Study On Methodology For Spectral Optimization Design And Color Quality Assessment Based On Multi-channel LED Light Sources

With the rapid development of light-emitting diode(LED)related technology,LED light sources have been widely used in daily life,industrial production as well as scientific research.LED,an electroluminescent solid-state light source,has the advantages of environmental friendliness,long lifetime,and high efficiency over traditional sources.In particular,its tunable spectral property and convenient dimming control offer tremendous possibilities to meet the needs of high color quality and diversity for light sources,which tends to be realized with multi-channel LED light sources.Accordingly,a comprehensive and in-depth study was carried out in this dissertation to investigate the spectral optimization design method and color quality assessment of light sources involved in these needs.The spectral optimization design method based on multi-channel LED light sources is mainly used to realize the reproduction of the target spectrum,i.e.spectral matching,and the optimal lighting quality of light sources,i.e.spectral optimization.The color quality assessment of light sources mainly involves the two aspects of the discussion of color quality evaluation methods and the modeling of perception-related color quality assessment.For spectral matching,a method of matching the target spectrum and illuminance by multi-channel LED light sources was proposed based on the differential evolution(DE)algorithm and photometric principle,while combining the cubic spline interpolation algorithm to solve the problem that the LED spectrum may shift nonlinearly with the control signal.Through a practical multi-channel LED array,the spectral power distributions of Commission Internationale de l'Eclairage(CIE)standard illuminants A and D65 were matched by the proposed method at the two illumination levels of 500 lx and 1000 lx,respectively,verifying the effectiveness of this method.Also,the DE algorithm has excellent global search capability compared to other direct search optimization algorithms.In order to further upgrade the spectral quality of the matched spectrum of D65,an improved method suitable for matching CIE daylight illuminants was developed through combining the metric for spectral matching error with the visible range metamerism index.The improved method was tested with four CIE daylight illuminants of D50,D55,D65,and D75 indicating that it can guarantee the spectral matching accuracy while significantly improving the quality grade of visible range metamerism index as well as the color quality of the matched spectra.With regard to the applications that the ideal spectral matching accuracy cannot be achieved through the actual multi-channel LED light sources while high requirements are still demanded for the lighting quality,taking a color quality evaluation index of light sources as the optimization target,a spectral optimization method under the photometric and colorimetric constraint(SOUPCC)composed of a specific chromaticity(or correlated color temperature CCT)and illuminance for multi-channel LED light sources was proposed in this study based on a constrained differential evolution algorithm.Again,by means of a multi-channel LED array,the spectral optimization with the CIE general color rendering index Ra as the optimization target was performed respectively under 16 groups of colorimetric constraints composed of 2 illumination levels of 500 lx and 1000 lx and 8 CCTs ranging from 2800 K to 6500 K.The results indicate that the optimized spectra present excellent color rendering performance under the condition of meeting the target colorimetric constraints,and that the adopted optimization algorithm also shows satisfactory global search potential in color quality optimization over other direct search algorithms.Considering that the color quality evaluation method of light sources(TM-30-15)issued by the Illuminating Engineering Society of North America(IES)incorporates the advantages of many color quality evaluation methods and has won widespread support,with the IES color fidelity index(IES Rf)and luminous efficacy of radiation(LER)as the optimization target,a two-index optimization method for multi-channel LED light sources was proposed utilizing a multi-objective optimization algorithm at a specific chromaticity(or CCT).This method can be employed to determine the optimal spectral parameters of individual channels and to test the performance of the IES Rf.Focusing on the four-channel LED light sources,the competitive relationships between LER and IESRf were revealed through optimizing the peak wavelengths,spectral bandwidths,and relative intensities of individual channels in the CCT range from 2800 K to 6500 K,which would not only provide theoretical and data support for balancing the color rendering performance and energy efficiency in the spectral optimization design of multi-channel LED light sources,but also verify the effectiveness of the improvement of IES Rf over CIE Ra.According to this optimization result,a set of optimal spectral parameters were determined for CCT-tunable four-channel LED light sources,including the optimal peak wavelengths and spectral bandwidths of individual channels.These spectral parameters produce excellent color rendering performance(IESRf values reach 94,all those of CIE Ra are about 96)and relative high energy efficiency(almost all the LER values are larger than 300 lm/W),which can provide theoretical guidance for developing new-type LED chips and choosing the chips for multi-channel LED light sources.Accordingly,this indicates that the IES Rf has good compatibility with the CIE Ra,and further validates that the improvement of IES Rf over CIE Ra is effective.Then,considering that the spectral shift and binning issue always exist in practical LED chips,the impact of the optimal peak wavelength shift of each channel on the IES Rf optimization score was analyzed in this dissertation.Besides,the optimization of the IES Rf was performed again by the proposed SOUPCC method with the practical multi-channel LED array covering the optimal peak wavelengths of individual channels,further examining the reliability of the conclusion drawn from the optimization of four-channel LED light sources,and validating the effectiveness of the SOUPCC method.In view of the fact that the calculation of CCT for characterizing the color appearance of light sources is usually a key step in implementing color quality evaluation methods for light sources,a CCT calculation method with high accuracy for a wide range of chromaticity was proposed based on the golden section search algorithm combining a small-scale look-up table.A large number of chromaticities were selected in the range of 1000 K<CCT<20000 K and-0.03<Duv 0.03 to test the proposed method,indicating that this method obtains excellent accuracy with the maximum calculation error being only 0.3 K,independent of the Duv values of light sources.On the basis of accurately calculating the CCTs of light sources,19 typical color quality evaluation metrics for light sources were investigated quantitatively by evaluating different types of illuminants and light sources through the Pearson correlation coefficient and cluster analysis method.The results indicate that there exist some correlations among most metrics,and that the combination of the color fidelity based method and color gamut based one could rather effectively characterize the color quality of light sources.Accordingly,an attempt was made to develop a comprehensive color quality index(CQI)based on the principal component analysis.The test of the proposed index shows its relatively high correlation with most metrics,verifying the feasibility of CQI.Based on the psychophysical method of categorical judgment,an immersive visual experiment of evaluating the perception-related color quality of light source was designed and conducted to further test the performance of the existing typical color quality metrics for light sources as well as the proposed comprehensive CQI in characterizing the subjective assessment.In the experiment,with the multi-channel LED array,the SOUPCC method was employed to optimally design 41 metameric spectra at each of the 4 target colorimetric parameters composed of the illumination level of 450 lx and 4 CCTs(2800 K,4000 K,5000 K,6500 K),and the common object colors were evaluated by 10 observers with normal color vision,in terms of naturalness,colorfulness,and preference according to their overall color appearance when rendered by the light source corresponding to each spectrum.Based on the visual data,the variance analysis verifies that C,CT has no significant effect on the observer judgment about the color quality attributes of light sources,and the correlation among different color quality attributes of light sources was discussed by means of Pearson correlation coefficient.Besides,based on Pearson correlation coefficient,the correlations between the 21 color quality metrics including CQI and these subjective evaluation data were also investigated in detail,confirming their limited characterization performance.Then with the multidimensional scaling analysis,the underlying relationship of these metrics and the subjective evaluation data was explored in depth,demonstrating that almost all metrics can correspond to one attribute of naturalness,colorfulness,and preference,and that the saturation level is identified as a critical factor affecting the color quality evaluation of light sources.On the basis of these analyses,a model for predicting the color quality of light sources was constructed utilizing a multiple nonlinear regression equation in this study.Compared with the aforementioned 21 metrics,this model can better characterize the subjective evaluation results,while being applicable to different CCTs.In the meantime,the effectiveness of the proposed model was further confirmed by the verification experiment.Finally,the research contents and achievements in this dissertation are summarized comprehensively,and the direction of future work is also forecasted.