Ground Point Removal From 3D Point Cloud Using 3d Ransac Based on Normal Vector

Abstract

Capturing objects in the real world using Light Detection and Ranging (LiDAR) is challenging due to the existence of ground points. The ground point tends to reduce the performance of computer vision-related tasks such as object recognition, detection, segmentation, and classification. This paper proposes a ground removal method for 3D Point Cloud using Random Sample Consensus (RANSAC). RANSAC algorithm effectively removes ground points on normal vector-based 3D points cloud objects. RANSAC fits data models containing significant inliers. By selecting the coordinate points, followed by calculating plane equation coefficients using normal vectors, removal is done by calculating each point's distance from the ground plane and identifying inliers based on a specified threshold. This analysis shows that the RANSAC algorithm effectively removes ground points.