Entropic image segmentation of sessile drops over patterned acetate

Abstract

An entropic segmentation method is presented and applied to the contour detection of water drops over patterned surfaces. Jensen–Shannon divergence is computed from a double sliding window in the image to get a real number matrix, in which a region growing procedure is performed in a similar way to usual watershed. Then a region merging process is achieved, and the optimal configuration is selected to obtain the complete drop contour. Once the drop contour is detected from top-view images, the contact angle might be readily computed from the area enclosed by the contour and the drop volume.From the results obtained the proposed method outperforms other usual methods, like watershed or Canny.