Extended depth‐of‐field reflected‐light microscopy with a low‐cost robotic microscope and Fiji focus stacking: Workflow integration and baseline evaluation

Optical reflected-light microscopy is widely used in mechanical and tribological testing laboratories, but its limited depth of field (DOF) restricts imaging of non-planar post-test surfaces when high-NA objectives are needed to resolve fine features. Here we present a low-cost robotic microscopy system (bill of materials around USD 300) that extends depth of field through controlled axial scanning and computational focus stacking. Built on the OpenFlexure platform and using open-source software tools, the system captures z-stacks and reconstructs all-in-focus micrographs in Fiji (ImageJ distribution) using extended-depth-of-field (EDOF) fusion engines. The workflow is demonstrated on mechanically tested surfaces (scratches and indentations), providing all-in-focus visualisation of features such as scratch lips, pile-up, indentation boundaries and microcracks on reflective textured surfaces, with Scanning Electron Microscopy used for cross-checking of feature detectability. Baseline characterisation includes operational DOF and resolving-power measurements, a geometric-distortion assessment using a micrometric grid target, and a time-cost analysis of stack acquisition. Comparative evaluation of three widely used fusion methods (Sobel, variance and complex wavelets) using image-quality metrics and visual inspection indicates that complex-wavelet fusion yields sharper reconstructions and improved edge preservation relative to Sobel and variance, at the expense of higher computation time. Finally, we summarise practical constraints and observed failure modes under representative conditions, providing baseline guidance for routine use, including deployment close to mechanical test stations. LAY DESCRIPTION: Low-cost reflected-light microscopy has shallow depth of field when high-NA objectives image rough surfaces. We demonstrate a fully open (∼USD 300) OpenFlexure-based robotic microscope with an end-to-end Fiji focus-stacking workflow that produces all-in-focus micrographs of scratches and indentations, cross-checked against SEM. Baseline characterisation covers DOF, resolving power, geometric distortion, and acquisition-time effects. Complex-wavelet fusion outperforms Sobel and variance in sharpness at higher computation time. Practical constraints are identified for post-test failure analysis.