There are three main methods for visual inspection of illuminated displays in high-speed production, whether it is in line, or at the end of the line in final inspection:
- Machine vision-based inspection – Extremely quick for simple tests. Does not reflect human visual experience for many tests
- Human inspection – Relatively slow and variable when compared to electronic testing methods but easily handles moderately complex testing requirements.
- Imaging colorimeter-based inspection – Somewhere between the preceding two techniques in speed. “Sees” like humans with a very high degree of repeatability and reliability.
The utilization of imaging colorimeter systems and associated analytical software to examine color uniformity, display brightness, and contrast, and to identify defects in displays, is well established.
A big difference between machine vision and imaging colorimetry is imaging colorimetry’s accuracy in matching human visual perception for color and light uniformity (and non-uniformity).
This article outlines how imaging colorimetry can be employed in a fully-automated testing system to identify and quantify defects in high-volume, high-speed, production environments. The test setup, and the range of tests which may be carried out – spanning complex mura detection and evaluation to simple point defect detection, are discussed.
Testing using imaging colorimetry is faster, more flexible, and more repeatable than human visual inspection, and more accurate in matching human visual perception than machine vision.
Imaging colorimeters accurately capture the spatial relationships in the variation of light and color across a display, making this measurement method ideal for assessing visual performance.
Imaging colorimetry systems are CCD-based imaging systems that are calibrated to possess the same response to brightness, light, and color as a standard human observer, as defined by CIE models.
Colorimeters supply accurate, simultaneous measurements of color and brightness and their spatial relationship. Data is produced, which can be readily utilized to establish display uniformity and contrast performance when they are employed to image displays.
Differences in uniformity can also be examined to identify and locate potential display defects. Three important challenges for display measurement and analysis are:
- Identify defects with a high correlation to human visual perception
- Assess the severity of the defects
- Perform analysis quickly and with high repeatability
The analysis and quantification of defects can form the basis for decisions relative to the display component that caused the defect and to establish the next steps, for example, to scrap the display or to return it for repair, heightening the quality testing effectiveness potentially decreasing costs.
Measurement Components and Set Up
An imaging colorimeter can be utilized to gather accurate, extensive, high-resolution data to describe the performance of a particular display, by specifying an appropriate automated test sequence.
Often, this measurement data can be gathered in a few seconds to a minute for typical test sequences, depending on the display technology and resolution. By employing new mura (blemish) analysis methods for uniformity defects, these images can be utilized to establish fine-scale differences between defects that are directly related to their physical cause.
Combination measurement control and analysis software is required for automated measurement and analysis of displays with an imaging colorimeter. The typical structure of the system that Radiant Vision Systems have developed for this application can be seen in Figure 1.
Figure 1. Flat panel display (FPD) AVI test set up with an imaging colorimeter under automated software control.
The key components of the system are:
- A scientific-grade imaging colorimetry system.
- PC-based measurement control software that controls both the imaging colorimeter and test image display on the device under test.
- A suite of image analysis functions that enables various tests to be performed.
The result is a system which can supply quantitative, automated inspection for a number of display defects, like line defects, point defects, and mura.
The automated test software architecture used in this instance is Radiant Vision Systems TrueTest™ Software. This software is made up of a core set of measurement control modules which supply the interface with the imaging colorimeter and the display under test.
Employing function calls to produce various measurements of white, red, blue, and green display screens at various brightness settings for uniformity analysis, or of checkerboard patterns for contrast measurement, a series of specific test functions is built on this base.
A partial list of tests implemented includes:
- Line defects
- Mura defects
- Color mura
- ANSI brightness
- Pixel defects
- ANSI color uniformity
- Black Mura
- Compare points-of-interest
- Diagonal pattern mura
- Blob analysis
- Checkerboard contrast
The TrueTest user interface enables tests to be chosen and sequenced, and test parameters and pass / fail criteria to be chosen where relevant. The user interface supports both an administrator mode with full control over test set-up, and an operator mode which only permits test execution for use in production applications.
Application to Display Defect Detection
A wide scope of display defects can be identified as line defects and pixel defects, damage to the screen (like scratches), physical imperfections in screen manufacture (such as delamination), and imperfections in image uniformity (like mura).
Using recent developments for quantifying visual perception, these defects can be classified numerically according to how noticeable they are (or are not) to human observers. This analysis method is highly repeatable and quick.
A second category of AVI tests detect display defects. Some defects have well-defined physical characteristics, such as point defects and line defects. Others are more random in structure, such as light leakage and mura.
This technique can be utilized with a number of display technologies, including OLED, LCD, LED, and projection displays. These defect detection and classification techniques are shown here through the analysis of multiple displays.
A photopic measurement of a display with a line defect can be observed in Figure 2. As seen in Figure 3, the analysis software identifies this defect and shows it on the display image. Line defects are an example of a defect for which identifying a root cause is simple.
Figure 2. Photopic measurement of a display screen with a line defect visible.
Figure 3. The line defect is identified by the imaging colorimeter AVI software; the location of the defect is identified on screen for the user.
A photopic measurement of a display with a point defect can be observed in Figure 4. As shown in Figure 5, the analysis software identifies this defect and indicates it on the display image. If the analysis establishes that the failure is the result of an LCD pixel being stuck on, point defects can be classified as a failed pixel.
Figure 4. Photopic measurement of a display with a point defect – can you see it?
Figure 5. The point defect is identified by the imaging colorimeter AVI software and marked on the display screen; we have zoomed in to make it more visible.
Yet, direct viewing from a single angle is unable to establish the difference between a particle on the back surface of the display glass and a dead pixel. In this instance, to discriminate and classify the cause, secondary examination is required.
Both detection and classification can be more complex for mura. Generally, they are non-uniformities in color or luminance which cover an extended, irregular area. Mura are detected by identifying luminance or color contrasts that surpass a perceivable threshold.
Yet, as human perception of this contrast depends on multiple factors including spatial frequency, viewing distance, and orientation – relevant mura cannot be identified by looking at simple, absolute values of contrast.
New advances in modeling human visual sensitivity to display defects permit the quantification of mura in terms of “just noticeable differences” (JND). The JND scale is defined so that a JND difference of 1 would be just noticeable based on a sampling of human observers.
Using an absolute scale, a JND value of 0 shows no visible spatial contrast and an absolute JND value of 1 represents the first noticeable spatial contrast – which enables the grading of display defects for display technologies.
So, an imaging colorimeter measurement of spatial distribution of color and luminance may be processed to produce a JND map of the image where mura defects are graded with a direct correlation to human visual perception.
Processing steps in identifying the mura are shown in Figures 8 and 9. A difference image is produced to show luminance deviations relative to a reference image as an intermediate step.
Figure 6. Imaging colorimeter measurement of a display with a mura defect – can you see it?
Figure 7. The mura defect is identified on the display by the imaging colorimeter AVI software. Its extent is shown, along with a JND value.
Figure 8. A difference image shows luminance deviations relative to a computed reference image. The location of the mura is highlighted.
Figure 9. A “false-color” JND map of the display is shown. Both light leakage at the edge of the display and a significant mura defect are identified with larger JND values.
A JND map of the display is then computed. It is worth noting that the mura test shown in Figure 7 ignored edge effects deliberately, which are readily apparent in the JND image. These effects can be identified easily and classified separately.
Imaging colorimeter-based AVI testing systems are able to identify and quantify display defects reliably and quickly. To establish or classify the root cause of the defect and decide the disposition of the display will occasionally need human inspection.
There is a 1-to-1 relationship between the identified defect and a cause in a lot of instances, like for the line defect shown in Figure 3. Classification is immediate and human inspection is not required in these cases.
There are a number of possible causes in other cases, like for some mura, so additional information is needed to complete classification. An efficient way to carry out this classification is to have the human operator make a choice about which of multiple causes could be the correct one.
TrueTest indicates for the operator the exact location and details of the defect that requires further examination, in order to optimize efficiency when human classification is required. Human judgment can be focused and accelerated by targeting the defect that requires classification specifically and by presenting appropriate detail.
For the point defect shown in Figures 4 and 5, the operator is presented the exact location and information on the dark point, enabling them to determine quickly whether this is a particle on the backside of the display glass or a dead pixel, for example.
The tests described in this white paper were performed using a Radiant Vision Systems TrueTest™ system: TrueTest AVI testing software together with a ProMetric® imaging colorimeter.
The imaging colorimeter AVI testing techniques which are outlined in this article can be applied to several display technologies and can be utilized for a range of displays (LCD, LED, OLED).
These techniques allow display defects to not only be identified, but also classified by cause, by supplying rapid, repeatable measurements which are correlated to human visual perception, and by being able to numerically characterize them.
This enables a consistent measurement of displays in manufacturing applications and permits automated determination of pass/fail in accordance with user-defined criteria. Crucially, this also enables automated determination of remedial action (e.g., rework or scrap).
1. “Methods for measuring display effect as correlated to human perception.” For more on imaging colorimetry as applied to display defect detection, see: H. Kostal, G. Pedeville, and R. Rykowski, SPIE Electronic Imaging Conf., (2009).
2. “The Spatial Standard Observer: A new tool for display metrology.” For more on JND analysis of display mura, see: A.B. Watson, Information Display, 23(1), (2007).
3. “Imaging Colorimetry: Accuracy in Display and Light Source Metrology.” For more on basic imaging colorimetry, see: R. Rykowski and H. Kostal, Photonics Handbook, (2008).
This information has been sourced, reviewed and adapted from materials provided by Radiant Vision Systems.
For more information on this source, please visit Radiant Vision Systems.