Toward Automated Detection of Oil Drops in Images of Multiphase Dispersions

A. Rojas Domínguez and G. Corkidi(Mexico)


Image segmentation, Circle detection, Hough transform, Fast radial symmetry, drop-distribution, multiphase dispersion.


The performance of three different circle-detection techniques is compared in the context of automated detection of oil drops in images obtained from a stirring tank containing a simple multiphase dispersion. The techniques compared are: the generalized Hough transform for detection of circles, the Fast Radial Symmetry operator proposed by Loyd & Zelinski, and a Template Matching procedure based on gradient orientation. A ground truth set formed by 1459 oil drops extracted from 241 images was used to compare the performance of the mentioned methods. The results reported indicate that Fast Radial Symmetry is superior in the task.

Important Links:

Go Back