UCLA

A new image segmentation algorithm that uses the simulation of a charged fluid is developed. Conceptually, a charged fluid consists of charged elements, each of which exerts a repelling electric force on the others. The charged fluid behaves like a liquid such that it flows through and around different obstacles. The boundary of the segmented object is determined by the image gradient, which is modeled as potential wells that stop the propagating front. The simulation is evolved in two steps that are governed by Poisson's equation. The first step distributes the elements of the charged fluid along the propagating interface until an electrostatic equilibrium is achieved. The second step advances the propagating front of the charged fluid such that it deforms into a new shape in response to the equilibrium electric potential and the image potential. The procedure is repeated until the propagating front resides on the boundary of objects being segmented. The electric potential of the simulated system is rapidly calculated using the finite-size particle (FSP) method implemented via the fast Fourier transform (FFT) algorithm. Experimental results using phantom images, photographic pictures, and medical images demonstrate the utility of this new algorithm in a wide variety of image processing applications. (C) 2006 SPIE and IS&T.