Abstract
This project was carried out to implement low-cost cameras in visual simultaneous localization and mapping (SLAM) for unmanned surface vessel (USV), by generating landmarks from stereo images. A program with pre-processing steps and a stereo block matching algorithm was developed. Gaussian filter, morphological transformation and binarization were applied in the pre-processing steps to segment the coastline from the stereo images. The block matching algorithm then computed the disparity and generated point clouds from the coastline pixels. The generated point clouds were noted to properly reflect the coastlines with an overlapping between different point clouds, inferring their potential to be used as landmarks in SLAM.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
