Normalization of fringe patterns using the bidimensional empirical mode decomposition and the Hilbert transform

Area of Science:

• Optical metrology
• Image processing
• Signal analysis

Background:

• Fringe patterns are crucial in optical metrology but often suffer from noise and illumination variations.
• Bias and modulation non-uniformities can significantly hinder accurate quantitative analysis of fringe patterns.

Purpose of the Study:

• To develop and evaluate a data-driven technique for bias suppression and modulation normalization of fringe patterns.
• To address challenges posed by varying fringe densities and illumination defects in fringe pattern analysis.

Main Methods:

• Utilizes bidimensional empirical mode decomposition (EMD) to decompose fringe patterns into intrinsic frequency modes.
• Employs the partial Hilbert transform to characterize local mode amplitudes for normalization.
• Tests the method using simulated fringe patterns with controlled defects and real-world data.

Main Results:

• The proposed technique demonstrates effective bias suppression and modulation normalization across various fringe densities.
• Successfully handles local variations in modulation caused by illumination defects.
• Validated performance on both simulated and real fringe pattern data.

Conclusions:

• The data-driven approach offers a robust solution for fringe pattern normalization.
• The method shows promise for enhancing the accuracy of quantitative analysis in optical measurement techniques.
• The technique's advantages and limitations are discussed, providing insights for practical applications.