【Paper Reproduction】——FEC: Fast Euclidean Clustering for Point Cloud Segmentation

1. Algorithm principle

1. Overview of the paper

  Segmentation from point cloud data is essential in many applications such as remote sensing, mobile robotics or autonomous vehicles. However, point clouds captured by 3D range sensors are usually sparse and unstructured, which poses challenges for effective segmentation. Fast solutions for computationally inexpensive point cloud instance segmentation are lacking. To this end, a new Fast Euclidean Clustering (FEC) algorithm is proposed, which applies a point clustering algorithm on the basis of the existing clustering algorithms and avoids traversing each point continuously.

2. Implementation process

  First, all points P i \mathbf{P}_i in the point cloud