Harbor detection in remote sensing images based on feature fusion

Harbor detection technology for remote sensing images has an important effect in military and civil affairs. By analyzing harbor´s prior knowledge, this paper proposes a harbor detection method of remote sensing images by fusing the geometry geographical features and jetty features. Firstly, geometry geographical information and jetty information are extracted separately from the sea-land segmented remote sensing images by scanning according to harbor´s prior knowledge. Secondly, jetties are merged or abandoned intelligently by fusing the two information based on their logical relativity. Lastly, identify the jetties and exactly locates the harbor target. By Comparing with Chen´s harbor detection algorithm, our method can make a more complete jetty scan and exact location, and then whole harbor information can be obtained. Also our method does not need to set the distance threshold to merge or abandon jetties, and thus is more general.