Differentiations of Image Analysis on Bubble Characteristics during Subcooled Flow Boiling of Seawater Using Various Automatic Algorithms

The present study analyzes the images of bubbles during subcooled flow boiling of 3.5 wt% artificial seawater using a free software Image J. The general strategies include collections of raw images from videos by a high-speed camera, preprocess of raw images, converting to 8-bit images, threshold processes by algorithms, watershed treatments of images, detections, and measurements of bubbles. 4 representative images including the slug-dominant behavior, weaker slug, foam-dominant behavior, and stable foam, are collected for the current datasets. Detections of bubbles depend on thresholding by algorithms and their results are studied after thresholding treatments. Differentiation and similarities of results of bubble image analyses using 17 algorithms in the literature are investigated by direct comparisons of montages and gray values between original images and thresholded images. Their effectiveness on different cases are assessed by the mean absolute errors. The algorithms of Yen, Huang, Li, Otsu, and Shanbhag all demonstrate a good result in catching the original images.