HIF: Height Interval Filtering for Efficient Dynamic Points Removal

Abstract

3D point cloud mapping is crucial for localization and navigation, but residual traces of dynamic objects compromise map quality, posing a key challenge for real-time applications in dynamic environments. Existing approaches, however, often incur significant computational overhead, making it difficult to meet the real-time processing requirements. To address this issue, we introduce the Height Interval Filtering (HIF) method, which constructs pillar-based height interval representations to probabilistically model the vertical dimension and updates interval probabilities using Bayes filter. Furthermore, we propose a low-height preservation strategy that improves the detection of unknown spaces, reducing misclassification in areas blocked by obstacles. Experiments on public datasets show that HIF achieves a 7.7× improvement in runtime while maintaining comparable accuracy and enhanced robustness in complex, dynamic environments. The code will be publicly available.