Study On The Recognition System For Ship Flag Based On Video Stream

With the increase demands for the development and utilization of marine resource comprehensively, the domestic coastal shipping lanes become particularly busy. Due to the fierce competition for maritime rights and interests, the maritime affair management become difficult. Therefore, how to construct an autonomous, quick and accurate management system for marine vessel recognition is a hot research topic currently.To the demands of maritime affair management, some concerned issues such as the target detection and tracking technology, the target segmentation and recognition technology, digital video storage and implementation scheme are studied deeply. In order to obtain the quantity and sailing trajectory of ships, the real-time video processing algorithm is exploited. For realizing the maritime traffic governance and enhance the coastal supervision effectively in specified waters, the ships identity is verified by distinguishing the flags. By integrating the above-mentioned technologies, a marine vessel recognition system is designed based on video stream. The main research contents and innovations of this paper are listed as follows:1.To manage a mass of video datasets effectively, the ship video database based on Visual Studio is exploited, making the retrieval of video data and process result at specific time be more convenient. By employing ADO(Active Data Object) access technology to exploit video database based on the ACCESS, the ship video database of ship discrimination system is established, realizing some system functions such as data add, inquiry, search and delete. According to the follow-up ships detection and tracking, image real-time storage and update, it is convenient for the user to see, query, forensics, and discussion.2. To obtain the quantity of ships and extract the foreground of moving marine flag, moving target tracking based on background subtraction and block tracking is studied in this paper. According to the background subtraction algorithm, the background updating algorithm and the thresholding method integrated the fixed value and OTSU method are proposed to detect and track a moving marine target. After extracting the centroid and area features, the block tracking algorithm is employed to detect and track multiple targets, then the trajectory of the centroid is obtained, achieving quantity statistics of passing ships.3. In order to identify the detected ships, a target segmentation algorithm based on connected regions and color features and principal component analysis is proposed in this paper to extract the ship flag. After segmenting by median filtering and morphological operations, an elaborate segmentation is obtained based on connected regions area and the factors of color feature. Afterward, principal component analysis is employed to extract features and the nearest neighbor classifier is used to recognize the target flag. In this paper, to improve the accuracy of flag recognition, the interferences and relative solutions in recognition based on principal component analysis is discuss.In view of the above content, a marine vessel recognition system based on video stream is designed in this paper. The system design idea, overall structure, function modules and key technologies are detailed. The experimental results show that the proposed system can effectively achieve the quantity statistics and the identity verification when marine vessels are sailing.