Monitoring of the railroad infrastructure is crucial for safety concerns and accident prevention. This task requires regular surveillance which is nowadays still carried out by expensive and time consuming manual visual inspections in many countries. The problem of railroad cable recognition (contact cables, catenary cables, return current cables) in rural areas based on LiDAR point clouds has it own methods, but current state of the art solutions suffer from extremely high computational load due the extensive size of the datasets. In this study we analyzed and compared novel robust solutions focusing on minimizing the assumptions (positions and distances of track bed, rail tracks, etc.) of the algorithms, and providing faster filtering methods that prepare the point cloud by removing the most of the points that cannot be part of a cable. Beside implementing the investigated methods based on open source libraries, we created a framework for algorithm pipelining, input, output and intermediate result management, and performance analysis.