Adaptive estimation of normals and surface area for discrete 3-D objects: application to snow binary data from X-ray...

Area of Science:

• Computer Vision
• Computational Geometry
• Image Analysis

Background:

• Accurate normal vector field estimation is crucial for 3D rendering and image measurements.
• Existing methods struggle with shapes exhibiting edges, leading to inaccuracies.
• Discrete object boundary analysis presents challenges for precise geometric computation.

Purpose of the Study:

• To develop a simple and accurate computational method for estimating normal vector fields on discrete 3D object boundaries.
• To address limitations of current algorithms in handling shapes with edges and varying convexity.
• To enable precise surface area derivation from discrete objects.

Main Methods:

• Utilizes gradient vector field analysis of the object's distance map.
• Employs adaptive filtering of the vector field using angle and symmetry criteria.
• Preserves object details while smoothing digitization artifacts.

Main Results:

• Achieved accurate normal vector field estimation across diverse shapes, including those with edges.
• Developed a projection method for direct surface area calculation from discrete objects.
• Demonstrated validity in the continuous limit with simulated and real-world data (X-ray tomography).

Conclusions:

• The proposed method offers a robust solution for normal vector field estimation on discrete 3D objects.
• It overcomes limitations of existing techniques, particularly for complex shapes.
• Enables more accurate surface area measurements and enhances 3D image analysis.