Despite the efforts on reducing maritime accidents, they still occur and, from time to time, have catastrophic consequences both in personal, environmental and financial terms. Structural failure is the major cause of ships wreckages and, as such, Vessel Classification Societies impose extensive inspection schemes for assessing the structural integrity of vessels.
The external and internal parts of the hull can be affected by different kinds of defects typical of steel surfaces and structures, such as cracks and corrosion. Nowadays, to detect these defects, visual hull inspections are carried out at a great cost. The goal of the EU-funded FP7 MINOAS project is to develop a fleet of robots for automating as much as possible the aforementioned inspection and maintenance operations.
Within this general context, the work presented constitutes a first attempt towards the remote visual inspection and documentation of hull surfaces. In this regard, the two main defective situations, cracks and corrosion, are expected to be autonomously or semi-autonomously detected by means of computer vision techniques.
In this work, several algorithms are presented for visual detection of the above mentioned two kinds of defects. On the one hand, a crack detector is described, which is based on a percolation process that exploits the morphological
properties of cracks in steel surfaces: dark, narrow and elongated sets of connected pixels.
On the other hand, two different approaches for corrosion detection are introduced and compared. While the first one takes profit from the distribution of color in corroded areas, the second one has been built around a weak classifier cascade scheme, separating the spatial and colour analysis in two different steps. As an final contribution, the crack detector is combined with the corrosion detector in order to guide the crack location and improve its performance. The obtained detectors have shown promising rates of detection as well as close to real-time performance.