The Parameters Affecting Accuracy of Reflected Color Measurements
Color is the sensation produced when light at visible wavelengths from 380-780 nm interacts with the human eye. Our perception of color is based on the wavelengths of light that are reflected from, absorbed by or emitted by an object. Light reflection can be specular or diffuse. Very smooth surfaces like mirrors or glass exhibit specular reflection, in which the angle of reflection is equal to the angle of incidence. Rough surfaces exhibit diffuse reflection, reflecting light through a wide range of angles.
The color we perceive from objects is mostly diffuse reflection. Many surfaces we perceive as being somewhat shiny or having gloss exhibit a mix of specular and diffuse reflection (Figure 1). When measuring reflection from an object, it is important to consider which type of reflection is dominant, and whether some or all the reflected light needs to be collected. The texture of the sample should be considered to select the best angle for a reflected color measurement.
Figure 1. Samples with very smooth surfaces exhibit specular reflection, although even glossy samples have a mixture of both specular and diffuse reflection.
Why Use a Spectrometer to Measure Color?
Spectrometers provide more information and flexibility than filter-based colorimeters commonly used to measure reflected color. Simple colorimeters use red, green and blue filters in combination with diodes or sensor pixels for measurement. More advanced systems use tristimulus filters that mimic CIE color matching functions. These setups work well for incandescent light sources, but are less accurate for LEDs. Handheld color meters may measure up to 20 wavelength bands, but this is not enough for research or high-accuracy measurements.
To detect small color changes, very high color resolution is necessary. By capturing the complete spectrum, the color measurement made by a spectrometer allows careful and detailed analysis of data. Color meters and analyzers based on filters or detection over specific bands do not provide as much information for the accurate determination of color as spectral measurements.
Also, some color analyzers are strongly dependent on lighting conditions, since objects tend to appear different colors under different illumination. With the optimum lighting, two objects can appear to be identical in color even if their reflected spectral power distributions differ, an effect called metamerism. If the lighting changes, however, the colors can look significantly different. This makes controlled lighting conditions essential to consistent results.
When color measurements are made with a spectrometer, a full reflected or emissive spectrum is the starting point for all calculations. Capturing the complete spectrum allows the data to be analyzed in different ways, and even recalculated later to change the observer, the illuminant, or the color space. This offers a degree of flexibility not available using other measurement methods.
How Do I Optimize a Reflection Measurement for Color?
There are several components to consider when optimizing your spectrometer setup for color measurements:
Light Source: A light source for reflected color needs to have good output from 380-780 nm, the range over which the human eye detects light. The broad, smoothly varying output of a tungsten halogen light source works well, and is economical and versatile for other applications. By comparison, a xenon light source is a poor choice for color due to its jagged, pulsed spectrum and the need for high averaging to get good quality measurements.
Sampling Optics: Either Vis-NIR or UV-Vis optical fibers can be used for color measurements, since both transmit well between 380-780 nm. UV-Vis fibers have slightly better transmission characteristics below ~400 nm, which may be useful if S:N for the system needs to be improved at shorter wavelengths.
If you’re using a reflection probe, remember that the probe will illuminate and detect from the same direction, “seeing” only part of the reflected light or color of the object. This is fine for most samples, but be careful when measuring iridescent samples and highly reflective ones where the color changes with the angle of illumination or viewing. This can happen with rough surfaces like brushed metal, plant seeds and reflective signs or materials.
An integrating sphere might be a better choice for these samples, since it both illuminates and collects light at all angles. An integrating sphere is also good for looking at convex curved surfaces, or to measure the color of objects that are small enough to fit into the sphere. Color measurements made using an integrating sphere with a lower port-to-diameter ratio yield the most accurate results, particularly on the L*a*b* scale. (Measurements made using the ISP-REF tend to show errors of ~5% or more, so this is not the preferred integrating sphere for color.) Also, it is important to consider whether specular reflection (gloss) should be included in your color measurements.
Collimating lenses can be used at the ends of individual fibers to customize the angle of incidence and angle of collection when making reflected color measurements, though the collimating lenses need to be adjusted carefully to avoid beam divergence and get good signal. We usually find that color measurements taken using collimating lenses and fibers are not as accurate as those made using an integrating sphere, so be sure that the extra fixturing and alignment is justified before using this method.
An example reflected color setup is shown in Figure 2.
Spectrometer Configuration: If you’re making color measurements using a tungsten halogen light source for illumination, consider using a UV-Vis grating like Grating #2, which covers 250-800 nm in the spectrometer. Grating #2 peaks at a shorter wavelength than the Vis-NIR grating (Grating #3, which covers 350-850 nm) used in preconfigured Vis-NIR spectrometer units. When combined with a silicon detector and a tungsten light source, the UV-Vis grating produces a spectral response that varies less with wavelength, resulting in slightly better S:N over the important 380-780 nm range. View these grating response curves to better appreciate the response characteristics of gratings typically specified for color measurements.
Color is a slowly varying spectral feature, so high resolution is not really needed to make good measurements. Resolution of ≤2.0 nm (FWHM) should suffice.
What Reflectance Standard Should I Use?
A reflectance system is not complete without a standard for reference. Reflectance measurements are a ratio of the reflected light spectrum to the incident light spectrum. Since there is no way to directly collect all the light incident on a surface, reflectivity is usually measured relative to a reference standard.
The standard chosen should be similar in reflectivity to the sample to keep signal levels about the same during measurements and thereby achieve the best signal to noise ratio (S:N). The WS-1 diffuse reflectance standard is a popular choice, since it is matte white in color and is >98% reflective from 250-1500 nm (and >95% reflective up to 2200 nm). The WS-1-SL can be a good choice when working in the field or in dirty environments, since it can be smoothed, flattened and cleaned if it gets pitted or smudged.
The STAN-SSH high reflectivity specular reference standard is the best choice when measuring very shiny surfaces, but it varies in reflectivity from 85% to 98% over its range of 250-2500 nm. This can be accounted for in OceanView software using the Non-Unity Correction feature. If no correction is applied, OceanView will assume the standard is 100% reflective at all wavelengths, giving distorted data.
At the other extreme, the STAN-SSL low reflectivity specular reflectance standard is best for surfaces with low specular reflectance values like thin film coatings, anti-reflective coatings, blocking filters and substrates. It has just ~4.0% reflectance from 200-2500 nm. It is also possible to purchase calibrated “gray” standards (diffuse reflectance standards at a variety of intermediate values) from our sister company, Labsphere.
Why use the Non-Unity Correction Feature?
Non-unity correction is a feature of OceanView software that compensates for reflection standards not reflecting 100% of the light that strikes them. The default assumption is that the reference standard has 100% reflection across the entire wavelength range, but no standard reflects 100% of the photons that strike it. Working without non-unity correction is fine for qualitative, comparative work, but it will not yield the most accurate quantitative results. You should apply non-unity correction for the most accurate results.
If you know the actual amount of reflectivity for a given reflection standard, you can correct for it. For example, during OceanView installation, standard files for the STAN-SSH, STAN-SSL, WS-1 and WS-1-SL reflection standards are stored on your computer. Those files can be used when applying the non-unity correction.
How Do I Take the Best Dark Measurement?
When taking a dark measurement with a reflection probe or integrating sphere, it is best to block the light at the light source. Turning the light source off and on again will throw the light source out of thermal equilibrium and require a new reference measurement.
Another option is to point the illuminated probe or opening of the integrating sphere into a dark space to take the dark measurement so that no light scatters back in. A background measurement taken this way is more accurate because it includes any scattered reference light that will be present in the sample measurements, allowing it to be subtracted properly.
Resist the urge to point the sampling optic at something black (like a piece of paper or a cover cloth). Objects that appear to absorb all wavelengths usually reflect some colors better than others.
Reflectance and Color
When properly measured, spectral reflectance can yield much of the same information as the eye, but it does so more objectively. Reflectance techniques can measure the color of a sample, or examine differences between objects for sorting or quality control. The samples may be as diverse as automotive parts, paint, coffee beans, dyed human hair or lizards, making it challenging to choose the best setup.
So, what’s the secret? As we’ve described here, the key is to understand how each part of the system works to measure the sample. Once you understand the trade-offs, configuring the optimum system becomes easier. We’d like to help you do that, and give you some tips on how to get the best results for your measurements.
- Measurement Techniques
- Color Setup Options and Trade-offs
- Color Applications and Tips
- Controlling the Color Consistency of Beverages
- Fiber Optic Reflectance Spectroscopy for Pigment Identification
- Determining Color Differences
- How is CIELAB color computed?
- Internal Shutter Improves Accuracy in Color Measurement
- Operation Gumball: The Sweet Science of Spectroscopy
- VIDEO: Perform a Reflected Color Spectroscopy Measurement with OceanView