Andrew Munro | Radiant Vision Systems: How do you program a machine to find a dead pixel when it's just a dark spot surrounded by more dark space?
00:05:53 - 00:07:19
Other snippets from this talk
Summary of the clip:
How do you program a machine to find a dead pixel when it's just a dark spot surrounded by more dark space?
A fundamental challenge in microLED inspection is algorithmically detecting a missing or dead emitter. Traditional defect detection algorithms struggle because the defect—a dark spot—is located within a pixel area that is already mostly dark space. This makes it incredibly difficult for software to reliably distinguish a true defect from the normal, non-emissive area between functioning LEDs.
The solution is to create what is called a "synthetic image." This innovative approach processes the high-resolution measurement by first isolating each individual emitter and then averaging its measured luminance across its entire designated pixel area. This computation effectively "fills in" the dark, interstitial space between the emitters, transforming the sparse array of light points into a continuous image.
The result is a synthetic image where each pixel is uniformly filled, closely resembling a measurement from a traditional LCD or OLED display. In this transformed view, a dead or missing microLED becomes the *only* dark pixel, standing out clearly against its illuminated neighbors. This allows standard, proven pixel defect algorithms to be applied effectively and reliably, solving the "darkness in darkness" problem.
In this short video, you can learn:
* The algorithmic difficulty of detecting defects in sparse-emitter displays.
* The concept of creating a "synthetic image" to simplify defect analysis.
* How this technique enables the use of traditional LCD-style defect detection algorithms for microLEDs.
📋 **Clip Abstract** Detecting dead microLEDs is challenging because they are dark spots surrounded by more dark space. This solution creates a "synthetic image" by averaging each emitter's light over its pixel area, which enables the use of traditional and reliable defect detection algorithms.
🔗 Link in comments 👇
#MicroLEDInspection, #DefectDetection, #SyntheticImaging, #SparseEmitterDisplays, #MicroLEDDisplays, #ARDisplays
This is a highlight of the presentation:
Addressing the Visual Inspection Challenges of microLED Displays & Lighting in Vehicles
More Highlights from the same talk.
00:03:32 - 00:05:49
How do you measure individual micron-sized LEDs when your camera sensor runs out of pixels?
How do you measure individual micron-sized LEDs when your camera sensor runs out of pixels?
To properly resolve and measure individual microLEDs across an entire display, an ultra-high-resolution imaging colorimeter is essential. Systems with 61 to 151-megapixel sensors are commonly used to achieve sufficient oversampling, for example, mapping approximately 3x3 camera pixels to every single display pixel on a 4K screen. This high sensor resolution is the first critical step in obtaining a detailed image of the display's emitters.
However, even with these powerful sensors, a challenge remains. The microLEDs themselves are often only tens of microns in size, occupying just a fraction of their total pixel area, with significant dark space between them. In these "resolution-starved" scenarios, the camera may still struggle to isolate each emitter cleanly, especially as LED sizes continue to shrink.
To overcome this, a clever technique called "spaced pixel patterns" is employed. Instead of turning on all pixels at once, the system iteratively illuminates and measures subsets of pixels—for example, turning on only pixel 1, 4, and 7 in the first measurement, then 2, 5, and 8 in the next. These individual measurements are then computationally stacked to create a composite image, which effectively increases the measurement resolution and allows for the precise characterization of each individual LED.
In this short video, you can learn:
* The megapixel count required for effective microLED display inspection.
* Why high sensor resolution alone may not be sufficient for tiny emitters.
* A "spaced pixel pattern" method to overcome hardware resolution limits for precise measurement.
📋 **Clip Abstract** Achieving the resolution to inspect individual microLEDs requires both ultra-high-megapixel imaging colorimeters and advanced techniques. The "spaced pixel pattern" method overcomes hardware limitations by iteratively measuring pixel subsets to create a high-fidelity composite image.
🔗 Link in comments 👇
#MicroLEDInspection, #ImagingColorimetry, #HighResolutionSensors, #SpacedPixelPatterns, #ARdisplays, #MicroLEDDisplays
00:08:12 - 00:10:21
Why does a standard display calibration tool give you the wrong color and brightness when measuring a microLED screen?
Why does a standard display calibration tool give you the wrong color and brightness when measuring a microLED screen?
Achieving accurate luminance and color measurements is critical for correcting non-uniformity (Mura) in microLED displays, but standard calibration methods fall short. Imaging colorimeters are typically calibrated in the factory using "Illuminant A," a NIST-traceable, broad-band tungsten light source. However, the narrow-band spectral emission of red, green, and blue microLEDs is completely different, and measuring them with an Illuminant A calibration will produce significant, unacceptable errors in both luminance and chromaticity data.
To solve this, one approach is an enhanced factory calibration. This method uses a set of narrow-band LED sources that are spectrally similar to those found in the target microLED display. This pre-calibration allows the system to measure narrow-band emitters with good out-of-the-box accuracy, closely matching industry standards and providing a solid foundation for Mura correction.
For the highest possible accuracy, the gold-standard solution is to perform an in-situ calibration using a spectroradiometer. This can be done with a separate spot spectroradiometer or, more effectively, with an integrated system that combines a colorimeter and a spectroradiometer sharing the same optical path. This integrated approach creates a perfect correction profile for the specific LEDs on the display being tested, ensuring the most accurate luminance and chroma data for optimal non-uniformity correction.
In this short video, you can learn:
* Why standard "Illuminant A" calibration is unsuitable for narrow-band microLEDs.
* The concept of an "enhanced color calibration" using representative LED sources.
* The benefits of using an integrated spectroradiometer for the highest-accuracy, on-the-spot calibration.
📋 **Clip Abstract** Standard calibration sources are mismatched for the narrow-band emission of microLEDs, leading to inaccurate color data. Accurate Mura correction requires either an enhanced factory calibration or, for the highest precision, an in-situ calibration using a spectroradiometer.
🔗 Link in comments 👇
#MicroLEDCalibration, #Spectroradiometer, #MuraCorrection, #ImagingColorimeter, #ARDisplays, #DisplayMetrology