利用頻譜分析影像流速之探討

Abstract

近年來，台灣發生土石流災害頻率日益增加且造成許多地區人員傷亡及財產損失。為了要減少土石流災害造成的影響，所以需要更加了解土石流發生的運動行為。本研究利用頻譜分析發展一套現地影像流場的速度量測法。該方法考慮流速與波速之物理差異，並針對流體表面因光線照射與流動速度的不同，以進行其明暗變化訊號之分析。 透過影像灰階化後，其灰階值為流體運動隨空間變化之資訊，而將流場空間資訊視為一複合波形，並透過傅立葉轉換至頻率域作訊號分析，其分析方式採用波浪理論中的延散關係式，擷取較慢波速之深水波訊號，且以傅立葉反轉換至空7間域進行波形之速度分析。 本研究之結果顯示，該方法能解決以往影像量測法缺少可分析質點之缺陷，而其演算法較為迅速；但其波形判識仍需利用人工分析處理，且流速分析結果與雷達波流速儀相比之平均誤差約為26%，希望未來針對其誤差來源作修正。

In recent years, debris-flow hazards occur frequently in Taiwan and cause casualties and property loss in many residential areas. In order to minimize the effect of such hazard, it requires more understanding of debris-flow’s behavior. The research is aim to develop an innovative image velocimetry, using spectrum analysis in the flow field. We took both celerity and water velocity into account to analyze the signal reflected from the water surface. After the images were transformed to gray level, the gray value represents the information of fluid motion. We first consider the spatial information of flow field as composite waveform and then transfer it to the frequency domain by using Fourier transformation. Afterwards, the dispersion relationship of wave theory was adopted to analyze the signal. For doing the velocity analysis, we first capture the signal of deep water wave and applied the inverse Fourier transformation to turn it into spatial domain. In this research, the issue existing in current image velocimetry that the flow field lacks of analyzable particle can be improved. However, there is still a room for waveform identification. The average error is 26% comparing to the results obtained via the instrument.