Jeudi 25 Octobre 2007
Orateur : Pablo Cancela
Titre : Ultrasound image segmentation with shape priors: application to automatic cattle rib-eye area estimation.
Résumé : Image segmentation in low SNR ultrasound (US) images requires a priori infromation to manage the lack of information in several zones produced by acquisition conditions. Rib-eye segmentation in US images is an example in which image information combined with rib-eye shape priors permits to achieve the task. Automatic segmentation is done by fitting a curve based on US image information and the geodesic distance between the evolving curve and the estimated mean rib-eye shape in a shape space. The mean shape and validation of the method is done based on a big database of tracings made by experts. Ongoing work is being done on determining the qualitiy of the US images as well as of automatic measures.
 |
CMLA UMR 8536 |
 |