<p>A micro-scale crack detection framework is proposed that integrates full-field, image-based displacement measurements with nonlocal differential operators for robust identification of discontinuities. Full-field displacement fields are first extracted from sequential micro-scale images using digital image correlation (DIC), yielding pixel-level accuracy without imposing continuity assumptions. A peridynamic differential operator is then applied to compute strain components and assess strain compatibility over a nonlocal neighborhood. In the presence of cracks, displacement discontinuities lead to violations of the compatibility condition in the computed strain field. Locations exhibiting pronounced incompatibility are subsequently analyzed using a Multivariate Adaptive Regression Splines (MARS)–based regression approach to identify crack location, geometry, and propagation paths. The proposed framework is demonstrated through tracking cracks in impact-induced bonded interfaces, with accuracy validated against experimental observations. The method enables reliable detection and tracking of cracks and other discontinuities at micrometer scales from arbitrary image sets, providing a quantitative characterization of interfacial damage. This capability is particularly relevant to the manufacturing and repair of electronic components.</p>