Extracting sub-pixel displacement measurements using visual vibrometry methods for non-destructive evaluation

Vibration-based methods are crucial for Non-Destructive Evaluation (NDE), with non-contact techniques valued for their adaptability to various materials and structures. This paper presents a cost-effective approach to vibrational analysis using off-the-shelf cameras to capture videos of forcibly excited structures. The method combines forced excitation with phase-based optical flow to extract sub-pixel displacements, eliminating the need for motion magnification. Forced excitation facilitates temporal Fourier analysis, optimising sub-pixel displacement extraction, enhancing sensitivity, and improving efficiency. In contrast to traditional methods that require high-frame rate cameras and extensive computational resources, this system operates effectively with a low-frame rate camera, making it well-suited for NDE applications in industrial settings. Visual vibrometry works on the principle that damage alters a structure’s vibrational response, using changes in pixel values as motion sensors for sub-pixel measurements. Forced excitation amplifies sensitivity by actively probing structures, aiding in early-stage damage detection. A case study of a cantilever structure demonstrated the system’s ability to resolve displacements as small as 0 . 65 μ m (1/200th of a pixel). Comparative analysis of defective and pristine cantilever samples revealed subtle differences in mode shapes, with the optimisation curve-fit method successfully identifying defects and showing promise for defect localisation. • Phase-based motion estimation with forced excitation for defect detection without magnification. • Temporal Fourier transform enhances computational efficiency and displacement accuracy. • Optimised curve fitting and sub-pixel displacement extraction for defect localisation. • Method works effectively with low-frame rate video and minimal hardware requirements. • Sensitivity analysis improves defect detection and localisation in visual vibrometry.