Model driven method for crack detection in robotic inspection

Nowadays civil infrastructure is exposed to several chal-lenges such as daily vehicular traffic and extreme weather conditions. It is well known that these may determine struc-tural deterioration and damages, which can even cause catas-trophic collapses related to significant socio-economic losses. For this reason automatic inspection and maintenance are expected to play a decisive role in the future. With the ob-jective of quality assessment, cracks on large civil structures have to be identified and monitored continuously. Due to the availability of cheap devices, techniques based on image pro-cessing have been gaining in popularity, but they require an accurate analysis of a huge amount of data. Moreover, frac-ture detection in images remains a challenging task due to high sensitivity to noise and environmental light conditions. This paper proposes a mathematical method for detecting cracks in images along with a parallel implementation on heterogeneous High Performance Computing (HPC) archi-tectures aiming both at automatizing the whole process and at reducing its execution time.