Autonomous UAV inspection in industrial refineries demands precise landing on pipelines under GNSS-denied conditions. This work presents a complete software architecture that integrates LiDAR-based localization with a novel multi-sensor framework for pipeline perception and tracking. The proposed method processes dual 2D laser scans and fuses them with depth data through a temporal filter. This enables consistent mapping and tracking of pipelines, effectively rejecting spurious detections and maintaining a reliable scene representation. The system is validated both in controlled indoor experiments and through fully autonomous landing missions in a real refinery, demonstrating robust performance and accurate alignment with the target infrastructure.