Motion-based enhancement of optical imaging

Optical imaging offers the possibility to study and image biology at molecular scale in small animal with applications such as oncology or gene expression studies. New device enable cinematic acquisitions but in some cases signals could suffer from weak spatial information. The aim of this paper is to improve localization of the luminescent data for a freely moving animal using motion information obtained with an additional camera. To this end, we propose an approach that performs a non-rigid registration in the scene video and use the corresponding motion vectors to filter the optical one. Due to the lack of contrast and texture, our method introduces silhouette constraints and landmarks on the mouse skin within a variation framework. Motion is represented using thin-plate-splines and the objective function is optimized using a gradient conjugate descent approach. Experimental results demonstrate the potentials of the proposed framework.