Checking railway status is critical to guarantee high operating safety, pro
per maintenance schedule, and low maintenance and operating costs. This ope
ration consists of the analysis of the rail profile and level as well as ov
erall geometry and ondulation, Traditional detection systems are based on m
echanical devices in contact with the track. Innovative approaches are base
d on laser scanning and image analysis. This paper presents an efficient co
mposite technique for track profile extraction with real-time image process
ing. High throughput is obtained by algorithmic prefiltering to restrict th
e image area containing the track profile, while high accuracy is achieved
by neural reconstruction of the profile itself.