“…Nevertheless, local or point signals are not sufficient to characterize the entire evolution of a cluster, as the cluster varies with time and its shape is hardly spherical, showing streamline, arch, horseshoeshape, and so on. 2,3 Visualization methods, such as the videography with optical-fiber micrograph probe, 23 high-speed videography, [24][25][26][27] and laser sheet technique, [28][29][30] have been widely employed for cluster analysis since the 1990s. In those methods, clusters are normally distinguished from the other phases by a grayscale threshold, such as the grayscale intensity of the practical boundary of the cluster, 25 the mean grayscale intensity of the whole image, 26 and the local mean grayscale intensity corrected by n-times its standard deviation.…”