SUBPIXEL EDGE DETECTION IN THE PROBLEM OF DETERMINING THE SURFACE TENSION FROM AN AXISYMMETRIC DROP PROFILE

摘要

We consider the determination of surface tension from the image of the profile of a sessile or pendant drop. The accuracy of drop edge detection is investigated and accuracy-enhancing methods are proposed, in particular, compensation of systematic errors associated with edge blurring and curvature. The proposed edge detector is virtually free from systematic error and can be used also in other problems. A convenient practical method has been developed for calibrating the camera scale with the aid of calibration objects of simple fixed shape, such as a disk or a rectangle. The method is based on overlapping the calibration object with the image. We also consider the stability of surface tension reconstruction in the presence of outliers, e.g., outliers due to the overlapping of the drop contour by a foreign object. An iterative algorithm is proposed for automatic outlier exclusion, which improves the accuracy of determination of the drop parameter.