Design And Construction Of A Sea Ice Monitoring System Based On Intelligent Video Technology

There are a lot of offshore oil platforms in Liaodong bay around Bohai Sea of China. However, because of the Siberian cold air,the phenomenon of different degrees of freezing appears every year during the winter in Liaodong bay. In the worst cases,the average thickness of ice was up to 30 cm and the sea ice could cause serious problems for offshore production activities and marine engineering facilities in the affect of wind, wave, tidal and other external forces. With the increasing of offshore oil exploration, it is required to fully consider the impact of sea ice on the oil platform, to strengthen the monitoring and forecast of sea ice, to understand the process of change of ice and to draw up preventive measures.Although satellite remote sensing and aircraft aerial has become the primary means of sea ice surveillance, it is difficult to uses these methods to extract the detailed spatial and temple information, such as the movement parameters of sea ice and sea-ice thickness, because the spatial resolution of the images are usually about several meters to several kilometers. On the other hand, in situ sea ice monitoring method based on digital camera technology could be used to obtain the higher resolution local scale sea ice characteristics than remote sensing image and aerial photographs. Therefore, it provides a means to supplement for large-scale space remote sensing data. This thesis is aimed to construct a sea ice imaging system based on high-resolution camera system and intelligent analysis methodology. It is focused on the image processing algorithms for sea ice type, ice thickness and ice drift speed and direction detection analysis around Bohai oil field platform. On the Measurements of ice movement, the method of Pyramid light flow estimate based on Harris corners to detect sea ice drift speed and direction automatically was implemented. On the measurements of ice thickness, a semi-automatic detection method based on image segmentation algorithm of Image entropy was presented. On the ice type Identification, an unsupervised learning method based on cluster analysis for sea ice recognition was proposed. On the measurements of ice density, it was based on the density statistics in camera's field by Image distortion correction based on principles of photogrammetry. The analysis software was developed with VC++, OpenVC and Matlab and the system has been installed on JZ20-2 platform of CNOOC in the Bohai Sea in Winter 2010 to carry out continuous observation of sea ice for three months. The practical results show that the developed system has the reliable performance for image upload, stable performance for image processing. It indicated that the system meets the application requirements of offshore oil platform and it forms the foundation for the development intelligent monitoring systems of Bohai Sea sea ice environment.