建築物外牆立面熱影像分析

紅外線熱影像法（Infrared Thermography）是以面為單位進行大範圍的檢測，且檢測時不需要直接接觸到受測物的表面，可以進行遠距離非接觸性的檢測，適合運用於建築物立面檢測。當紅外線熱影像法在白天進行檢測時，由於陽光照射角度與物體外型與材料特性等因素，可能產生強烈反射，導致拍攝的熱影像中，呈現不規則的受測表面溫度分布，甚至影響檢測結果的判讀。為了解反光與表面放射率之間的關聯性，本研究利用有限元素軟體ANSYS針對大樓牆面玻璃與磁磚兩種材質之放射率進行加熱模擬，發現磁磚的放射率調升時，表面溫度有下降的情形，另外在玻璃的放射率調降時，表面溫度有上升的情形，由此推論在拍攝大樓外牆的熱影像時，因為玻璃與磁磚的放射率有所不同，以及拍攝角度的差異，可能引發反射之現象，使得熱影像圖中磁磚與玻璃的溫差增大，而影響牆面磁磚缺陷區域之辨識。進一步利用紅外線熱像儀拍攝大樓外牆的一系列熱影像圖，再運用主成分分析法(Principal component analysis, PCA)進行分析，提取每張熱影像圖中的特徵，重新組成特徵影像圖，為了使特徵影像圖與紅外線熱影像圖的資訊統一，將特徵影像圖中數值經過尺度轉換為假溫度值(Pseudo Temperature)，並與實驗熱影像圖進行溫度比較，由結果得知窗戶反光區域之溫度有明顯下降趨勢，說明主成分分析法可以改善反光對紅外線熱影像的影響。

關鍵字

紅外線熱影像法；放射率；反光；主成分分析法

並列摘要

Infrared thermography is an efficient tool for surface temperature measurement. Subsurface defects can be found based on the images produced by an infrared camera. When the Infrared thermography is used during the daytime detection, it is often subject to the sunlight illumination angle and the surface geometry and emissivity. The thermal image captured often displays irregular surface temperature distribution due to reflection or glare effects. Numerical simulations are applied to investigate the relation between the surface temperature distribution and the emissivity of the glass and tiles on the exterior wall of the building. The results show that the surface temperature is lowered as the emissivity of the tile is increased. Furthermore the surface temperature is increased when the emissivity of the tile is increased. It is inferred that the reflection may be caused by the difference in the emissivity of the glass and the tile, and the difference in the shooting angle. The temperature difference between the tile and the glass in the thermal image is increased, and the identification of the defect area of the wall tile is affected. Furthermore, a series of images were recorded so as to investigate the transient variation of temperature distribution of the exterior surface of a building. Principal Component Analysis (PCA) was applied to extract the features in each thermal image and to reconstruct the feature images. Pseudo temperature scale was obtained so as to compare the featured image to the original thermal images. The temperature distribution in part of the reflective area has a significant decreasing trend. The results show that the effect of reflection on thermal images can be reduced.