In this paper, a vision-based system for underwater object detection is pre
sented. The system is able to detect automatically a pipeline placed on the
sea bottom, and some objects, e.g. trestles and anodes, placed in its neig
hborhoods. A color compensation procedure has been introduced in order to r
educe problems connected with the light attenuation in the water. Artificia
l neural networks are then applied in order to classify in real-time the pi
xels of the input image into different classes, corresponding e.g. to diffe
rent objects present in the observed scene. Geometric reasoning is applied
to reduce the detection of false objects and to improve the accuracy of tru
e detected objects. The results on real underwater images representing a pi
peline structure in different scenarios are shown. The presence of seaweed
and sand, different illumination conditions and water depth, different pipe
line diameter and small variations of the camera tilt angle are considered
to evaluate the algorithm performances.