Search

A new approach to image segmentation based on entropic region growing and merging, which is useful in drop shape analysis, is presented in this paper. The procedure works in three steps. First, a normalized divergence matrix is obtained which gives the likelihood of being a boundary pixel for each pixel in the image. Second, a region growing algorithm is carried out on the divergence matrix, keeping a record of boundaries between adjacent regions. Third, some regions are merged by following a combined entropic criterion, based on both the divergences of the matrix along the common boundary and the global divergence between two adjacent regions. The final contour is adapted by a dynamical spline fitting. This general purpose algorithm is presented here applied to drop shape analysis.

The rough material can improve the friction, the adhesion, as well as reduce the wetting which were characterised by Brownian surface. The work is to analyse the possible relationship between different classical roughness parameters, the maximum surface amplitude, the Hurst exponent, the lacunarity. The topographies were obtained using the atomic force microscopy.