MURA defect detection requires a high quality sensor since it has to challenge the eyes sensibility which is in many aspects very high. In this respect, UMaster is certainly the most sensitive solution of the market thanks to its cooled CCD sensor, high quality optics and high transmission filters. The other difficult task is to automatically analyze, quantify and classify the MURA defects. ELDIM provides in option a full automated solution EZMURA for this task. The software performs automatically the following operations:

Segmentation allows finding defect entities in 2D luminance images from the display. A wide range of segmentation algorithms have been developed: local threshold, region growing, watershed, fuzzy clustering, edge detection, active contour (snakes), Markov random fields, neural networks… Because of MURA family complexity, it is impossible to find a single ideal algorithm able to detect simultaneously all kinds of MURA defects. Each segmentation approach has its own advantages and drawbacks and allows focusing only on a given kind of MURA defect. EZMURA software uses six different algorithms to detect different family of MURA defects.

Once a defect has been identified, it has to be quantify so as to decide whether its intensity is acceptable or not. EZMURA uses SEMI quantification or custom quantification so as to fit an existing quotation system (based on operator visuals). Critical values can be found (one for each sub classes of defect) and applied for final pass/fail assessment.

Light can have different states of polarization. It can be randomly polarized (or unpolarized). This is generally the case of natural light. It can be also linearly polarized. In this case the electric field is oscillating always in the same plane. In any case the electric field characterizing any light wave can be separated in two components:

The polarized component can be defined by its elliptical coefficients (ellipticity ε and orientation α) as shown hereafter. Unpolarized light component is defined by the degree of polarization ρ given by the ratio of the intensity due to polarized component over the total light intensity. The three previous parameters can be combined with the intensity to provide Stokes vector.

The polarization option of UMaster includes three polarizers at different orientations (0, 45 and 90°) and two wave-plates at different orientation (45 and 135°). The system makes automatically seven measurements with different polarization configurations and computes automatically the polarization parameters and the Stokes vectors. The measurement is made at a given wavelength defined by a band pass filter (for example 550nm).

Up to now imaging polarization has not been used to characterize display homogeneity. It is surprising since LCD are essentially polarizer modulators but no practical system was available up to now. UMaster covers this gap and allows polarization characterization of displays and their components. For examples CCFL backlights are always partially polarized and can exhibit complex structures on their surface. That will impact the performances of the LCDs.

LWAP software interface for calibration coefficient computation