LED illuminant based on statistical color quality metrics

Abstract

The color rendition properties of the illuminant are quantified by a set of novel Statistical Color Quality Metrics (SCQM), which measure the color rendition properties not only by the ability of the light source to reveal the true colors of objects, but also by the ability to saturate, desaturate, and distort colors of objects. The corresponding metrics are the Color Fidelity Index (CFI), Color Saturation Index (CSI), Color Dulling Index (CDI) and Hue Distortion Index (HDI), respectively. Based on these figures of merit, a variety of color LEDs and phosphor-converted white LEDs are used to constitute desired SPDs of white lights. The constituted white lights are also able to work together with other light sources for achieving various lighting tasks.

This work also investigated spectral engineering of narrow-band LEDs. In contrast to conventional color LEDs with a spectral width of 30nm, recently developed research narrow-line phosphor color LEDs have the spectral width of 7nm. These new light sources enable new features for LED illuminants. We demonstrate that using a narrow bandwidth green phosphor with the peak wavelength closely aligned with the peak in the human eye sensitivity significantly improves the Light Efficacy of Radiation (LER) for Red-Green-Blue (RGB) emitters. Compared to the traditional RGB sources, the improvement in LER of 35 to 50 lm/W can be achieved. The color rendition property of light estimated by the Color Rendering Index (CRI) also significantly improves. Combining the narrow band green phosphor with a conventional wide band green phosphor allows for trading off these improvements against the deviation from the Planckian locus. The light sources with the narrow band green phosphor could have considerably better energy efficiency and therefore beneficial for high intensity illumination (such as street lighting).

To seek wider range of application of tunable LED illuminant, an ambient light augmenting theory has been developed. Equipped with Statistical Color Quality Metrics, this theory provides procedures to dynamically integrate the ambient light with LED light to tailor the overall illumination for optimal quality. As a result, lighting tasks are achieved with lower energy cost, better color quality and tunable color rendition properties. Based on this theory, a camera flashlight and a table lamp have been designed. They both learn ambient lighting conditions and generate white light with optimal SPD according to different lighting applications.

A human factor research has been conducted to validate the Statistical Metrics and the power of Color Rendition Engine. During the visual experiment, observers from Chinese and U.S. cultural backgrounds were invited to recognize “natural” lighting as well as preferred lighting with objects illuminated by the Color Rendition Engine. The result shows that the recognized “natural” lighting well with the Color Fidelity Index, while the preferred lighting is correlated with observers’ cultural background, objects familiarity to the observers and the content of the objects. Those recognized and preferred SPDs were recorded by quantification of Statistical Color Quality Metrics. Proper illuminants were designed by representing these quantifications.

This thesis is devoted to an intelligent control of white LED Spectral Power Distribution (SPD). LED illuminants have been designed to meet different lighting requirements including luminance level, Correlated Color Temperature (CCT) and color rendition properties.

The first design of the illuminant was a white light source with continuously tuned color rendition properties (called the Color Rendition Engine (CRE)). This source was composed of four color LED channels: Red, Amber, Green and Blue. Varying the four spectral components, the constituted white light had the tunable ability to saturate and desaturate color objects as well as the ability to reveal the color fidelity, while the total illuminance level and CCT were kept constant.