Abstract:We describe surface area measurements based on local estimates of isosurfaces originating from a marching cubes representation. We show how improved precision and accuracy are obtained by optimizing the area contribution for one of the cases in this representation. The computations are performed on large sets (approximately 200,000 3D objects) of computer generated spheres, cubes, and cylinders. The synthetic objects are generated over a continuous range of sizes with randomized alignment in the digitization g… Show more
“…It also allows a comparison of the various methods for surface estimation. As shown in Ziegel and Kiderlen (2009), the maximum asymptotic relative error for general sets X is 12.8 % for the weights suggested in Lindblad and Nyström (2002) The approach based on the Crofton formulae has a number of advantages. First, the method of surface area estimation can simply be extended to arbitrary dimensions and to most of the other intrinsic volumes, in particular the integral of the mean curvature which measures e.g.…”
Section: Discussionmentioning
confidence: 99%
“…The probably most intuitive method for measuring the surface area is based on rendering data, see, e.g., Lindblad and Nyström (2002), where the areas of the surface patches serve as weights for computing the surface area from local knowledge, i.e., from the numbers of pixel configurations. However, this type of estimator is not multigrid convergent.…”
Section: Discussionmentioning
confidence: 99%
“…We are starting from the local knowledge represented by the numbers of 2 × 2 × 2 pixel configurations on cubic primitive lattices and ask for the best choice of surface weights. The weights suggested by Lindblad (2005) minimise the estimation variance of the surface area of a plane with random normal direction uniformly distributed on the unit sphere. This idea goes back to Mullikin and Verbeek (1993), see also the discussion in Windreich et al (2003).…”
Section: Discussionmentioning
confidence: 99%
“…Moreover, compared to empirical marching cubes methods like in Windreich et al (2003), only a small number of pixel configurations has to be taken into account. In Schladitz et al (2006) the weights for the surface are compared with those of Lindblad and Nyström (2002), see also Ziegel and Kiderlen (2009) for a sound investigation of surface area estimation. This paper is organised as follows: First we repeat some facts about homogeneous lattices and summarise results on adjacency systems and Euler number estimation from Nagel et al (2000); Ohser et al (2002;; Schladitz et al (2006).…”
The densities of the intrinsic volumes -in 3D the volume density, surface density, the density of the integral of the mean curvature and the density of the Euler number -are a very useful collection of geometric characteristics of random sets. Combining integral and digital geometry we develop a method for efficient and simultaneous calculation of the intrinsic volumes of random sets observed in binary images in arbitrary dimensions. We consider isotropic and reflection invariant Boolean models sampled on homogeneous lattices and compute the expectations of the estimators of the intrinsic volumes. It turns out that the estimator for the surface density is proved to be asymptotically unbiased and thus multigrid convergent for Boolean models with convex grains. The asymptotic bias of the estimators for the densities of the integral of the mean curvature and of the Euler number is assessed for Boolean models of balls of random diameters. Miles formulae with corresponding correction terms are derived for the 3D case.
“…It also allows a comparison of the various methods for surface estimation. As shown in Ziegel and Kiderlen (2009), the maximum asymptotic relative error for general sets X is 12.8 % for the weights suggested in Lindblad and Nyström (2002) The approach based on the Crofton formulae has a number of advantages. First, the method of surface area estimation can simply be extended to arbitrary dimensions and to most of the other intrinsic volumes, in particular the integral of the mean curvature which measures e.g.…”
Section: Discussionmentioning
confidence: 99%
“…The probably most intuitive method for measuring the surface area is based on rendering data, see, e.g., Lindblad and Nyström (2002), where the areas of the surface patches serve as weights for computing the surface area from local knowledge, i.e., from the numbers of pixel configurations. However, this type of estimator is not multigrid convergent.…”
Section: Discussionmentioning
confidence: 99%
“…We are starting from the local knowledge represented by the numbers of 2 × 2 × 2 pixel configurations on cubic primitive lattices and ask for the best choice of surface weights. The weights suggested by Lindblad (2005) minimise the estimation variance of the surface area of a plane with random normal direction uniformly distributed on the unit sphere. This idea goes back to Mullikin and Verbeek (1993), see also the discussion in Windreich et al (2003).…”
Section: Discussionmentioning
confidence: 99%
“…Moreover, compared to empirical marching cubes methods like in Windreich et al (2003), only a small number of pixel configurations has to be taken into account. In Schladitz et al (2006) the weights for the surface are compared with those of Lindblad and Nyström (2002), see also Ziegel and Kiderlen (2009) for a sound investigation of surface area estimation. This paper is organised as follows: First we repeat some facts about homogeneous lattices and summarise results on adjacency systems and Euler number estimation from Nagel et al (2000); Ohser et al (2002;; Schladitz et al (2006).…”
The densities of the intrinsic volumes -in 3D the volume density, surface density, the density of the integral of the mean curvature and the density of the Euler number -are a very useful collection of geometric characteristics of random sets. Combining integral and digital geometry we develop a method for efficient and simultaneous calculation of the intrinsic volumes of random sets observed in binary images in arbitrary dimensions. We consider isotropic and reflection invariant Boolean models sampled on homogeneous lattices and compute the expectations of the estimators of the intrinsic volumes. It turns out that the estimator for the surface density is proved to be asymptotically unbiased and thus multigrid convergent for Boolean models with convex grains. The asymptotic bias of the estimators for the densities of the integral of the mean curvature and of the Euler number is assessed for Boolean models of balls of random diameters. Miles formulae with corresponding correction terms are derived for the 3D case.
“…For an alternative voxel-based surface area estimation method, see e.g. [6]. Table 1 shows the frequency of the nine surface voxel types in a 160 × 200 × 160 segmented white matter MR brain image (in the grey-white matter interface).…”
Section: The Estimator Of Mullikin and Verbeek [14]mentioning
Abstract. Consider a complex, convoluted three dimensional object that has been digitized and is available as a set of voxels. We describe a fast, practical scheme for delineating a region of interest on the surface of the object and estimating its original area. The voxel representation is maintained and no triangulation is carried out. The methods presented rely on a theoretical result of Mullikin and Verbeek, and bridge the gap between their idealized setting and the harsh reality of 3D medical data. Performance evaluation results are provided, and operation on segmented white matter MR brain data is demonstrated.
Spatial correlation of stratigraphic units quantified from geological maps. Comput. Geosci., 4, 515-526. 4 Aguilera, A., Rodríguez, J., and Ayala, D. (2002) Fast connected component labeling algorithm: A non voxel-based approach.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.