Understanding False Positives in Acoustic Imaging: Artifacts, Reflections & How to Avoid Them

You’re scanning overhead pipework in a compressor room when a bright hotspot appears on your acoustic camera’s display — right on a steel support beam. You walk over, listen carefully, run your hand along the surface. Nothing. No hiss, no vibration, no leak. The screen is showing you a sound source that doesn’t exist.

Before you question your equipment, take a breath. That phantom hotspot isn’t a malfunction. It’s physics — and understanding it will make you a significantly better acoustic imaging operator.

False positives are an inherent characteristic of how acoustic cameras reconstruct sound fields. Every vendor’s product generates them, though the type and severity vary with array design, algorithm quality, and operating environment. Field experience across industrial facilities suggests that 15–30% of initial acoustic imaging indications in reflective environments — plants with metal ducting, concrete floors, and steel enclosures — turn out to be false positives rather than real sources. Most manufacturers avoid this topic entirely. We think that’s a mistake, because the engineers who understand false positives don’t just avoid misdiagnosis — they use these artifacts to locate sources more effectively.

This article covers the three types of false positives you’ll encounter, how to identify them, three techniques to minimize them, and — perhaps most importantly — how to turn them into diagnostic allies.

What Are False Positives in Acoustic Imaging?

False positive (acoustic imaging): An apparent sound source indication on an acoustic camera display that does not correspond to an actual physical source at that location. Caused by physical phenomena including sound wave reflections, beamforming algorithm sidelobes, or environmental noise interference — not by equipment defect.

Three related terms often get used interchangeably, but they describe different phenomena:

• False positive: Any indicated source that isn’t real at the shown location
• Artifact: A systematic error pattern produced by the beamforming algorithm itself (e.g., sidelobes)
• Ghost image: A reflected or mirrored source — real sound arriving from an indirect path

Understanding these distinctions matters because each type has different causes, different on-screen characteristics, and different solutions.

Three Types of False Positives

Reflections (Ghost Images)

Sound waves bounce off hard, smooth surfaces — metal walls, concrete floors, glass panels, polished pipes — just like light reflects off a mirror. When your acoustic camera picks up both the direct sound and the reflected sound, the reflected path appears as a second source at a location where nothing is actually producing noise.

Typical scenario: You’re imaging a compressed air manifold mounted near a stainless steel wall. The display shows two hotspots — one on the manifold (real) and one on the wall behind it (ghost). The ghost image appears at roughly the same intensity and frequency as the real source. On-screen signature: Ghost images tend to appear at geometrically symmetrical positions relative to the reflecting surface. They share the same frequency spectrum as the real source and often appear at similar or slightly reduced intensity.

Sidelobe Artifacts

This is the most technically nuanced type. Beamforming algorithms work by mathematically “focusing” the microphone array on each point in the field of view. But just as a flashlight can’t produce a perfectly sharp beam edge, beamforming produces a main lobe (the focused area) surrounded by sidelobes — weaker response regions that can register false sources.

Typical scenario: You’re imaging a single loud leak, but the display shows the main hotspot surrounded by a ring or pattern of secondary spots. These sidelobe artifacts are always clustered around the true source and become more pronounced when the source is loud relative to surrounding noise. On-screen signature: Sidelobes appear as a repeating pattern around the main source — often a ring, halo, or radial spoke pattern. Their intensity is always lower than the main lobe, and they maintain a fixed geometric relationship to the primary source regardless of scanning angle. Key factor: The number of microphone channels directly affects sidelobe levels. A 64-channel array produces more prominent sidelobes than a 128-channel array, which in turn produces more than a 200-channel array. Higher channel counts provide narrower main lobes and lower sidelobe floors. Advanced algorithms like CRYSOUND’s HyperVision processing further suppress sidelobes beyond what standard delay-and-sum beamforming achieves.

Environmental Noise Interference

Not every unwanted indication is a reflection or algorithm artifact. Sometimes, your acoustic camera is accurately detecting a real sound — just not the one you’re looking for. Background noise from HVAC systems, nearby machinery, overhead cranes, or even wind can register as apparent sources that get confused with your target.

Typical scenario: During a compressed air leak survey in a manufacturing hall, you see multiple hotspots across a wide area. Some are genuine leaks. Others are background machinery noise that happens to fall within your selected frequency band. On-screen signature: Environmental noise sources typically have broader, more diffuse patterns than leaks (which appear as tight, focused hotspots). They also show different frequency characteristics — machinery noise tends to be narrower-band and harmonic, while leak noise is broadband and turbulent.

Quick Comparison

How to Identify False Positives: A 4-Step Process

When you see an indication you’re unsure about, run through these steps before logging it as a confirmed source:

1. Angle change test. Move 2-3 meters to the side and re-scan the same area. Real sources stay in the same physical position on the image. Reflections shift or disappear as you change the angle of incidence. Sidelobes rotate with the main source.

1. Frequency signature check. Switch to spectrum view (if your camera supports it) and examine the frequency profile. Compressed air leaks produce broadband, turbulent noise typically above 20 kHz. Machinery noise has distinct tonal peaks. If the “source” shares an identical spectrum with a known real source nearby, it’s likely a reflection.

1. Distance validation. If your camera allows distance-to-source measurement, check whether the indicated distance matches the physical geometry. A reflection off a wall 3 meters behind you will show a source distance that doesn’t match the wall’s position from your scanning location.

1. Ultrasonic listening. Most acoustic cameras offer a downshift playback feature that converts the directionally-captured ultrasonic signal into audible sound through headphones. Point the camera at the suspect indication and listen. A real compressed air leak produces a distinctive broadband hiss. Sidelobe artifacts and reflections carry no independent acoustic signature — they sound identical to the main source. Environmental noise sounds tonal and harmonic. This lets you verify from your scanning position without walking to the indicated location.

Techniques to Minimize False Positives

1. Select the Right Frequency Band

Most acoustic cameras allow you to filter by frequency range. Narrowing the band to your target application reduces interference dramatically. For compressed air leaks, focus on 20–50 kHz. For partial discharge, 20–100 kHz. Excluding lower frequencies cuts out most machinery noise.

2. Use Advanced Beamforming Algorithms

Not all beamforming is equal. Standard delay-and-sum (DAS) algorithms are computationally simple but produce higher sidelobe levels. Advanced algorithms apply spatial filtering to suppress sidelobes at the processing level. CRYSOUND’s HyperVision algorithm, available on the CRY8124 and CRY8120 series, provides up to 10x processing power over standard beamforming, reducing sidelobe artifacts significantly without sacrificing real-time performance.

3. Adopt a Multi-Angle Scanning Protocol

Make it standard practice to scan critical areas from at least two different positions. Compare the results. Sources that appear consistently in the same physical location are real. Sources that shift, disappear, or change intensity are false positives. This takes an extra 30 seconds per area and dramatically reduces misdiagnosis.

Turning Artifacts into Allies: Using False Positives to Locate Sources

Here’s where experienced operators separate from beginners. False positives aren’t just noise to eliminate — they carry information about the acoustic environment that you can exploit.

Reflection Mapping

If you see a ghost image on a metal wall, you’ve just learned something valuable: there’s a real sound source at the geometrically mirrored position. The reflection tells you the sound wave’s travel path. In complex piping environments where direct line-of-sight to a leak is blocked, reflected images on nearby surfaces can reveal the general direction and approximate distance of sources you can’t see directly.

Pro technique: When scanning in confined spaces with multiple reflective surfaces, intentionally note where ghost images appear. The pattern of reflections triangulates the true source location, even when the source is behind equipment or above a ceiling panel.

Sidelobe Pattern Reading

Sidelobes always radiate symmetrically from the main lobe. When you see a sidelobe pattern, the center of that pattern is your real source — guaranteed. In environments where obstructions partially block your view, the visible sidelobes can confirm the direction of a source that’s partially hidden. If the sidelobe “ring” is only visible on one side, the true source is on the opposite side, behind whatever’s blocking your view.

Spatial Focusing (ROI Isolation)

In noisy industrial environments, you can’t shut down surrounding equipment just to get a cleaner acoustic image. Instead, use your camera’s Focus Function: draw a region of interest (ROI) directly on the screen around the area you want to investigate. The camera restricts its beamforming analysis to that region only, effectively suppressing sound sources outside the selected area.

This is particularly powerful when combined with frequency band filtering. Narrow the frequency range to your target application (e.g., 20–50 kHz for leak detection), then draw an ROI around the suspect zone. The double filter — spatial plus spectral — dramatically reduces environmental noise interference without requiring any change to the plant’s operating conditions. What previously looked like a cluttered display with overlapping hotspots becomes a clean, focused view of the area that matters.

Quick Reference Checklist

Eliminating false positives:

• Set correct frequency band for target application
• Scan from at least two angles
• Check frequency spectrum of ambiguous sources
• Use ultrasonic listening to verify auditory signature before logging

Leveraging false positives:

• Note reflection positions to triangulate hidden sources
• Read sidelobe patterns to confirm source direction
• Use Focus Function (ROI) to isolate target area from surrounding noise

Need help diagnosing false positives? Talk to our application engineers →

Frequently Asked Questions

How common are false positives in acoustic imaging?

Every acoustic camera produces some false positives — it’s an inherent characteristic of beamforming physics, not a defect. The frequency and severity depend on the array’s channel count, algorithm quality, and operating environment. With proper technique, experienced operators quickly learn to distinguish real sources from artifacts within seconds.

Can reflections be completely eliminated?

No. As long as there are hard, smooth surfaces in the environment, sound will reflect. However, you can minimize their impact by scanning from multiple angles, using frequency analysis to match spectra, and applying software post-processing. Advanced algorithms reduce reflection sensitivity, but they cannot violate the physics of wave propagation.

Does a higher microphone channel count reduce artifacts?

Yes. More channels provide a narrower main lobe and lower sidelobe floor. A 200-channel array produces significantly cleaner images than a 64-channel array. This is one of the most meaningful specification differences when comparing acoustic cameras across manufacturers — channel count directly impacts false positive rates.

How do I handle false positives in noisy factory environments?

Combine three techniques: narrow your frequency band to the target application range (e.g., 20–50 kHz for leak detection, which excludes most machinery noise), use the camera’s Focus Function to draw an ROI around the suspect area and suppress surrounding interference, and scan from multiple angles to confirm source persistence. The spatial-plus-spectral double filter handles most noisy environments without requiring any change to plant operations.

Can acoustic reflections actually help locate the real source?

Yes — and this is an underappreciated technique. Reflections reveal the sound wave’s travel path, which means a ghost image on a wall tells you the real source is at the mirrored position relative to that surface. In complex industrial environments with obstructions, deliberately noting reflection patterns can triangulate sources that are not in direct line of sight.