Acoustic Imaging and Source Localization Techniques

Acoustic imaging and source localization encompass a range of methods designed to reconstruct the spatial distribution of sound sources and their radiated fields. Central to these approaches is the use of microphone arrays, where signals recorded by multiple sensors are processed to infer source positions and strengths. Conventional beamforming steers an array’s sensitivity pattern to focus on specific directions, yielding a map of acoustic intensity but suffering from limited resolution and sidelobe artefacts. Deconvolution algorithms refine these maps by suppressing spurious lobes and improving dynamic range. Near-field acoustic holography extends spatial reconstruction to pressure and particle velocity fields close to vibrating structures, often employing the equivalent source method to model complex geometries. Recent advances exploit non-synchronous measurements, synthetic apertures and volumetric mapping to extend bandwidth and three-dimensional coverage. Emerging inverse schemes solve Helmholtz equations under realistic boundary conditions via finite-element or Bayesian frameworks, and machine-learning architectures have begun to incorporate physical constraints to enhance reconstruction fidelity. Together, these techniques have found application in aeroacoustics, underwater acoustics, industrial noise control and structural health monitoring, delivering higher resolution, wider frequency ranges and more robust performance in challenging environments.

Research from Nature Portfolio

Recent studies have refined wideband near-field acoustic holography through an adaptive equivalent source method. By combining iterative weighted and steepest-descent algorithms below and above a transition frequency, respectively, the approach achieves superior reconstruction accuracy across a broad frequency spectrum. The method also applies principal-component truncation to input data, enhancing stability and reducing computational cost, with experimental validation confirming improved performance over traditional regularization schemes.

Research from all publishers

Researchers have demonstrated volumetric source mapping by merging asynchronous microphone array recordings. By combining cross-spectral matrices from non-synchronous measurements, the technique suppresses sidelobes and refines main-lobe resolution in both two- and three-dimensional configurations, enabling accurate mapping of distributed and discrete sources such as aerodynamic surfaces and multirotor drones.

A physics-informed neural network has been introduced for near-field acoustic holography, embedding the Kirchhoff–Helmholtz integral within a convolutional autoencoder loss function. This integration of forward propagation physics with data-driven reconstruction yields significant improvements in estimated velocity fields and pressure distributions on vibrating structures compared with state-of-the-art methods.

Non-synchronous measurements have also been extended to three-dimensional beamforming. By virtually enveloping the sound field with a moving array, the method increases effective aperture and frequency range, reducing the number of microphones required while maintaining high spatial resolution. Simulations and experiments confirm that both orthogonal and non-orthogonal motion trajectories deliver enhanced 3D localisation performance over stationary arrays.

Acoustic Imaging and Source Localization Techniques publication trend

The graph below shows the total number of articles in acoustic imaging and source localization techniques across all publications each year (not limited to Nature Index journals).

Technical terms

Beamforming: A signal-processing technique that steers the sensitivity of an array towards desired directions to map acoustic intensity.

Near-field acoustic holography: A method for reconstructing pressure and velocity fields close to sources, often using inverse algorithms.

Equivalent source method: A modelling strategy that represents complex sound sources by a set of discrete, mathematically defined sources.

Cross-spectral matrix (CSM): The frequency-domain covariance matrix of sensor signals, used to characterise source strength and coherence.

Point spread function (PSF): The response pattern of an imaging system to a point source, which determines resolution and localization accuracy.

Non-synchronous measurements: Sequential recordings at different array positions to create a synthetic aperture and enhance resolution.

References

1. A refined wideband acoustical holography based on equivalent source method. Scientific Reports (2017).
2. Combining asynchronous microphone array measurements for enhanced acoustic imaging and volumetric source mapping. Applied Acoustics (2021).
3. A Physics-Informed Neural Network Approach for Nearfield Acoustic Holography. Sensors (2021).
4. Achieving 3D Beamforming by Non-Synchronous Microphone Array Measurements. Sensors (2020).
5. Inverse Scheme for Acoustic Source Localization using Microphone Measurements and Finite Element Simulations. Acta Acustica (2018).
6. A High-Resolution and Low-Frequency Acoustic Beamforming Based on Bayesian Inference and Non-Synchronous Measurements. IEEE Access (2020).