Image Segmentation using Active Contours Without Edge (original) (raw)

In this paper, we propose a segmentation method based on an active contour model without edges which, given an input image or image sequence, generates a mask of the desired object(s). This model is then extended to be used in multi-object tracking. Our method allows us to detect objects which are not necessarily delimited by gradient. In other terms, we do not need to detect edges of the object we want to extract from the image to perform our segmentation. To perform this segmentation, we base our model on techniques of curve evolution, the Mumford-Shah functional and level sets. Our problem can be seen as a particular case of the minimal partition problem in minimizing energy domain. The goal of this survey is also to present an interactive and parametrizable software which helps the understanding of the model and allows to test its limits. After presenting our model, we will give the numerical algorithms and the discrete approximations we used. Finally, various experimental results will be presented and commented.