Approximate extraction of late-time returns via morphological component analysis.

Area of Science:

• Acoustic data processing
• Signal separation
• Sonar imaging

Background:

• Separating complex acoustic signals into distinct components is a fundamental challenge.
• Measured time series are often additive mixtures of signals and noise, posing an ill-posed inverse problem.
• In active sonar, distinguishing early-time returns (object geometry) from late-time returns (wave coupling) is crucial.

Purpose of the Study:

• To compare two morphological component analysis (MCA) models for separating early-time and late-time acoustic responses.
• To evaluate the effectiveness of MCA in separating signals without time-gating.
• To assess the compatibility of MCA-based signal separation with sonar image reconstruction.

Main Methods:

• Utilized morphological component analysis (MCA) framework.
• Employed short-duration and long-duration responses as proxies for early-time and late-time returns.
• Applied methods to both analytic data (Stanton's elastic cylinder model) and experimental data (in-air circular synthetic aperture sonar).

Main Results:

• MCA successfully separated early and late-time responses in both analytic and experimental datasets without time-gating.
• The separation process was shown to be compatible with image reconstruction, forming separated time series into imagery.
• Optimal separation was achieved using a computationally intensive frame-based model, while a faster Fourier transform-based method offered competitive results.

Conclusions:

• MCA provides a viable method for separating early and late-time acoustic signals in sonar applications.
• The technique enhances the ability to differentiate between geometric and wave coupling effects in elastic object sensing.
• MCA facilitates improved sonar image reconstruction by enabling robust signal separation.