Roads and pipes detection within LADAR intensity images through fuzzy techniques
Pilar Sobrevilla, James M. Keller, Eduard Montseny · 2002
There is uncertainty in many aspects of image processing and computer vision. Fuzzy set theory and fuzzy logic are ideally suited for dealing with such uncertainty. Image segmentation is an important step in many computer vision algorithms, and errors made in this stage will impact all higher-level activities. This paper extends our earlier and on-going work in image-labeled segmentation (E. Montseny and P. Sobrevilla, 1998), wherein methods which incorporate the uncertainty of object and region definition and the faithfulness of the features to represent various objects were considered. To apply our previous system and framework to LADAR (LAser raDAR) images, it has been modified and improved. We have introduced new fuzzy morphological structural elements to eliminate "vertical noise" and false detections, and we have improved the segmentation of the histogram for dealing with problems due to the wide range and variability of gray levels of the elements appearing within these images.