Extraction Technology Of Inshore Target From Remote Sensing Images

Target extraction and recognition has always been a focused and difficult problem in the field of photogrammetry and remote sensing.Inshore target extraction is fundamental for the construction of sea battlefield environment and the development of marine economic.The rapid development of satellite remote sensing technology provides data support for inshore target extraction using remote sensing images.In addition,the continuous development of image interpretation and other related technologies makes it possible to extract inshore targets rapidly and automatically.Research in this dissertation is mainly focused on waterline extraction,breakwater and dock extraction,ship detection.The major work and innovations included in this dissertation are listed as follow:1.A waterline extraction method based on quad-tree and multiple active contour model is proposed in this dissertation to address the difficulty of accurately extracting weak edges or serious concave edges,and improving the automation level of waterline extraction.The experimental results show that the proposed method can effectively extract the weak edges and serious concave edges,and shows good characteristics such as sub-pixel accuracy,high efficiency and robustness.2.A breakwater extraction method from remote sensing image based on waterline is designed on the basis of the characteristics of the breakwater on the remote sensing image are analyzed,and the available features are summarized,after which various features are used to achieve the breakwater extraction.Experiments show that this method is feasible,and can achieve good results on breakwater extraction while even dealing with complex backgrounds.In addition,it shows high efficiency,strong robustness and simple operation.3.A coast dock extraction method based on multiple features is designed and a segmentation method based on the structural features of the coastal dock in remote sensing images is presented for the extraction of coast dock.The experiment results show that the first method can achieve good effect on coast dock extraction even when the background environments are complex.The second method can divide dock zones from waterlines,showing good performance for both breakout docks and the dock along the shore.4.To address the problem that ships and their background share similar patterns in color and texture and the situation of inshore ships docked side by side,an inshore ship detection method based on line segmentation detection and shape features is proposed.The experimental results show that ships can be accurately,stably and efficiently detected with this method,even under the situation that the texture and color of background environments are highly similar with the ships,or when ships are closely-docked.5.Two optional solutions for ship detection are designed for engineering practice.In the first one,inshore ships are detected according to the position difference between the coastline and the waterline.The second framework is based on the object-oriented information extraction software “eCognition”,in which rule sets are designed to detect ships according to their features.Experiment results show that both solutions are feasible.