In the context of semiconductor 3D packaging and MEMS (Micro-Electro-Mechanical Systems) device manufacturing, the attributes of deep trenches created through deep reactive ion etching (DRIE)—such as their depth, sidewall inclination, and waviness—exert a direct influence on the performance of the devices. Nevertheless, the pattern dependency effects lead to considerable discrepancies in the etching topography across trenches with varying opening sizes. Presently, destructive cross-sectional observation methods are not only time-intensive but also ill-suited for online production environments. On the other hand, non-destructive optical inspection techniques are constrained by their resolution limitations, posing challenges in accurately reconstructing the internal details of deep trenches. Consequently, there is an urgent demand within the industry for an online measurement technology that is capable of high-throughput, high-precision, and entirely non-destructive operations.