Active contour models, and in particular snakes, have proven to be very effective tools for image segmentation. An active contour model is essentially a curve that evolves from an initial position (user-specified) toward the boundary of an object within an image. The popularity of this semiautomatic approach may be attributed to its ability to take advantage of a priori knowledge and user interaction in the segmentation process.

In biology, and more precisely in the cell segmentation context, multiple targets are needed to be segmented. The first approach is to use a standard snake to segment every single target separately. Unfortunately, this approach bring unsuccessful results in heavily crowded scenes. The proposed solution lies in perform a joint optimization and to include contextual constraints.

In this project the student will implement in Java a multiple-target snake and study study the impact of different couplings in terms of efficiency and performance.