Using Image Processing Technique to Analyze Surface Streaky Structure on Numerically Simulated Wind-Driven Turbulent Flow

• Main Authors: Hong-Yu Liu, 劉鴻煜
• Other Authors: Wu-Ting Tsai
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/zw3af2

碩士 === 國立臺灣大學 === 工程科學及海洋工程學研究所 === 106 === In this thesis, the characteristics of streamwisely elongated streaky structures of numerical wind-driven free-surface turbulent shear flow are analyzed using image processing techniques. At first, adaptive histogram equalization and gamma correction are employed to enhance the contrast of image of streamwise velocity distribution. To segment the higher velocity regions, Otsu’s method is applied to automatically determine the threshold of image, which separates the image into higher and lower velocity regions. A streak detection method is developed to identify streaks automatically from the higher velocity regions. Streak spacing and the corresponding distributions are then derived after the streak identifications. The result shows that the mean streak spacing decreases as the wind speed increases. By the analysis of chi-square goodness-of-fit test, it shows that, comparing to the other chosen theoretical probability distributions, lognormal distribution is closer to the distributions of streak spacing, which is similar to that of the no-slip wall turbulent boundary layer.