Segmentation and wake removal of seafaring vessels in optical satellite images
conference paper
This paper aims at the segmentation of seafaring vessels in optical satellite images, which allows an accurate length estimation. In maritime situation awareness, vessel length is an important parameter to classify a vessel. The proposed segmentation system consists of robust foreground-background separation, wake detection and ship-wake separation, simultaneous position and profile clustering and a special module for small vessel segmentation. We compared our system with a baseline implementation on 53 vessels that were observed with GeoEye-1. The results show that the relative L1 error in the length estimation is reduced from 3.9 to 0.5, which is an improvement of 87%. We learned that the wake removal is an important element for the accurate segmentation and length estimation of ships.
TNO Identifier
479496
Publisher
SPIE
Article nr.
88970B
Source title
Electro-Optical Remote Sensing, Photonic Technologies, and Applications VII, 25-September 2013, Dresden, Germany
Editor(s)
Kamerman, G.W.
et al
et al
Place of publication
Bellingham, WA