High-resolution LiDAR technology has revolutionized sinkhole detection, offering a more efficient and accurate alternative to traditional, labor-intensive visual inspections. The aim of this paper is to develop an automated approach to detect sinkholes in the railway environment by extracting relevant spatial features. Our methodology starts by the precise extraction of ground points from LiDAR point clouds. Efficiently isolating ground points enables the detection and analysis of ground deformations and subsidence patterns near railway tracks. The Voxel Based Ground Filtering algorithm (VBGF) is compared to the robust Cloth Simulation Filter (CSF). High-resolution Digital Elevation Models (DEMs) are subsequently generated from the LiDAR ground points, and sinkholes are detected using a segmentation-based approach. This study presents a comprehensive method for sinkhole detection, facilitating the early identification of potential sinkhole formations and offering a proactive strategy for maintaining the safety and integrity of railway infrastructur.