Matching design-intent planar, curved, and linear structural instances in point clouds

The lack of timely progress monitoring and quality control contributes to cost-escalation, lowering of productivity, and broadly poor project performance. This paper addressed the challenge of high-precision structural instance segmentation from point clouds by leveraging as-designed IFC models in Scan-vs-BIM contexts. We proposed an automatic method to segment the entire points corresponding to the as-designed instance. The workflow contains: 1) Instance descriptor generation; 2) PROSAC-based shape detection; 3) DBSCAN-based cluster optimization. The method matches design-intent planar, curved, and linear structural instances in complex scenarios including: 1) the as-built point cloud is noisy with high occlusions and clutter; 2) deviations between as-built instances and as-designed models in terms of position, orientation, and scale; 3) both Manhattan-World and non-Manhattan-World instances. The experimental results from five diverse real-world datasets showed excellent performance with mPrecision 0.962, mRecall 0.934, and mIoU 0.914. Benchmarking against state-of-the-art methods showed that the proposed method outperforms all existing ones.