(英)基于横纵多尺度灰度差异加权双边滤波的弱小目标检测
投稿时间:2019-08-21  修订日期:2019-12-29  点此下载全文
引用本文:
摘要点击次数: 170
全文下载次数: 3
作者单位E-mail
朱含露 中国科学院上海技术物理研究所 中国科学院智能红外感知重点实验室上海 200083
中国科学院大学 北京 100049 
zhuhanlu19920@163.com 
张旭中 中国科学院湖州应用技术研究与产业化中心 浙江 湖州 313000  
陈忻 中国科学院上海技术物理研究所 中国科学院智能红外感知重点实验室上海 200083  
胡亭亮 中国科学院上海技术物理研究所 中国科学院智能红外感知重点实验室上海 200083  
饶鹏 中国科学院上海技术物理研究所 中国科学院智能红外感知重点实验室上海 200083  
中文摘要:为了有效地检测复杂背景下的红外弱小目标,提出了一种基于横纵多尺度灰度差(HV-MSGD)的方法来增强弱目标,并通过距离和像素差异来实现对背景强边的抑制。目标区域与周围区域之间存在不连续性,为了加强它们的差异,HV-MSGD与双边滤波(BF)相结合,可以在抑制背景的同时提高目标强度。进一步通过自适应局部阈值分割和全局阈值分割来提取候选目标。为了进一步验证对单帧检测的影响,将上述单帧检测算法与改进的无迹卡尔曼粒子滤波器(UPF)相结合,实现轨迹检测。实验结果表明,该方法在弱信噪比(SNR)下优于其他方法,在抑制背景的同时可以增强目标,增强效果是其他方法的6-30倍。在实验中,输入信噪比分别为2.78,1.77,1.79,1.13和1.16。图像处理后,背景抑制因子(BSF)分别为13.48,21.33,11.73,20.63和121.92,信噪比增益(GSNRs)分别为40.09,71.37,27.53,12.65和131。该方法的检测概率(Pd)也优于其他算法。当误报率(FAR)为, , , ,计算五组真实序列图像的Pd为94.4%,92.2%,91.3%,95.6%和96.7%。
中文关键词:弱目标检测  多尺度灰度差  距离和像素差  局部阈值分割  全局阈值分割
 
Dim Small Targets Detection based on Horizontal-Vertical Multi-scale Grayscale Difference weighted Bilateral Filtering
Abstract:We propose a single-frame method for the effective detection and tracking of targets that emit weak-intensity infrared radiation in images with a complex background. The method is based on horizontal-vertical multi-scale grayscale difference (HV-MSGD) to enhance the intensity of the target and strong edges in the background and uses the difference between the distance and grayscale value of a pixel. The discontinuity between the target and the surrounding area is exploited to enhance these differences. HV-MSGD combined with bilateral filtering (BF) is used to improve the target intensity while suppressing the background and the candidate target extracted by adaptive local and global threshold segmentation. The single-frame detection algorithm is combined with an improved unscented Kalman particle filter (UPF) to implement trajectory detection to verify the effectiveness. The experimental results show that the proposed method is superior to other methods for weak signal-to-noise ratio (SNR) detection, and can enhance the target by 6-30 times (compared to other methods) while suppressing the background. After image processing, the background suppression factors (BSF) values were 13.48, 21.33, 11.73, 20.63, and 121.92, and the gains of signal-to-noise ratio (GSNRs) were 40.09, 71.37, 27.53, 12.65, and 131, respectively. The probability of detection (Pd) of this method was also superior to that of other algorithms. The Pd values of five sets of real sequence images were counted, when the false alarm rate (FAR) values were , , , , and , the Pd values were 94.4%, 92.2%, 91.3%, 95.6%, and 96.7%.
keywords:dim target detection  multi-scale grayscale difference  distance and pixel difference  local threshold segmentation  global threshold segmentation
  HTML  查看/发表评论  下载PDF阅读器

版权所有:《红外与毫米波学报》编辑部