Local hull-based surface construction of volumetric data from silhouettes

Area of Science:

• Computer Vision
• Computer Graphics
• Geometric Modeling

Background:

• The Marching Cubes (MC) algorithm is widely used for surface reconstruction from volumetric data.
• MC can produce surface discontinuities, especially with noisy or imperfect input data.
• These discontinuities arise from the sensitivity of surface extraction to noise in volumetric data.

Purpose of the Study:

• To develop a robust surface construction algorithm that overcomes the discontinuity issue in MC.
• To improve the quality of reconstructed surfaces from shape-from-silhouette methods.
• To ensure continuous surface generation even with imperfect volumetric data.

Main Methods:

• A novel algorithm aggregates local surfaces constructed via the 3-D convex hull algorithm.
• The method classifies local convexities from imperfect MC vertices using sliced volumetric data.
• It addresses noise sensitivity inherent in MC surface extraction.

Main Results:

• Continuous surfaces were successfully obtained from imperfect silhouette images.
• The algorithm demonstrated effectiveness for both convex and nonconvex objects.
• The proposed method enhances the reliability of surface reconstruction from noisy volumetric data.

Conclusions:

• The proposed surface construction algorithm effectively generates continuous surfaces.
• It provides a robust alternative to standard MC for imperfect volumetric data.
• This method improves the quality and continuity of reconstructed object surfaces.