Metodología para el desarrollo de nuevas técnicas y algoritmos en el reconocimiento inteligente de imágenes submarinas

Abstract

This thesis presents a collection of techniques and algorithms that in the past have been used in digital filtering of optical images and this time have been applied in acoustic images, more specifically, in underwater images from a sidescan sonar of single beam. For this, a program has been implemented in Matlab with an intuitive and user-friendly interface where the user can try out different filters and see the effect they have on the image. This process is performed in order to optimize the image, but taking special care in removing noise and finding edges. After the image pre-processing, it is time for the image segmentation applying a recognition filter of man-made objects, which is very useful for inspecting the seabed.