| Summary: | 碩士 === 國立臺灣科技大學 === 光電工程研究所 === 103 === There are many problems which do not comply with the visual evaluation when making color rendering index predictions to modern light sources (e.g. LEDs) characterized with narrow band by applying the current CIE color rendering index method. For this reason, several researchers have proposed various methods to predict the color rendering index including four aspects in color fidelity, color preferences, color discrimination and color harmony respectively. For the different characteristics of light source a different procedure will be used.
This research mainly based on color fidelity. Aim to evaluate the performance of 8 different color rendering indices (CIE-Ra, CRI-CAM02UCS, CRI2012, CQS, GAI, MCRI, FCI_CIELAB, and FCI_CIECAM02). Through psychophysical experiments to investigate the correlation between color rendering index and visual evaluation.We apply two sets of Telelumen® light source system as the test light source and the reference light source respectively. Each system has 16 narrow band LEDS that can be adjusted to produce a variety of spectral power distributions (SPD). We obtain 12 test light sources via mixing up 4 different CCTs plus color rendering property (CIE-Ra) in high, middle and low levels. Both test light source and reference light source have the same CCTs and reference light sources have much higher CIE-Ra (>95). All light sources are fixed at a luminance of 100 cd/m2.
By changing of the 12 different pairs of the light sources, we conducted the psychophysical experiment to derive the visual color differences of 27 pairs test color samples. Then we average the visual color differences of 27 pairs test color samples as visual evaluation to the test light sources. And proceed the correlation coefficient analysis of visual evaluation and 8 kinds of color rendering indices to compare the color fidelity performance. The results show that the color rendering index (CRI-CAM02UCS, CIE-Ra and CRI2012) with color fidelity as evaluating target have a significant best performance (r = 0.98) and CQS (r = 0.93) followed. The others are not satisfied. This result indicates that CIE-Ra is still an accurate color rendering index for color fidelity.
|
|---|
