• No results found

A Dynamic Surface Reconstruction Framework for Large Unstructured Point Sets

N/A
N/A
Protected

Academic year: 2022

Share "A Dynamic Surface Reconstruction Framework for Large Unstructured Point Sets"

Copied!
10
0
0

Laster.... (Se fulltekst nå)

Fulltekst

(1)

A Dynamic Surface Reconstruction Framework for Large Unstructured Point Sets

R´ emi All` egre, Rapha¨ elle Chaine, Samir Akkouche

LIRIS CNRS / Universit´e Lyon 1, France

Snapshots of selective reconstructions

Symposium on Point-Based Graphics 2006

(2)

Bimba

(1.9M points→31k points)

(3)

Bimba

(1.9M points→58k points)

(4)

Asian Dragon

(3.6M points→185k points)

(5)

David

(3.6M points→175k points)

initial reconstruction

(6)

David

(3.6M points→175k points)

local refinement

(7)

Thai Statue

(5M points→572k points)

(8)

Lucy

(14M points→551k points)

out-of-core reconstruction

(9)

Lucy

(14M points→551k points)

out-of-core reconstruction

(10)

St. Matthew

(26M points→1.2M points)

out-of-core reconstruction

Referanser

RELATERTE DOKUMENTER

The local update is performed in 11 seconds using our accelerated selective reconstruction algorithm... The local update is performed in 11 seconds using our accelerated

We tested the rendering performance for three different cases: for the case that all points were rendered as point sprites, the hybrid rendering approach that uses textured splats

Implicit surface fitting methods based on signed distance functions are very sensitive to outliers (shot noise), since the shortest distance from all input data points is computed..

Using the Voronoi diagram of the input point set, we deduce a tensor field whose principal axes and eccentricities locally represent respectively the most likely direction of the

The streaming surface reconstruction algorithm alternates the incremental Delaunay triangulation of the loaded data, a Delaunay refinement process that splits Delau- nay

Finding locations of the moving object that contact the offset surface only at a single point is essentially the prob- lem of dynamic collision detection.. In dynamic collision

In this section we will first briefly review the kd-tree based neighbor search and then present how to take advantage of the spatial coherence of the queries using our novel

Fi- nally, in order to reconstruct structured shape and animation of the subject from video, we present a dense 3D correspondence finding method that enables spatio- temporally