Accès libre

Quantitative Comparison of Deformable Models in Range Segmentation

À propos de cet article


In this paper we segment range images by applying three deformable models (the classical Active Contour, the adaptive active contour and the Level Set method). These three methods are used to segment images with planar and curved surface scenes. Then the numerical results obtained are compared in order to find the best technique of deformable models adapted to segmentation of range images. Despite of the good experimental results on simple objects, we have noted that the adaptive and classical snake methods have a few limitations and cannot detect discontinuities in curvatures and some items do not always converge. However, the level set method is very efficient for segmenting range images with curvature and complex forms.

4 fois par an
Sujets de la revue:
Computer Sciences, Information Technology