Advanced Likelihood Generators for Boundary Detection.
ROCHESTER UNIV NY DEPT OF COMPUTER SCIENCE
Pagination or Media Count:
Previously discussed were the advantages of feature detectors that return likelihoods and some preliminary results with a boundary point detector that returns likelihoods. Now I have developed boundary point detectors that have these properties Return probabilities Can be combined robustly Potential sub-pixel precision Work with correlated noise and Can handle multiple gray levels tested with 255. These detectors were applied both to aerial images and to test images whose boundaries are known. They are compared to established edge detectors. Keywords Edge detection Template Likelihood Bayesian reasoning.