3D Railroad Ballast Reconstruction by Defocusing

R. Lorusso, M. Nitti, E. Stella, N. Ancona, and A. Distante (Italy)


Depth-from defocusing, 3d reconstruction, railway monitoring


Ballast attitude estimation is a fundamental task in railroad infrastructure monitoring, in order to prevent railway accidents. In this paper an approach to recover depth of ballast, by analyzing the blur of corresponding image is presented. This is performed by a "depth from defocusing" technique which is based on the comparison of two images of the same scene, acquired at different focus distances. All these assumptions represent an advantage respect to methods based on triangulation, because problems like occlusion or shadowing are avoided. Moreover, limitations imposed by structured light based techniques are overcome. In order to manage the common difficulties in depth from defocusing technique, related to textures in the scene, we have calibrated the imaging system only on one specific texture (ballast). However, it will be possible to generalize this aspect, on our method, applying others texture analysis based techniques.

Important Links:

Go Back