A Study on Sea Surveillance System with Flapping Reference Image

Our goal is to realize sea surveillance systems in which an image processor detects intruding objects such as boats in the sea. A difficulty of the sea surveillance systems is to detect moving objects such as boats inhibiting false detection caused by flapping, splashing and shiny surface of the sea. Firstly we pick up factors of false detection in sea surveillance. Secondly we propose a new surveillance system considering these factors. Our proposed system includes three methods for detecting boats of various speeds with inhibiting false detection in various condition of the sea such as heavy waves, splashing waves and shiny waves against a light. The three methods are detection of Differences between Surveillance images and a Reference image, abbreviated by DSR, and detection of Contours from Averaging images, abbreviated by CA and Silhouette Detection, abbreviated by SD. The proposed system detects boats of various speeds by DSR and CA under normal lighting conditions, while by SD under backlighting conditions. The proposed system decides the lighting conditions and select the three detecting method automatically. Lastly we discuss performances of our proposed system about actual sea images. Then we can show a successhl performance of it about the actual sea.