Recovering Inner Slices of Layered Translucent Objects by Multi-Frequency Illumination

Area of Science:

• Computer Vision
• Optics
• Image Processing

Background:

• Translucent objects exhibit complex light scattering, making it difficult to visualize internal structures.
• The appearance of layered translucent objects is a summation of blurred appearances from each layer, influenced by depth-dependent point spread functions (PSFs).

Purpose of the Study:

• To develop a computational method for recovering the appearance of individual inner slices within translucent objects.
• To address challenges posed by non-uniform transmission and material inhomogeneity in translucent materials.

Main Methods:

• A multi-frequency illumination approach utilizing varying spatial frequencies of checker-pattern illumination.
• Exploiting differences in low-pass characteristics of depth-dependent PSFs for layer separation.
• Estimating pixel-wise PSFs and assuming spatial smoothness for accurate inner slice recovery.

Main Results:

• Successful recovery of individual inner slice appearances from layered translucent objects.
• Demonstrated robustness to non-uniform transmission effects.
• Quantitative validation through simulations and qualitative assessment on real-world scenes.

Conclusions:

• The proposed multi-frequency illumination method effectively recovers inner slice appearances of translucent objects.
• The technique provides a reliable way to visualize internal details, overcoming challenges of light scattering and non-uniformity.