Just Noticeable Distortion Profile for Flat-Shaded 3D Mesh Surfaces

Area of Science:

• Computer Graphics
• Human-Computer Interaction
• Perceptual Computing

Background:

• 3D meshes often experience geometric distortions due to lossy operations like compression and transmission.
• Evaluating the visibility of these distortions is crucial as human users interact with 3D models.

Purpose of the Study:

• To develop a model for computing the Just Noticeable Distortion (JND) profile for flat-shaded 3D meshes.
• To quantify the perceptual visibility of vertex displacement in 3D models.

Main Methods:

• Investigated human visual system properties, including contrast sensitivity and masking, for flat-shaded surfaces.
• Conducted psychophysical experiments to determine human detection thresholds for vertex position changes.
• Defined local perceptual properties on 3D mesh surfaces.

Main Results:

• Successfully computed a JND profile for flat-shaded 3D meshes based on experimental data.
• The JND model quantifies the minimum detectable vertex displacement.
• Demonstrated the model's application in guiding 3D mesh simplification and quantization.

Conclusions:

• The proposed JND model accurately reflects human perception of geometric distortions in 3D meshes.
• The JND profile is valuable for optimizing 3D mesh processing tasks, ensuring visual quality.
• This research provides a perceptual basis for handling geometric distortions in 3D graphics.