Vessel detection and classification from spaceborne optical images: A literature survey

Remote Sens Environ. 2018 Mar 15;207:1-26. doi: 10.1016/j.rse.2017.12.033.


•A review of 119 papers on ship detection and classification from optical satellite.•From 1978 to March 2017, showing an exponential growth in the number of papers.•Most published methods have very limited validation.•While big steps have been made, automatic algorithms are still far from perfect.•Increase in new observation and processing capabilities promises rapid advances.

This paper provides an overview of existing literature on vessel/ship detection and classification from optical satellite imagery. Although SAR (Synthetic Aperture Radar) is still the leading technology for maritime monitoring, the number of studies based on optical satellite data is quickly growing. Altogether we analysed 119 papers on optical vessel detection and classification for the period from 1978 to March 2017. We start by introducing all the existing sensor systems for vessel detection, but subsequently focus only on optical imaging satellites. The article demonstrates the temporal development of optical satellite characteristics and connects this to the number and frequency of publications on vessel detection. After presenting the methods used for optical imagery-based vessel detection and classification in detail, along with the achieved detection accuracies, we also report possibilities for fusing optical data with other data sources. The studied papers show that the most common factors greatly influencing the vessel detection accuracy are the following: different weather conditions affecting sea surface characteristics, the quantity of clouds and haze, solar angle, and imaging sensor characteristics. All these factors bring great variations in the selection of the most suitable method; some still continue to pose unsolved challenges. For higher relevance and wider usage, we suggest that the algorithms for detection and classification should support a variety of targets and meteorological conditions, and ideally also a variety of optical satellite sensors. At least, they should be tested on many images under different conditions. This is not usually the case in the existent literature. We also observed that many authors omit an appropriate performance quantification, which is critical for a practical assessment and a numerical comparison of the presented algorithms. Overall it can be seen that vessel monitoring from spaceborne optical images is a popular research topic and has a great operational potential in the near future due to the large amount of satellite data, much of it free and open.

Keywords: Maritime domain awareness; Object recognition; Optical satellite data; Sea target detection; Ship classification; Ship detection; Vessel classification; Vessel detection.

Publication types

  • Review