A Fast Point Cloud Ground Segmentation Approach Based on Coarse-To-Fine Markov Random Field

Ground segmentation is an important preprocessing task for autonomous vehicles (AVs) with 3D LiDARs. However, the existing ground methods are very difficult to balance accuracy and computational complexity. This paper proposes a fast point cloud approach based on coarse-to-fine Markov random field (MRF) method. The method uses coarse result of improved local feature extraction algorithm instead prior knowledge initialize MRF model. It provides initial value fine dramatically reduces graph cut then used minimize proposed model achieve segmentation. Experiments two public datasets tests show that our more accurate than both features faster graph-based methods. can process Velodyne HDL-64E data frames in real-time (24.86 ms, average) only one thread I7-8700 CPU. Compared deep learning, it has better environmental adaptability.