Phase retrieval with only a nonnegativity constraint

Area of Science:

• Signal Processing
• Image Reconstruction
• Fourier Optics

Background:

• Reconstructing objects from Fourier transform data is crucial in various imaging fields.
• Traditional methods often rely on support constraints, limiting applicability.
• Noise in Fourier intensity data poses a significant challenge.

Purpose of the Study:

• To investigate the feasibility of reconstructing 2D objects using only nonnegativity constraints.
• To determine if object reconstruction is possible without explicit support constraints.
• To assess the impact of noise on reconstruction accuracy.

Main Methods:

• Utilized the magnitude of the Fourier transform of a real-valued, nonnegative 2D object.
• Applied a nonnegativity constraint during the reconstruction process.
• Introduced varying levels of noise to the Fourier intensity data.

Main Results:

• Demonstrated successful reconstruction of 2D objects solely based on nonnegativity.
• Showed that support constraints are not always necessary for reconstruction.
• Confirmed that the method is robust even with significant noise in the data.

Conclusions:

• Object reconstruction from Fourier magnitude is achievable with nonnegativity constraints alone.
• This approach offers a viable alternative to traditional methods requiring support information.
• The technique shows promise for applications dealing with noisy Fourier intensity measurements.