基于视觉注意机制的UWB SAR叶簇隐蔽目标变化检测

发布时间：2018-03-09 09:54

本文选题：低频超宽带合成孔径雷达　切入点：叶簇隐蔽目标检测　出处：《电子学报》2016年01期 　论文类型：期刊论文

【摘要】：在超宽带合成孔径雷达叶簇隐蔽目标检测中,传统的UWB SAR图像变化检测方法易受图像灰度值起伏和成像条件变化的影响,致使现有的变化检测算法的性能下降.本文根据人类视觉系统的生理结构和认知特点,提出了一种基于视觉注意机制的叶簇隐蔽目标变化检测算法.该方法使用视觉注意模型,将图像的多尺度特征信息融合为单幅视觉显著图像,并利用图像局部邻域信息和目标的空间相关特性对视觉显著图中视觉注意焦点进行分层筛选和变化检测.实验结果表明:本文中基于视觉注意机制的变化检测方法可以有效检测多时相UWB SAR图像中的叶簇隐蔽目标,较之传统的基于统计原理的变化检测方法,其检测速度更快,且对场景复杂的UWB SAR图像亦具有鲁棒性.
[Abstract]:In UWB synthetic Aperture Radar (SAR) leaf cluster covert target detection, the traditional UWB SAR image change detection method is easily affected by the fluctuation of image gray value and the change of imaging conditions. According to the physiological structure and cognitive characteristics of human visual system, a new change detection algorithm based on visual attention mechanism is proposed in this paper, which uses visual attention model. The multi-scale feature information of the image is fused into a single visual salient image. The local neighborhood information of the image and the spatial correlation characteristics of the target are used to screen and detect the visual attention focus in the visual salient image. The experimental results show that the change detection method based on the visual attention mechanism in this paper is based on the visual attention mechanism. The method can effectively detect the covert targets of leaf clusters in multitemporal UWB SAR images. Compared with the traditional change detection method based on statistical principle, its detection speed is faster, and it is also robust to UWB SAR images with complex scene.
【作者单位】：国防科学技术大学电子科学与工程学院;空军预警学院信息对抗系;
【基金】：国家自然科学基金(No.61302194;No.61302146)
【分类号】：TN957.52

【参考文献】