Research On Video-based Sea Surface Moving Target Detection

Sea surface moving target detection technology is a critical component in sea surface target identification and behavior analysis with a wide range of applications, such as marine environmental protection, sea area monitoring, marine safeguarding rights and sea island protection, thus possessing important significance and broad application prospects.Video detection technology is an effective means of moving target detection with low cost, strong adaptability, high real-time and high spatial resolution advantages. On the base of deeply research on video-based moving target detection and detailed analysis of marine moving target detection, this paper proposed a new ship target detection algorithm, meanwhile, a fast and effective target detection method fused with AIS message is presented.The main work and innovation include the following two aspects:1. Combined with temporal-spatial correlation of video images, the parallelogram regional characteristics of a moving target at sea in time-stack images is put forward, and accordingly a ship target detection method is proposed. In the stage of image pre-processing to extract the sea area, a shoreline detection method based on L*a*b*color components and a sea-sky-line detection method based on color gradient are proposed. In the stage of target area detection, Hough transform combined with oblique line detection is adopted. In the stage of post-processing, median filtering and morphological featured are used to identify ship targets. The experimental results demonstrate that this method has a strong adaptability and anti-interference ability, especially suitable for low-contrast and small targets.2. A ship target detection method based on video images fused with AIS message is proposed. Firstly the ship location is determined according to AIS message, then an improved salient region detection method is used to get the saliency map, finally a double threshold algorithm is introduced in the ship target segmentation. The experimental results demonstrate that this method performs accurately and quickly in different conditions.In this paper, a detailed analysis of the characteristics of video-based sea surface moving target detection technology is made and an effective solution is proposed, which lay a solid foundation for the following research of sea surface target recognition and behavior analysis.