Résumé
Among the issues raised by Computer Vision, the tracking of one or several regions of interest is a major consideration. Achieving accuracy and robustness is critical in Visual Servoing and more generally in vision-based applications because of the sensibility of the algorithms. To address those issues in the specific case of MIS heart surgery, we developed a texture-based approach that provide good results on experimental beating heart images. Our approach consists in integrating textural characterization of the region of interest to reinforce the tracking procedure based on classical approaches such as correlation. We show that an appropriate combination of image descriptors can improve the efficiency of the tracking, especially on low-contrasted medical images.