基于形态字典学习的复杂背景SAR图像舰船尾迹检测

发布时间：2018-05-28 01:15

本文选题：SAR图像 + 舰船尾迹检测　；参考：《自动化学报》2017年10期

【摘要】：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.
[Abstract]:Ship wake detection in SAR images can not only be used to retrieve the heading information of moving ships, but also to find small and weak ship targets. However, the existing methods of warship wake detection are generally only suitable for SAR images in the background of simple sea conditions, and the detection results under complex sea conditions are difficult to meet the requirements of application. In this paper, a complex background ship wake detection method based on morphological component analysis and multi-dictionary learning is proposed. Aiming at the complex variability of sea background and the finiteness of ship wake type, the method constructs a sea surface texture dictionary by offline learning, constructs a wake structure dictionary by analytic method and updates iteratively. The image is decomposed into the structural components containing the ship wake and the texture components containing the sea background. The high frequency coefficients of the structural components are reconstructed by shear wave transform to enhance the structural components. Radon transform is used to detect the wake of the enhanced structure. The experimental results show that the proposed method is more effective than the existing methods in detecting ship wake in complex background SAR images.
【作者单位】：清华大学电子工程系;北京市遥感信息研究所;北京工业大学计算机学院;
【基金】：国家自然科学基金(61501008) 首都卫生发展科研专项(2014-2-4025)资助~~
【分类号】：TN957.52

【相似文献】