基于形态字典学习的复杂背景SAR图像舰船尾迹检测

SAR图像舰船尾迹检测不仅可用于反演运动舰船的航速航向信息，也有助于发现弱小舰船目标.然而现有舰船尾迹检测方法一般仅适用于简单海况背景下的SAR图像，复杂海况背景下的检测效果难以满足应用需求.本文提出一种基于形态成分分析与多字典学习的复杂背景舰船尾迹检测方法.该方法针对海况背景的复杂多变性以及舰船尾迹类型的有限性，通过离线学习方式构建海面纹理字典，通过解析方式构建尾迹结构字典并迭代更新，将图像分解为包含舰船尾迹的结构成分与包含海面背景的纹理成分，利用剪切波变换对结构成分高频系数重构以增强结构成分，并通过Radon变换对增强后的结构成分进行尾迹线检测.实验结果表明，本文所提方法对于复杂背景SAR图像舰船尾迹检测的效果明显优于现有方法.

Ship Wake Detection in SAR Images with Complex Background Using Morphological Dictionary Learning

Detection of ship wakes in SAR images is helpful not only in estimating the speed and the direction of moving ships, but also in finding small ship objects. The existing ship wake detection methods for SAR images can achieve satisfactory results only for simple background, but can hardly work for complex background. In this paper, a novel ship wake detection method for complex background based on morphological component analysis (MCA) and multi-dictionary learning. In this method, a SAR image is decomposed into a cartoon component containing ship wakes, and the process of the decomposition is supported by a ship wake dictionary built analytically and renewed iteratively. At the same time, the SAR image is also decomposed into a texture component supported by a sea-surface texture dictionary learnt off-line. Then, the cartoon component is enhanced by the shearlet transform and the high-frequency coefficient reconstruction. At last, the ship wake lines are detected from the enhanced cartoon component by Radon transform. Experimental results show that the performance of the proposed method outperforms other state-of-the-art methods for detection of ship wakes in SAR images with complex background.