MAPClean – Microstructurally Adaptive Pixel-Level Cleaning

A modular MATLAB pipeline for automated cleaning of Electron Backscatter Diffraction datasets MAPClean is a modular MATLAB pipeline, built on the open-source MTEX toolbox, for automated cleaning of Electron Backscatter Diffraction (EBSD) datasets. The workflow operates at the pixel level, combining phase-aware and orientation-aware logic to remove noise, correct misindexed pixels, and reconstruct missing data. It employs an intelligent Strict vs. Relaxed protocol selection based on data quality to apply the most appropriate cleaning strategy. The pipeline features Mean Angular Deviation (MAD) filtering, sample-mask generation for physically grounded restriction of the mapped specimen region, phase and orientation wild spike removal (WSR), adaptive hole filling using parallelised Breadth-First Search (BFS, Strict mode) or Multi-Pass Filling (MPF, Relaxed mode), and a protected hole filling for conservative recovery of uncertain grain interior regions. Parallel execution via MATLAB's Parallel Computing Toolbox is used in the strict BFS hole-filling stage and in the orientation wild-spike detection stage.