手機高光譜影像應用於顯示器光輻射危害評估

Abstract

在本研究中，我們設計並實現了兩種基於智慧型手機的高光譜成像系統：第一種結合電控平移台，構建推掃式高光譜成像架構；第二種基於深度學習技術，利用 RGB 影像重建高光譜影像，無需外接分光模組。此外，我們開發了名為 VisionSpec 的瀏覽器應用程式，可在行動裝置上即時操作與分析高光譜數據。通過電子顯示器光譜檢測案例研究，我們量化了可見光進入視網膜的輻射劑量，並定義累積劑量指標以評估光暴露對視覺健康的影響。實驗結果顯示，部分電子顯示器的在低劑量光輻射下連續長時間使用 (約6 小時)可能超過安全閾值，對視網膜造成潛在損傷。本研究結果反映出光危害評估的重要性，且與使用者使用情況較為相關，應考慮使用時間以及環境光條件。

In this study, we designed and implemented two smartphone-based hyperspectral imaging systems. The first system integrates an electronically controlled translation stage to construct a push-broom hyperspectral imaging framework. The second system employs deep learning techniques to reconstruct hyperspectral images from RGB images, eliminating the need for external dispersive modules. Additionally, we developed a browser-based application named VisionSpec, which enables real-time manipulation and analysis of hyperspectral data on mobile devices. Through a case study of spectral measurements on electronic displays, we quantified the radiative dose of visible light reaching the retina and introduced a cumulative dose metric to assess the impact of light exposure on visual health. The experimental results indicate that under low-dose light radiation, prolonged exposure (within six hours) to certain electronic displays may exceed safety thresholds, posing potential risks of retinal damage. These findings highlight the significance of light hazard evaluation, which is closely related to user habits. Factors such as usage duration and ambient light conditions should be carefully considered in risk assessments.