Phase unwrapping following a multi-seed short-path constructed by undirected graphs

Area of Science:

• Image Processing
• Computational Imaging
• Signal Processing

Background:

• Traditional phase unwrapping algorithms face efficiency limitations.
• Path-dependent methods can be computationally intensive and slow.

Purpose of the Study:

• To develop a faster and more accurate phase unwrapping algorithm.
• To improve the computational efficiency of phase unwrapping techniques.

Main Methods:

• A graph-traversal-based path algorithm is proposed.
• The wrapped phase map is modeled as a pixel-level undirected graph.
• Reliable seed points are identified using an auxiliary map and gradient consistency.
• Block-wise parallel unwrapping is achieved through locally constructed paths.
• A pixel-wise fusion strategy ensures global phase continuity and consistency.

Main Results:

• The proposed method achieves fast and accurate phase unwrapping.
• Computational speed is improved by nearly an order of magnitude compared to established algorithms.
• The algorithm demonstrates comparable accuracy to existing methods.
• Scalability and robustness are inherent features of the new approach.

Conclusions:

• The graph-traversal algorithm offers an effective solution for phase unwrapping.
• It significantly enhances computational efficiency without sacrificing accuracy.
• The method provides a robust and scalable approach for constructing unwrapping paths.