Fast computational depth segmentation using orthogonal fringe patterns without pattern sequence changing | JoVE Visualize

Fast computational depth segmentation using orthogonal fringe patterns without pattern sequence changing

Yu Xiao, Wenzhong Han, Xuejing Zhang

Journal of the Optical Society of America. A, Optics, Image Science, and Vision|April 2, 2021

View abstract on PubMed

Summary

A new fast computational depth segmentation (FCDS) method improves speed and reduces complexity. It uses fewer patterns and phase singularity points for precise object segmentation, even with abrupt depth changes.

Area of Science:

Optical metrology
Computer vision
Image processing

Background:

Omnidirectional depth segmentation methods (ODSM) offer robustness but suffer from slow speeds and high computational costs due to multiple fringe patterns and sequence changes.
Traditional methods often struggle with segmentation speed and computational complexity, limiting practical applications.

Purpose of the Study:

To propose a fast computational depth segmentation (FCDS) method that enhances segmentation speed and reduces computational complexity compared to existing ODSM.
To address limitations of previous methods by utilizing phase singularity points and a modified Fourier transform algorithm (MFTA).

Main Methods:

The FCDS method employs only five fringe patterns without sequence changes, utilizing phase singularity points as markers for depth segmentation.
A modified Fourier transform algorithm (MFTA) calculates wrapped phase sequences using orthogonal phase-shifting fringe patterns and a DC component pattern.
An optimization algorithm refines depth segmenting lines to handle objects with abrupt depth changes and prevent mis-segmentation.

Main Results:

The FCDS method successfully segments objects at different depths into isolated regions without pattern sequence alteration.
It achieves precise depth determination by extracting segmenting lines along orthogonal directions, overcoming phase insensitivity and reducing calculation costs.
Experimental results show a 120% increase in segmentation speed compared to previous methods, with successful segmentation of similarly colored objects.

Conclusions:

The proposed FCDS method offers significant improvements in speed and computational efficiency for omnidirectional depth segmentation.
It effectively segments objects with varying depths, including those with abrupt changes and similar colors, demonstrating high precision and practical viability.