whereas detecting Mura patterns on patterned backgrounds is highly affected by the masking effect. For static pictorial backgrounds, detecting Mura patterns even relies very much on cognition and thus causes huge problems for observers to recognize them. This situation only improves when the position correlation between a Mura pattern and its background changes from static to dynamic; and it is until then detecting Mura patterns seems to be a quantifiable task again. A Mura detection model was constructed based on knowledge of the human visual system, which is mainly concerned with how a static image pattern projected on the retina is converted into neural signals, and how these signals are interpreted by our brains. Our analyses show that the model delivers more reliable results than the S-CIELAB model when masking effect take places.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available