Ship detection in port surveillance based on context and motion saliency analysis

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

“…Video-based techniques where sequences of image frames are analysed are well studied in road vehicle detection, but for vessel detection, video analysis still remains a challenging and complex task. Recently, several papers discussed the promising usage of video cameras to perform detection of vessels (Arshad et al, 2010, Bao et al, 2013, Fefilatyev et al, 2010, Hu et al, 2011, Rodriguez Sullivan and Shah, 2008). Camera-based vessel detection is attractive due to its low cost and ease of management, both in installation and maintenance.…”

Context modeling combined with motion analysis for moving ship detection in port surveillance

“…Therefore, the performances of different ship detection techniques are difficult to compare. To analyze our improved approach, we compare it with our previous algorithms Existing 20 and Cabin detector.…”

“…Additionally, we compare this approach with the ship detection based on HOG 15 and support vector machine (SVM) 14,19,24 and an initial version of the concept discussed in this paper. 20 Finally, Sec. 6 presents conclusions and discusses future work.…”

“…In our earlier work, 19,20 we presented a preliminary framework of context-based ship detection by extracting the water region as contextual information. Although the context is not fully modeled because only water regions are labeled, the performances of those approaches are promising for the context-based detection.…”

“…We further develop a context-based method to perform automatic moving ship detection, which completes the context model explored in our previous work. 19,20 Additionally, the temporal properties of moving ships are exploited to fuse the visual, spatial, and temporal features in a single framework. Although some previous work also extracts supporting information to facilitate ship detection, such as modeled water region, 9,21 the water region is only used as a background model and modeled at the pixel level.…”