New Methodology For Evaluating Color Rendering Indexes

• Main Authors: Yen-Hsiang Chao, 趙彥翔
• Other Authors: Hung-Shing Chen
• Format: Others
• Language: en_US
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/67743987595324361413

碩士 === 國立臺灣科技大學 === 應用科技研究所 === 101 === The aim of this research was to evaluate variety types of color rendering characteristics of light sources using color rendering models, such as CIE-CRI, CRI-CAM02UCS (nCRI) and CQS. A psychophysical experiment was conducted to compare between the subjective assessment and physical quantity. Furthermore, their performance was analyzed and compared via statistic methods. Three major color rendering algorithms were implemented. Those three methods are based on color fidelity, which concept has been adopted in CIE-CRI for many years. A psychophysical experiment was conducted. Overall, 14 widely used commercial light sources (e.g. CIE-F11, CIE-F2) including white LEDs with different correlated color temperature and D 65 daylight simulators where evaluated. Each of 10 normal color observers was asked to visually assess the color shift of one of the 25 test samples between a ‘Test’ source and ‘Reference’ source generated by 16-band LEDs which has the close chromaticity match to the test illuminants following the CIE-CRI method. The test-samples were well distributed in CIELAB space and were selected from the NCS color atlas. The results showed that all four indices gave similar performance (i.e. correlation coefficients (r) of CIE-CRI, CRI-CAM02UCS, CQS, nCRI are 0.92, 0.88, 0.84 and 0.85 respectively). In this study we also evaluated the performance among UCSs (i.e., CIELAB, CIEUVW, CIEDE2000 and CAM02-UCS), the results showed CAM02-UCS performed slightly better than the other UCSs. With the using of chromatic adaptation transformation CAT02, the performance improved a bit for all UCSs. Attempts were also made to investigate a fixed reference illuminant in chromatic adaptation model (i.e., CAT02). It was found that 4000K gave a best fit to the visual data, but its performance is nowhere close to the present method to calculate the ideal reference illuminant from the Plankian and CIE illuminants.