摘要
针对传统的区域生长法需要人工交互找到种子点的缺陷,提出了一种结合视觉显著性和区域生长的红外目标提取方法。首先提出了一种基于频谱残差原理改进的视觉显著性模型,用于提取红外图像的显著图,接着对显著图进行简单的阈值分割进而获得二值图像,然后在对二值图像进行形态学处理的基础上根据目标的数量找到对应数量个最大连通区域,最后分别以各个连通区域的重心为种子点进行区域生长从而实现红外目标提取。通过仿真实验证明该方法能有效消除复杂背景的干扰,准确提取出红外目标。
Due to the requirements of traditional regional growth methods for selecting the seed points manually,a method is proposed to combine the visual saliency and the regional growth. First,based on spectral residuals,an improved visual saliency model is proposedto extract the saliency map of infrared image. Second,the simple threshold segmentation on the saliency map is used to obtain a binary image. Then,morphological processing of binary image is employed to find the top n（ representing the number of targets） largest connected areas. Finally,the center of gravity of each connected area as the seed point for regional growing and the infrared target extraction is achieved. The simulation experimentsshow that this method can effectively eliminate interference from complex backgrounds and accurately extract target.
作者
杨爽
徐宏宇
唐泽坤
YANG Shuang;XU Hong-yu;TANG Ze-kun(College of Electronic and Information Engineering,Shenyang Aerospace University,Shenyang 110136,China)
出处
《沈阳航空航天大学学报》
2018年第4期85-89,共5页
Journal of Shenyang Aerospace University
关键词
红外图像
目标提取
视觉显著性
频谱残差
区域生长
infrared image
target segmentation
visual saliency
spectral residual
regional growth
