Simplification of polygons from point cloud data for automated floorplan generation

This paper investigates methodologies for identifying significant edges in polygons, with a particular focus on applications in floor plan analysis. The primary objective is to develop and evaluate an algorithm that can accurately simplify and refine polygonal representations of rooms, ensuring that key structural details are preserved. The study introduces a novel simplification algorithm that combines holistic filtering, regression techniques, and a mechanism for closing polygons. This algorithm is designed to address the shortcomings of existing methods by providing a more comprehensive approach to reducing complexity while maintaining essential geometric features. The algorithm is validated through testing on two datasets of floor plans derived from real laser scans. The findings contribute to the field of automated floor plan generation, offering a scale-independent solution for simplifying redundant polygonal data while preserving spatial accuracy.