PSI - Issue 77

Francisco Afonso et al. / Procedia Structural Integrity 77 (2026) 584–592 F. Afonso et al. / Structural Integrity Procedia 00 (2026) 000–000

591

8

3.4.2. Digital Image Correlation In this test, Digital Image Correlation is used as the processing method to measure vibration. The same optical setup was used, but with a printed gradient target which, at high magnification, resembles a pseudo-speckle pattern. Figure 9 shows the region of interest, the measured displacement over time and the corresponding vibration power spectrum. Although 33.0 Hz can be distinguished in the power spectrum, it is not a clearly dominant frequency. These measurements indicate that while the pseudo-speckle pattern is usable, its quality may be sub-optimal.

(a) Region of interest displaying a close-up gradient.

(b) Displacement over time and power spectrum vibrations.

Fig. 9: Gradient target as seen from the frame camera and DIC results.

4. Conclusions and future works

This paper explored several optical approaches to vibration measurement. Di ff erent setups were assembled using neuromorphic and conventional frame cameras, with either a standard 8 mm lens or a macro lens setup with 4.0 × magnification. The neuromorphic configurations estimated the predominant vibration frequency through the camera’s events stream, while the frame camera produced measurements using image tracking and DIC. A tachometer first established the reference rotational speed for the balanced and unbalanced fan, providing an expected dominant frequency for comparison. When balanced, the fan rotated at 32.5 rps and when unbalanced, 31.8 rps. Using a tri-axial accelerometer with the unbalanced fan and an earlier version of the holder configuration, it exhibited approximately 31.0 Hz which was also used as a reference to validate the optical methods. Table 1 compiles the dominant frequencies obtained by each setup.

Table 1: Dominant frequencies per method.

Setup

Specimen

Dominant frequency [Hz]

Neuromorphic camera with standard lens Neuromorphic camera with macro lens

Unbalanced fan and bands target (earlier holder configuration)

31.6 32.3 33.0 33.0 1

Balanced fan and grid target Balanced fan and grid target Balanced fan and gradient target

Frame camera with macro lens Frame camera with macro lens

1 This frequency is visible but it is not clearly dominant.

Overall, the results agree acceptably across event-based imaging, accelerometry, image tracking and DIC. It should be noted, however, that accelerometer measurements introduced mass to the specimen, which may have slightly al tered the readings. Among the optical methods employed, image tracking provided the richest results; however, the neuromorphic camera enabled the simplest hardware setup (single USB connection) and low-latency operation. In practice, standard event-based frequency analysis returns the dominant vibration frequency and does not typically provide a full frequency spectrum without additional processing. Additionally, vibrational amplitude is not trivial to obtain, even as displacement. By contrast, the conventional frame-based approach requires additional synchronization hardware to achieve the required 300 fps.