摘要: In image processing, models are used to improve robustness of algorithms by introducing a priori knowledge. Deformable models, frequently in the field medical images, described means energy functionals with data attachment terms and regularising terms. The express constraints relating expected shapes. shape blood vessel segment 3D images obtained Magnetic Resonance Imaging or helicoidal Computed Tomography is often implicitly generalised cylinder model, i.e. an association axis (vessel centreline) surface boundary). this context, involve, for candidate points, measure likelihood being located on centreline boundary. Such can use reflecting low-level local photometrical properties brightness patterns. This presentation will give overview recently be illustrated authors' contribution.