Review and comparative study of three local based active contours optimizers for image segmentation

In this paper, we present a review of three local based formulations of the active contours model (ACM) as a mean of image segmentation. In real world images, especially in the field of medical images such as Magnetic Resonance Imaging (MRI), Computed Tomography (CT) etc. intensity inhomogeneity and noise are two major concerns that make segmentation a very challenging task. In order to assess the robustness of each method to noise and intensity inhomogeneity, we start by intensity inhomogeneity and noise free synthetic image samples, and apply different levels of added inhomogeneity varying from small to severe intensity, or randomly distributed additive gaussian noise with gradually varying variance. The parameters of the energy formulation of each method are initially tuned for the clean sample images and then kept constant for all of the experiments. Our goal is to assess how robust each method is, to still overcome the added noise and/or intensity inhomogeneity and produce the desired segmentation.