主成分分析を用いた天空要素の分光放射輝度の推定法 ESTIMATION OF THE SPECTRAL RADIANCE OF A SKY ELEMENT USING PRINCIPAL COMPONENT ANALYSIS

この論文にアクセスする

この論文をさがす

著者

    • 八塚 秀樹 YATSUZUKA Hideki
    • 京都大学大学院工学研究科建築学専攻 修士課程 Graduate Student, Dept. of Architecture, Graduate School of Eng., Kyoto University
    • 上谷 芳昭 UETANI Yoshiaki
    • 京都大学大学院工学研究科建築学専攻 准教授・工博 Assoc. Prof., Dept. of Architecture, Graduate School of Eng., Kyoto University, Dr. Eng.

抄録

In the previous paper<sup>7)</sup>, we have developed a method of measuring the spectral radiance of all sky elements using a general purpose digital camera and a circular fisheye lens. In this method, the fisheye camera measured the XYZ tristimulus values of each sky element. The XYZ tristimulus values were transferred into the spectral radiance in two ways: (1) by using the CIE daylight illuminant and (2) by using the eigenvectors obtained by principal component analysis of spectral radiance of sky elements measured by a spectroradiometer. The method (1) was mostly accurate, but was sometimes unable to estimate spectral radiance of blue skies because the CIE daylight illuminant was defined with the correlated color temperature being within the range of approximately 4,000 to 25,000Kelvin. The correlated color temperature of the method (2) was unlimited, but the accuracy was lower than the method (1). In this study, to improve the method (2), we obtained eigenvectors by principal component analysis of normalized spectral radiance of sky elements. The XYZ tristimulus values were transferred into the xy chromaticity coordinates, and the relative spectral power distribution was computed. Then it was transferred into the spectral radiance by tristimulus values <i>Y</i> [cd/m<sup>2</sup>]. We found the improved method (2) was able to estimate spectral radiances more accurately than the original method (2).

収録刊行物

  • 日本建築学会環境系論文集

    日本建築学会環境系論文集 79(697), 227-232, 2014

    日本建築学会

各種コード

  • NII論文ID(NAID)
    130004937102
  • NII書誌ID(NCID)
    AA11830377
  • 本文言語コード
    JPN
  • ISSN
    1348-0685
  • NDL 記事登録ID
    025327528
  • NDL 請求記号
    Z74-D484
  • データ提供元
    NDL  J-STAGE 
ページトップへ