poXNOR Morphological Transform based Feature Extraction for Mobile Robot Applications
Abstract
Efficient feature extraction algorithms are needed for localization, navigation of the autonomous mobile robot and fast mapping of the environment. This paper presents an experimental implementation of the morphological transform to extract the features of the environment. A novel percentage occupancy XNOR (poXNOR) approach has been proposed to address the same. To enhance the computational speed, compressed images of different sizes were generated using the laser range finder (LRF) data. Structure elements proportional to this image size were used for performing the poXNOR based morphological transformation. A significant 70% positive hit of the features with ±0.05 m accuracy was observed.
Keywords
Feature Extraction, Morphological Transform, Localization, Mobile Robots, Laser Range Finder