A Method for Pose Estimation of Ship Target from SAR ROI Based on Ellipse Fitting
Pose estimation of target ROIs (region of interest) is an essential approach in Synthetic Aperture Radar (SAR) image interpretation. It brings high efficiency and fine accuracy in the process of target recognition and classification with accurate pose estimation. In this paper, firstly, the typical methods of target pose estimation are introduced, and the advantages and disadvantages of them are analyzed. Then, focusing on ship targets in SAR images, a novel method based on ellipse fitting is proposed to estimate the pose of the ship. According to the fact that the contour shape of ship target is generally similar to an ellipse in medium resolution SAR images, the principal axis of the fitted ellipse can be used to estimate the attitude angle of the ship. The ship target ROI is separated from the background and the contour of the target is extracted to fit an ellipse through a least squares ellipse fitting method. Although the contour of the ship target is generally coarse, the ellipse can be stably achieved by the complex real contour and a minority of singular points does not affect the final result significantly. At last, a number of experiments with ship target ROIs from TerraSAR-X SAR data are explored, and the mean error is less than 1.0°, which shows the robustness and validity of the proposed method in ship pose estimation with SAR images.
Xiaoqiang Zhang Boli Xiong Ganggang Dong Gangyao Kuang
National University of Defense Technology, China
国际会议
Progress in Electromagnetics Research Symposium 2014(2014年电磁学研究新进展学术研讨会)
广州
英文
1077-1080
2014-08-01（万方平台首次上网日期，不代表论文的发表时间）