風浪表面因水下流場的渦流結構而形成與風場同向且間距相近的條痕結構,條痕區域的溫度因熱量的傳輸方向而改變,當熱自水體傳輸至空氣時,條痕區域的溫度較周邊低,而於熱圖像形成低溫條痕。本研究發展一影像辨識法,以自動擷取實驗室風浪表面熱圖像的條痕結構,進而分析條痕間距的特性。我們先以「經驗模態分解法」濾除熱圖像中因熱輻射轉換至電訊號過程所產生的短波雜訊,並透過辨識熱圖像中跨流向上的相對低溫點位,運用適當點位連結範圍將點位連結而形成條痕,進一步分析跨流向上條痕間距的分佈統計特性。結果顯示條痕間距之機率密度分佈近似於對數常態分佈,且與無滑移邊界的結果相似;無因次化之平均條痕間距隨風速增大而愈大,然而於無滑移邊界流場之結果則趨於一定值 ((λ^+ ) ̅=100)。
Thermal streaky structures can be observed on wind-wave surface. They are induced by the underlying coherent eddies in parallel with the wind. The temperature in these streaks is lower than that in the surrounding area when the heat flux is upward from the water to the air, and vice versa. Cold streaky structures, therefore, are observed on infrared thermographic images. In this study, an image recognition method is developed to automatically capture these streaky structure on thermographic images of laboratory wind waves. The method of empirical mode decomposition is first applied to filter out the short-length noises in the thermographic images. The local temperature minima in the spanwise direction are then identified. A streak passing a local temperature minimum is formed by connecting the neighboring downstream/upstream local temperature minima within a chosen radius. Spanwise spacings between the neighboring streaks can then be calculated and analyzed. It is found that the probability density distribution of the streak spacing is close to lognormal distribution, similar to the streaks observed next to a no-slip wall. The non-dimensional mean streak spacing based on friction length, however, increases with the friction wind speed. This is different from the flow next to a no-slip wall in which the non-dimensional mean streak spacing approximates 100 friction unit.