Quantification of Specular Image Distinction in Avionics
Active Matrix Liquid Crystal Display (AMLCD) Applications
Peter N. Wyckoff, Steven D. Ellersick
Boeing Commercial Airplane Group, MS 02-JH, Box 3707, Seattle, WA98124-2207
Flight Crew Operations Integration: Display Optics & Lighting Systems
ABSTRACT
With the proliferation of liquid crystal displays as the primary instrumentation interface on aircraft flight decks, it becomes important to quantify and control the unique optical parameters, associated with AMLCDs, that affect readability in a critical environment. It has been discovered that image distinctness, in addition to reflected energy intensity, can be a significant detriment to display readability. Consequently, efforts are being made to establish measurement procedures and quantify distinctness-of-image to help control the detrimental effects of this characteristic. Two measurement methods will be discussed that are currently being considered for use in describing the degree of specular distinctness-of-image reflections on avionics AMLCDs.
Keywords: Specular Reflection, Reflected Image, Distinctness, Clarity, AMLCD, Avionics
- BACKGROUND
Most of the instrumentation in today’s modern commercial aircraft is presented to the crew by means of computer screens. As the flexibility and reliability to present all flight information to the crew electronically surpassed that of electromechanical dials and gauges, the flight deck panels became a mosaic of computer screens. Hence, the term “glass cockpit” came to appropriately describe this transition. First came CRTs (cathode ray tubes) with new presentations of flight information using color and dynamic symbology enhancing the value and readability of the information presented to the crew. Then, about ten years ago, LCDs began replacing CRTs because of their promise of lighter weight, smaller volume, and lower power consumption.
By looking at the pros and cons associated with CRT versus LCD optics in the flight environment, an interesting optical trade-off becomes apparent. The lack of diffuse reflectance, relative to the avionics CRT and its phosphors, enables the LCD daytime optical performance to be superior when exposed to direct and bright indirect scattered sunlight. However, because of the LCD’s composition of smooth surfaces, it is quite a specular reflector also. This means that under conditions where any object is sufficiently bright and within the field of view of the LCD, a strong virtual image forms which competes with or partially obscures the information presented on the display. In contrast, CRT symbology washes out under direct sunlight illumination but so do specular reflections (with the exception of a cover glass surface with very low reflectance).
During implementation of AMLCDs as primary flight displays, first on the Boeing 777 and subsequently on the Boeing 737, it was determined that there is an acceptable limit to the degree of specular reflection allowed.1 In addition, recent experience with the Boeing 767 has established there is an acceptable limit to the degree of clarity, or distinctness-of-image, of the virtual images produced.
From the inception of LCD applications in aerospace, it has always been known that specular reflection should be controlled i.e. the amount of energy allowed to exit the display surface as a result of light incident thereon.2 It was also known that under the extreme sunlight conditions encountered in the flight deck environment, any degree of diffusivity or haze could mask the display information being portrayed (contrast reduction and color desaturation).3 Hence, LCDs used in aircraft make use of antireflective coatings on smooth surfaces versus the treatments used on desktop displays. However, the specular reflectance metrics in place did not yield any degree of discernment between two LCD constructs - one being acceptable and the other being unacceptable with respect to distracting reflections in the flight deck environment. After some investigation, the difference between the displays was not the specular reflective energies but rather the degree of distinctness of the images reflected off each of the displays.
The standard practice for quantifying specular reflectance off of a typical aircraft LCD is described in SAE ARP 4260, MIL-C-14806, VESA FPDMS as well as in many of the supplier and manufacturer documents. Basically, there are two means of controlling specular reflections and they are both related to exitant energy as a function of incident energy. The first test procedure deals with total photopic reflectance off a given test display. The second procedure is concerned not only with the proportional exitant energy off a test display but with its associated wavelength. Wavelengths in the middle of the visible spectrum are specified with lower reflectance limits.
Both of the two LCD constructs, mentioned above, were subjected to standard specular reflectance procedures. Figure 1 shows the test results for both photopic and spectroreflective measurements. Both displays are comparable, within measurement tolerances, with respect to total photopic reflectance, and they are within acceptable limits. It can be seen that display ‘B’, which proved unsatisfactory, reflects too much energy at the lower wavelengths and suggests the problem was mostly one of reflective coating color control. Yet, another display was submitted with the reflective spectrum being redistributed to better meet the spectroreflective requirements, especially bringing down the reflective energy in the lower wavelengths. However, once the LCD was evaluated in flight, it was still unacceptable to the crew. This led to consideration of the remaining difference between the displays: reflected image distinction.
Of course, if there is a problem of reflected image distinction, the age-old solution is to ‘diffuse’ or soften the edge distinction of the image by some optical means. The solutions of adding some texturing to the first surface of the display as well as splitting the image have been observed on commercial LCD applications. Both of these solutions have been found satisfactory for reducing the impact of distinctness-of-image upon display readability. The optimum amount of scattering for the reflected specular images and reflected image splitting is subjective, hence the topic of this paper.9
There have been several metrics and test methodologies established for quantifying the degree of surface diffusivity. Methods range from taking ratios of energy off the test surface versus a known standard to energy ratios through a combed shutter device.4, 6 All these methods are based on attempting to define the degree of diffusivity through the scattering of reflected energy. They say nothing about how the exitant energy is distributed. Furthermore, if the solution to softening the distinctness-of-image effects is done through image splitting, the above methods are not sensitive enough to sufficiently categorize and control the degree of splitting that is satisfactory.
Lately, the recognition that nonspecular surfaces need to be further characterized have been addressed in Kelley and Becker.5, 7 More emphasis is put on the distribution of exitant energy off of the reflective surface as a means of characterizing the surface with more accuracy. The concept of the BRDF is presented as a good metric for characterizing reflectance and can be categorized in terms of its specular, haze (semi-diffuse-specular), and diffuse components.5 Recent instrumentation using video conoscopic photometry has allowed the collection of BRDFs with relative ease and in little time. Because of this, it looks plausible to specify and control the texturization of surfaces to a fine degree based on reflectance distribution profiles.
Quantification of non-specular surface reflectivity is highly desired in the case of flight deck LCD applications because the window between being too specular (distinctness-of-image) and too diffuse (veiling & indirect glare) is narrow when dealing with critical readability issues under strong sunlight environments. The attempt to correct specularity could easily turn into an “over-attempt” resulting in glaring diffusivity; something for which LCDs are generally seen as a solution.
Good quantification of specular surface reflectivity with image splitting capability is also desired in order to provide enough image break-up to bring down the distinctness of image to a satisfactory level. The BRDF approach cannot address this need since the surface reflectance distribution may not be very different from a pure specular surface. Another approach is needed to quantify a surface in this category along with a metric that enables the tester to determine that the category of reflectance is, indeed, specular or non-specular.
Total Photopic Specular Reflectance @15°
Display A / Display B
.65% / .56%
Total Photopic Specular Reflectance @30°
.61% / .50%
Figure 1 – Spectral and Total Photopic Reflectance of Flight Deck AMLCDs at 15° and 30° Incident Angles.
- Proposed Quantification
The best way to address the quantification of minimally acceptable image splitting is to specify an imaging pattern that is sensitive to image location and to the direction and degree of splitting. The pattern should then be quantifiable by typical photometric instrumentation. A variable frequency, high contrast, bar pattern can be used to locate the degree of splitting which is desired by being reflected off of the test surface. The virtual pattern image will be split into primary and secondary images, 180° out of phase, that actually alter the pattern contrast as the split secondary image overlays the adjacent bar to the primary image (see Figure 2). Energy lost off the primary image to the secondary image also alters the primary image. The overall effect is one of reduced contrast from that of a nonsplit image and creates a fuzzy image. The degree of splitting can be specified by the frequency that should be filled by the image. The strength of the image split (in terms of reflected energy divided between the secondary images) can be specified by a maximum contrast threshold. All of this is measurable by a photometer at the appropriate viewing angle.
Figure 2 – Variable Ruling and Offset Superposition of Multiple Images.
A hazed surface will inhibit the ability to view more detailed resolution and, hence, a test can be devised to determine whether the surface is hazed or specular by depending on a minimum resolution threshold of a reflected image. The variable frequency, high contrast pattern can be reflected in the display surface and the category of the surface ascertained by where the resolution of the pattern is lost. If the surface is determined to be specular, then any further investigation into distinctness-of-image control shall be addressed under the image splitting approach.
If the surface is determined to be hazed, then obtaining the BRDF profile of the reflection off of the surface at predetermined angles should yield information as to the degree of haze that is present. This test would also give information as to whether the surface treatment involved is too diffuse or too specular. The essential metric is a region of allowable distribution away from the specular angle that will result in acceptable image spread without creating unwanted glare. The degree of deviation away from the specular angle of the reflected energy gives a BRDF profile. As mentioned in Becker,7 the use of a digital conoscopic-measuring device enables a BRDF profile to be obtained quickly and easily. Figure 3 shows examples of BRDF profiles obtained from glass surfaces of varying degrees of specularity. Dispersion of reflected energy can be used as a related characteristic that helps to define acceptable hazing. The resulting effect on distinctness-of-image is very sensitive to the degree of haze. With some calibration, conoscopic distributions can also be analyzed for luminance levels.
Figure 3 – Reflective dispersion about the specular angle off various glass surfaces.
Additionally, control of non-specular reflection can be further checked by measuring the remaining types of reflection namely specular and diffuse via the accepted methods.6, 8 Although haze is used on LCD surfaces to reduce image distinction, it is equally important to continue to keep specular and diffuse components below defined threshold reflectances. This will prevent the situation where haze is acceptable but is part of a total BRDF that is too bright or too diffuse.
- TEST PROCEDURES
In order to obtain the three metrics discussed above: surface categorization, degree of image splitting, and haze profiling, the following instrumentation is the most practical. The test procedures, which come after description of the instrumentation, are designed with this in mind.
Figure 4 – Test set-up for measuring degree and strength of image splitting.
Measurement of the degree of image splitting on aircraft LCDs requires using a strong illumination source to put enough luminance on the target to create a measurable image off the display and associated antireflection coating. In addition to the illumination source, a photometer is needed which is capable of measuring luminance down to 6.8 cd/m2. Related apparatus includes a high contrast, variable frequency ruling or a pattern consisting of several of these rulings in various directional orientations, and a means to get to the functional eye point of the application since the degree of image splitting is dependent on view angle.
As is illustrated in Figure 4, the test LCD should be placed in front of the photometer and positioned at the proper view angle for the application involved. The variable ruling should be placed alongside the photometer so that its reflected image falls onto the lens of the photometer for measurement. The illumination source should be placed behind the display and aimed at the variable ruling pattern until a virtual image appears in the LCD of sufficient intensity so as to be measurable with the photometer.
The image of the ruling in the display is observed. The directionality of the image splitting on the display can then be determined and the ruling bars aligned to be perpendicular to the direction of splitting. The image split should fit into a specified minimum frequency ruling, e.g. lines/cm, at the point of measurement in the photometer. If it does not, then the splitting is not sufficient to reduce distinctness-of-image. In other words, find the part of the pattern where the secondary image of the dark bar entirely overlaps the white bar beneath. The previous white bar secondary image should entirely overlap the dark bar making it appear lighter. The contrast ratio of the dark bar to the light bar can be measured. If the contrast ratio is too great then the splitting is not sufficient to reduce distinctness-of-image. This procedure is repeated at various parts of the display.
For determining which distinctness-of-image test to use, the display should first be categorized into the area of either a specular or a semi-specular surface. The variable, high contrast ruling should be placed in front of the LCD with a strong illumination source behind the display. The illumination source should be placed so that a visible image of the pattern appears in the surface of the display. If any frequency higher than a specified maximum, e.g. 8 cycles/degree, can be distinguished, then the display should be classified as specular. If the display is semi-specular, then the following apparatus and procedures should be used for determining distinctness-of-image control.
As is illustrated in Figure 5, the test LCD should be placed in front of an apparatus built to determine incremental solid angle of reflected energy around the principal specular angle of reflection. An incremental surface area is illuminated with a source of fairly collimated white light at 15° and 45° incident angle ensuring that the area being measured is contained entirely within the beam of the incident light. The lamp or the photocell is moved around incremental angles, at a fixed distance, above the test surface and around the peak reflectance until reflectance becomes insignificant. Enough data point density is taken to describe a ‘solid’ of light distribution centered on the specular path or peak reflectance. The use of a video conoscope makes this procedure easy and quick but sacrifices in accuracy must be known.
Figure 5 – Concept of analyzing semi-diffuse or semi-specular reflection.
As the display surface becomes increasingly diffuse, the BRDF profile of a reflected beam changes from one extreme of a specular ‘delta’ function to that of a complete flat response over angles (Lambertian character). The metric for a semi-specular surface must specify the maximum and minimum acceptable haze components in the BRDF profile which yields a display with enough spread to soften distinctness-of-image but not create a diffuse veiling glare problem. These haze components could be specified in terms of a luminance function over angle from the peak component or as a frequency function based on transforming the luminance profile.
As previously shown in Figure 3, various BRDF profiles for key surface treatments have been obtained. The sharp profile of the black glass standard is as expected for a very shiny surface. We already know that a surface of this type are what got us into the problem of image distinction in the first place and will not be acceptable for use on aircraft LCDs. The next profile is from the surface texturization that is currently approved and flying on some Boeing aircraft LCDs. As can be seen, this profile contains some significant flares at its extreme angles that are identified as the haze component. The last profile is from an even more diffuse glass treatment that is evident from the thickening and lengthening of the profile flares in comparison to the previous BRDF profile. By the time the diffusivity of the profile becomes this significant, it has been determined that the veiling glare problem already makes this degree of dispersion unacceptable for the application.
At this point, we only know the center Figure 3 BRDF profile to be acceptable and so will start with specifying a tolerance around that profile’s haze component (Figure 6). The tolerance will be specified with upper and lower bound equations in and . If the specification is uniform around about the specular path of inclination, s, then the equations can be written only as a function of . The haze reflectance bounds can then be stated as follows: