Estimation of Light Source Color Rendition with Low-Cost Sensors Using Multilayer Perceptron and Extreme Learning Machine

Botero J. -S.; Valencia Aguirre J.; Vargas Bonilla J. -F.

首页> 外文期刊>Leukos: The journal of the Illuminating Engineering Society of North America >Estimation of Light Source Color Rendition with Low-Cost Sensors Using Multilayer Perceptron and Extreme Learning Machine

【24h】

Estimation of Light Source Color Rendition with Low-Cost Sensors Using Multilayer Perceptron and Extreme Learning Machine

机译：Estimation of Light Source Color Rendition with Low-Cost Sensors Using Multilayer Perceptron and Extreme Learning Machine

获取原文

获取原文并翻译 | 示例

掌桥外文数据库（机构版） >>

开具论文收录证明 >>

文献代查 >>

页面导航

摘要
著录项
相关主题

摘要

Artificial lighting was an engine of economic progress since its development allowed to extend human activities at night and to spaces where natural light did not reach. Thenceforth, artificial electric lighting has constantly evolved, in principle oriented toward improving energy efficiency, and recently giving more importance to improving color reproduction for the observer. In that sense, it has been necessary to constantly improve the understanding of the observer model and consequently enhance the measures that allow to characterize the illumination sources according to their ability to reproduce the colors. Currently, the most used measures are those derived from TM 30-18 (R-f, R-g) and the CIE CRI (R-a). On the other hand, the development of controllable LED sources and the integration of environments with mixed lighting (daylight-artificial) have opened a new field of research aimed at controlling LED sources, thinking about energy efficiency and maintaining controlled the chromatic reproduction conditions. For this reason, in this paper is proposed an R-f, R-g and R-a estimation model, using artificial neural networks with different architectures, and where RGB sensor signals or low-cost multiband signals are used as inputs. The estimation was made by training and evaluating 120 MLP and ELM architectures, and the training set was constructed from combinations of the SPDs available in TM 30-18, finally obtaining more than 300,000 spectra. The absolute error for the best performing MLP was less than 1% and the correlation between the actual and estimated measures was close to 99%. The model found as a result of the estimation in this work, can be implemented in a low-cost microcontroller unit (MCU) to be integrated into an intelligent control system, an loT or Edge system.

著录项

来源
《Leukos: The journal of the Illuminating Engineering Society of North America》 |2021年第4期|280-290|共11页
作者
Botero J. -S.; Valencia Aguirre J.; Vargas Bonilla J. -F.;
展开▼
作者单位

Inst Tecnol Metropolitano, Fac Ingn, Grp AEyCC, Carrera 31 54-10, Medellin, Colombia;

Univ Antioquia UdeA, Fac Ingn, Grp SISTEMIC, Medellin, Colombia;

展开▼
收录信息
原文格式 PDF
正文语种英语
中图分类
关键词
Color render index (CRI); extreme learning machine; multilayer perceptron; gamut area (GA); photodetectors; RGB sensors; TM-30;

Estimation of Light Source Color Rendition with Low-Cost Sensors Using Multilayer Perceptron and Extreme Learning Machine

摘要

著录项

相关主题

期刊订阅