UAV Hyperspectral Corrosion and Coating Spectral Libraries — Itzenplitz Headframe
Year: 2026
Type: dataset
Abstract: This dataset contains the spectral libraries, inter-class separability matrix, and structural delineation masks supporting the study "UAV Hyperspectral Corrosion and Coating Characterization of Post-Mining Steel Infrastructure Using Hierarchical Spectral Libraries" (submitted to Mining, MDPI). Hyperspectral data were acquired with a HAIP BlackBird V2 pushbroom sensor (100 bands, 502–997 nm) mounted on a DJI Matrice 350 RTK platform, using both nadir and oblique viewing geometries. Oblique acquisition was applied to image the vertical and near-vertical steel surfaces of the Itzenplitz headframe, a post-mining structure with visible coating degradation, staining, and corrosion.Contents Grouped global spectral library — six consolidated corrosion and coating classes derived from ROI spectra of six reference steel samples, provided at class level and source level. Itzenplitz site-specific spectral library — seven surface-state classes derived through a staged, criteria-driven procedure (structural delineation, unsupervised clustering, consolidation by spectral angle, acceptance by within-class angular dispersion, cross-scene verification, and targeted refinement), provided at class level and source level. The source-level tables preserve full traceability of scene, source name, source type, class assignment, wavelength, mean reflectance, dispersion, and pixel support. Inter-class spectral angle matrix — complete 7×7 separability matrix in radians. Analysis masks — polygon definitions and rasterized boolean masks for the four field scenes (STRONG, PAINTED, EARLY, COMPLEX), enabling exact reproduction of the structural delineation step. Reflectance values are given in percent and follow the standardized UAV hyperspectral preprocessing workflow described in the authors' companion framework paper. A README documents the schema of every file and its correspondence to the tables and sections of the associated manuscript. This dataset was produced within the KONSENZ project (Concept Development for Sensory Condition Monitoring of Industrial Cultural Heritage), funded by the RAG-Stiftung under project reference number 25-00599, at the Research Center of Post-Mining (FZN), THGA Bochum.
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7207565486