Understanding Acoustic Cameras: Effective Use, Limitations and Common Misconceptions
Acoustic cameras, widely known for their ability to visualize sound sources using beamforming algorithms, have significantly impacted various fields from environmental noise monitoring to industrial diagnostics. However, despite their capabilities, these devices are often misused or pushed beyond their intended limits, leading to inaccurate results and misconceptions. This blog post outlines the main dos and don'ts of using acoustic cameras for solving problems in the audible domain, emphasizes their limitations, as well as introduces viable alternatives for certain applications.
Historical Perspective
In the 1970s, multichannel microphone arrays were first applied to sound source localization, although the concept dates back to World War I. The first microphone antenna, or "acoustic telescope," was invented by Billingsley in 1974. Since then, the use of multichannel products has grown substantially with improvements in data acquisition systems, computing hardware, and localization algorithms. Since 1999, devices known as "acoustic cameras" have been marketed as solutions for detecting and localizing noise sources in a sound field, with Gfai being the first company to introduce such a commercial tool.
How does an acoustic camera work?
Acoustic cameras use combinations of acoustic signals captured from different sensors to localize sound sources. The so-called beamforming techniques allow them to have a highly directive response, effectively steering a "virtual microphone" across the testing environment. This process enables the system to virtually listen to different areas and identify noise sources by amplifying sound from specific directions while suppressing noise from others. The resulting noise maps are computed almost in real time and often overlaid on a video image, enabling the visualization of the noise emitted by various sound sources.
Ideal Conditions for Acoustic Cameras
Most acoustic cameras are designed based on a theoretical framework primarily focused on localizing far-field point sources in free-field conditions. This means that ideally the sound sources should be omnidirectional, sufficiently far from the array, and have a wavelength much smaller than the array size. Under these conditions, an acoustic camera can accurately detect and quantify sound sources. The technology works best when measurements are conducted in environments that closely match the free-field assumption, such as outdoor facilities or anechoic test chambers.
DO
- Use acoustic cameras for outdoor measurements where scattered sound sources are far from the array.
- Focus on mid to high-frequency ranges where the wavelength is much smaller than the array size.
- Utilize the technology to localize stationary or transient noise sources with low directivity.
DON'T
- Use acoustic cameras for near-field measurements or indoor settings without considering significant limitations.
- Expect accurate results for low-frequency sound sources due to the large wavelength relative to the array size.
- Rely on acoustic cameras for quantifying complex sources distributed across a large area and/or with high directivity.
Challenges with Low-Frequency Sound
A critical limitation of acoustic cameras is their difficulty in resolving low-frequency sounds. The ability to focus on different directions is highly dependent on the ratio between the array size and the wavelength. Low-frequency sounds, which have longer wavelengths, do not allow for the strong directivity needed for accurate localization, especially with portable microphone arrays, resulting in poor spatial resolution and very low dynamic range. While large microphone arrays can somewhat mitigate these issues, they remain limited by the physical principles governing their operation.
Technological Limitations
- Aperture Size: The size of the acoustic camera’s aperture affects its angular resolution. Larger apertures provide finer resolution, but this can make the cameras bulky and cumbersome, limiting their mobility.
- Frequency Range: The maximum frequency an acoustic camera can detect is limited by the spatial separation between adjacent microphones. To sample higher frequencies, the microphones must be spaced closer together. The lower effective cut-off frequency is defined by the total array size.
- Miniaturization Challenges: While there is ongoing research to miniaturize acoustic cameras for better portability, this often involves trade-offs in directivity and noise suppression capabilities, especially at lower frequencies.
Application Limitations and Environmental Factors
Acoustic cameras can more easily resolve multiple sources of equal power than sources with vastly different levels. The directivity pattern of the array introduces leakage when the main lobe is too wide or the side lobes are strong, masking weaker sources near louder ones. Advanced signal processing algorithms such as deconvolution techniques can extend the effective range but often introduce other challenges, such as increased sensitivity to noise and discrepancies between theoretical models and practical conditions.
Indoor Measurements and Acoustic Reflections
The propagation path between the sources and the measurement array can be strongly affected by the testing environment where the acoustic camera is used, which can effectively distort the resulting acoustic images. In environments with complex acoustics, such as architectural spaces, sound behavior is far from the predefined ideal free-field models used in the imaging algorithms, making it challenging to accurately map sound sources. Acoustic cameras can help verify compliance with noise limits but may not capture all noise emission points or provide a reliable sound pictures in acoustically complex environments.
Indoors, acoustic reflections from floors, walls, and other surfaces can create ghost sources that interfere with the accurate characterization of noise sources. These reflections complicate the analysis and often lead to artifacts in the resulting sound maps, making it difficult to link the maps to physical sources. Advanced signal processing algorithms can extend the effective range but often introduce other challenges.
Beamforming and Near-Field Acoustic Holography (NAH)
Far-field beamforming, the underlying algorithm used by most acoustic cameras, is a spatial filtering operation applied to sensor data to extract the localization information of noise sources. An array of transducers captures certain acoustic properties of the sound field at discrete positions. The data is processed using wave propagation models to estimate their direction of arrival, acting as a spatially discriminating filter.
Near-field acoustic holography (NAH), on the other hand, can be applied if the array is placed close to the noise sources, capturing both the evanescent and propagating sound waves, allowing for an accurate sound field reconstruction and quantification. Coherent sources are easily assessed, while incoherent sources require more complex processing methods. Commercial solutions using this technology are commonly referred to near-field acoustic cameras. Both beamforming and NAH are very powerful for multiple applications, especially in assessing non-stationary sources. However, their practical limitations, such as the finite, discrete aperture and sensor calibration errors, should not be disregarded.
Direct Sound Mapping: A Simple Alternative
Direct sound mapping offers a viable alternative to acoustic cameras, particularly for near-field applications under time-stationary conditions. Unlike acoustic cameras, direct sound mapping measures sound directly near the source, obtaining visual representations based on the measured sound that are not constrained by the frequency of interest (i.e. suitable for low-frequency noise problems). Physically measuring at a discrete set of points near the target device allows for accurate quantification of noise emission. This approach avoids the theoretical assumptions used by beamforming algorithms related to propagation path or testing environment that can lead to inaccuracies. Measuring close to the sources provides unparalleled spatial resolution with a very high dynamic range due to the intrinsic spatial variability of the sound field and near-field evanescence energy. Depending on the visualized quantity, the spatial resolution varies, with particle velocity being the superior quantity for sound localization, enabling the resolution of sources as close as the measurement distance. Furthermore, if sound intensity is captured all around the target device/source, accurate sound power characterization can be performed in-situ. This makes direct sound mapping very attractive for both troubleshooting and benchmarking products.
STAY UPDATED WITH THE LATEST SOUND AND VIBRATION RESEARCH
Subscribe to our Newsletter
Concluding remarks
Acoustic cameras are powerful tools for visualizing and localizing sound sources, especially in ideal conditions. However, their limitations must be acknowledged, particularly in low-frequency ranges, indoor environments, and scenarios with significant power disparities between sources. By understanding these limitations and considering alternatives like direct sound mapping, users can make more informed decisions and achieve more accurate results in their acoustic measurements.
While acoustic cameras offer valuable insights, they should not be regarded as infallible. Each sound map produced is a combination of measurement data and computational models that impose assumptions, and deviations from these assumptions can lead to misleading results. Therefore, acoustic cameras should be used carefully and preferably in conjunction with other methods to ensure reliable and comprehensive acoustic analysis.
Sound Source Localization
Near-Field Acoustic Camera