An in-depth assessment of simplification algorithm performance across diverse feature types and scales, employing a range of threshold values

No previous study has simultaneously examined the influence of diverse feature types, scales, and thresholds on simplification performance. We compared ten algorithms using six measures, seven datasets, and eleven thresholds. Results revealed that ‘Ramer-Douglas-Peucker,’ ‘Sleeve-fitting,’ and ‘Before Opening Window’ produced the largest changes in angularity and vector displacement, while minimizing percentage changes in coordinates and curvilinear segments. The opposite holds for Triangular Routine, Euclidean Distance, and Perpendicular Distance. Performance varies with feature type, scale, and threshold, revealing anomalous trends requiring further investigation. Statistical validation supports these findings, and guidelines help avoid misleading comparisons of newly developed algorithms.