| 摘要: | 在光學與色彩計算領域中，透過反射率轉換至色度空間（如CIEXYZ或CIELAB）雖可得出精確的色彩數值，但實務上，特別是在反射率極低或材質特殊的樣品中，所計算之亮度值Y常低於人眼實際感知，導致顯色結果與觀察者主觀視覺經驗出現落差。針對此問題，本研究從人眼視覺系統對亮度的適應性出發，提出兩種補償策略，分別針對自然光與燈管照明環境下的人眼感知行為進行亮度調整。
在自然光場景中，研究指出，當樣品反射率極低時，僅透過線性轉換所得之 Y 值將導致色彩顯示過暗。根據 CIELAB 模型中 L* = 50 所對應的亮度為 Y=18.42，本研究取其為依據，尋找趨勢，並定義 Y 數值作為辨識彩色的亮度基準。另一方面，在燈管照明條件中，因畫面中存在極亮的燈管倒映，人眼視覺系統會將此處視為基準亮度，並對整體視野亮度壓縮。因此，本研究建立以最亮區域定義 Y=100，並依比例修正的亮度補償模型，以模擬人眼的相對亮度感知行為。
本研究結合實景拍照、反射率量測、CIE色彩轉換與視覺比對，透過亮度補償有效改善了低反射率樣品顯色過暗與感知不一致的問題，提升色彩呈現的視覺準確性，亦為未來進行色彩模擬與人因一致性設計提供應用依據。;In optical and colorimetric applications, converting surface reflectance to colorimetric values using models such as CIEXYZ or CIELAB enables objective color computation. However, in practice—especially for low-reflectance or optically complex materials—the calculated luminance value (Y) often underrepresents the brightness perceived by the human eye, resulting in color displays that appear darker than expected. To address this discrepancy, this study proposes two luminance compensation strategies based on human visual adaptation mechanisms, targeting both natural illumination and fluorescent lighting conditions.
Under natural lighting, low-reflectance samples tend to yield extremely low Y values, often displaying as near-black despite being perceptually colorful. Referring to the CIELAB color space, where a lightness value of L* = 50 represents perceptual mid-gray, the corresponding Y value is approximately 18.42. Based on this, we define Y for visible color, and adopt it as a luminance compensation baseline for natural viewing conditions.
In fluorescent lighting environments, intense specular reflections (e.g., tube light reflections) cause high-brightness regions in the visual field, leading the human visual system to adapt by normalizing the brightest area to the perceptual maximum. Accordingly, we propose a second strategy that sets Y = 100 for the brightest sample and linearly adjusts other samples proportionally, modeling the brightness compression experienced by the human eye under such conditions.
By combining photographic simulation, reflectance measurement, color conversion via CIE models, and perceptual color patch comparison, the proposed compensation methods significantly improve visual accuracy for low-luminance samples. The findings offer a practical basis for color correction techniques that align better with human visual perception, with potential applications in display calibration, optical coating analysis, and perceptual color rendering. |
