What Is The Mura Test?

The Mura Test is a specialized display inspection procedure used to detect, quantify, and classify mura defects, which are regions of luminance or chromaticity non-uniformity that appear as cloudy patches, streaks, spots, or blotches on a display panel. The term mura comes from the Japanese word 斑 (mura), meaning "unevenness" or "irregularity," and has become the universal industry term for these subtle, low-contrast defects.

Unlike dead pixels or hot pixels, which are discrete, binary failures of a single picture element, mura is a regional defect caused by process-induced gradients in the panel's manufacturing. A proper mura test must therefore detect not just whether a defect exists, but how visible it is to the human eye and how it should be classified for quality grading.

What Mura Looks Like and Why It Matters

Mura defects typically present with three characteristics that make them uniquely challenging:

• Low contrast: Often only 1 to 5% luminance deviation from surrounding areas.
• Blurry contours: Unlike sharp pixel defects, mura has soft, undefined edges.
• Irregular shapes: No fixed geometry, making template matching ineffective.

Research published in the SID Journal notes that mura is "small, and the gray change is unknown," which is precisely why manual inspection misses defects and automated detection is technically demanding.

Mura Classification

Industry literature, including research published on ResearchGate, classifies mura defects by shape and size:

Mura Type	Description	Typical Cause
Spot Mura	Small, localized bright or dark spots	Particle contamination, photoresist defects
Line Mura	Horizontal or vertical streaks	Coater stripes, TFT gate/source line defects, mechanical scratches
Region Mura (Cloud Mura)	Large, blurry blotches	Cell-gap non-uniformity, backlight unevenness, alignment film irregularities
Edge Mura	Brightness fall-off near panel edges	Sealing process variation, backlight light-guide issues
Color Mura	Chromaticity non-uniformity	Color filter thickness variation, white balance shift

The Industry-Standard Metric: SEMU

The most widely cited quantitative metric for mura is SEMU (Spatial Enhancement Mura Unification), proposed by SEMI (Semiconductor Equipment and Materials International) in 2002 under specification SEMI D31-1102. SEMU translates measured luminance differences into a perceptual index that correlates with how visible a defect appears to a human observer.

The general principle is:

• SEMU = 0 means no detectable mura.
• SEMU = 1 represents the just-noticeable difference (JND) threshold.
• SEMU > 1 indicates progressively more visible defects.

A typical production specification might use a SEMU threshold of 1.7 as referenced in machine-learning-based detection research, with stricter thresholds for premium panels and more relaxed thresholds for industrial-grade displays.

Newer perceptual metrics published in Measurement journal (ScienceDirect) build on SEMU by incorporating Human Visual System (HVS) modeling, contrast sensitivity functions, and oblique-effect compensation to achieve correlation up to 0.77 with subjective inspector rankings for line mura.

How the Mura Test Is Performed

Manual Visual Inspection

The traditional method, still used as a verification step, requires:

• A dark room with ambient light below 10 lux.
• Standardized test patterns (full-field gray at multiple levels: typically 32, 64, 128, 192 gray).
• Trained inspectors viewing from multiple angles and distances.
• Defect documentation against a written specification.

The drawbacks are well-known: subjectivity, fatigue, lack of repeatability, and inability to quantify defect severity consistently across shifts.

Automatic Optical Inspection (AOI)

Modern mura testing is performed using AOI systems combining high-resolution imaging colorimeters, controlled lighting, and sophisticated image processing. The typical AOI workflow includes four stages:

1. Image Acquisition: A high-resolution industrial camera (typically 29 MP or higher) captures the panel displaying calibrated gray-field test patterns.
2. Preprocessing: Algorithms remove illumination non-uniformity, shading, and camera noise. Techniques include morphological filtering, Gabor filtering, and gamma correction.
3. Feature Extraction: The system identifies regions of non-uniformity using spatial frequency analysis. Common techniques include Fourier transforms, wavelet decomposition, PRESS residual analysis, and level-set segmentation.
4. Classification and Grading: Detected defects are scored against SEMU thresholds and pass/fail criteria, typically yielding Grade A, B, or reject classifications.

Algorithmic Methods

Modern mura test algorithms span several technical approaches:

• Gabor Filtering: Excellent at separating mura from textured backgrounds by tuning to specific spatial frequencies.
• Multivariate Statistical Analysis: Hotelling T² statistics integrate color coordinates to detect chromatic mura.
• Wavelet Transforms: Effective for low-frequency (large, blurry) defects like cloud mura.
• Level-Set Methods: Region-based active contours improve defect boundary delineation.
• Deep Learning: Recent published research demonstrates YOLOv4-tiny with spatial pyramid pooling and piecewise gamma correction with adaptive particle swarm optimization achieving detection accuracy above 93% on low-contrast images.
• Support Vector Machine (SVM) Classification: USPTO-published methods use feature vectors combining texture and image moments for white-spot mura detection.

When the Mura Test Is Performed

Mura inspection happens at multiple stages of display manufacturing:

• TFT Array Process: Inline micro-defect inspection using nonlinear principal component analysis catches the root causes that later manifest as cell-process mura.
• Cell Process: After liquid crystal injection or OLED deposition, panels are tested with controlled gray patterns to identify cell-gap and emission uniformity issues.
• Module Assembly: Final mura inspection after polarizer lamination, driver IC bonding, and backlight integration.
• Outgoing Quality Control (OQC): A final mura test confirms compliance with the customer's specification before shipment.

Mura Test Challenges Specific to Flexible OLED

Recent industry analysis highlights a unique challenge in flexible OLED manufacturing: traditional white alumina ceramic chucks reflect stray UV light during photolithography and reduce signal-to-noise ratio during AOI, making accurate SEMU quantification difficult. Black alumina chucks are being adopted as an absolute light absorber to solve this problem and improve mura detection accuracy.

Mura Test vs. Demura

It is important to distinguish two related but distinct processes:

• Mura Test (this article): An inspection step that detects, quantifies, and classifies mura defects to grade or reject a panel.
• Demura: A correction process that uses uniformity measurement data to compute per-pixel compensation coefficients applied through the timing controller or external IC, masking mura that cannot be physically eliminated.

The mura test typically feeds the demura process: measure first, then correct what cannot be reworked at the cell level.

Engineering Takeaway

The Mura Test is the front-line quality gate of every display fab, transforming subjective human perception of "uneven brightness" into objective, repeatable, quantifiable metrics. It has evolved from manual visual inspection in dark rooms to AI-driven AOI systems running deep neural networks on multi-megapixel images. As display resolutions push higher and panel sizes grow (Gen 8.7 OLED lines), mura test throughput, sensitivity, and classification accuracy have become competitive differentiators among display manufacturers. Mastering mura detection is essentially mastering the limits of human perception, which is why the field continues to be a rich area of academic research and industrial investment.