摘要
该文提出一种基于频谱残留变换的红外遥感图像舰船目标检测方法。该方法首先根据海洋红外图像中自然背景和干扰的特性设计频谱残留变换的模型参数;然后对海洋红外图像实施频谱残留变换;最后在变换图像上进行目标检测。实验结果表明:该方法可以有效消除红外图像中的大尺度干扰和图像噪声,增强图像中舰船目标的信杂比,提高舰船检测的准确性。
A ship detection algorithm based on spectral residual transform is presented to detect ship in infrared remote sensing images. Firstly, the model parameters of spectral residual transform are designed according to the prior knowledge of ship and its natural backgrounds. Secondly, the spectral residual transform of sea infrared image is implemented. Thirdly, ship detection is done on the spectral residual transform image. Experimental results reveal that the new detection algorithm can remove large scale image interference and the image noise and improve the SCR of ship image. The detecting probability of the new algorithm is higher than other conventional methods.
出处
《电子与信息学报》
EI
CSCD
北大核心
2015年第9期2144-2150,共7页
Journal of Electronics & Information Technology
基金
国家自然科学基金(61101185
61302145)
国家专项课题(0404040604)资助课题
关键词
红外遥感图像
舰船目标检测
信杂比
视觉显著性
谱残留模型
Infrared remote sensing images
Ship target detection
Signal Clatter Ratio(SCR)
Visual saliency
Spectral residual model