红外视频图像运动弱目标检测方法研究

CORC > 沈阳自动化研究所 > 中国科学院沈阳自动化研究所 > 光电信息技术研究室

题名	红外视频图像运动弱目标检测方法研究
作者	马天磊
学位类别	博士
答辩日期	2016-05-26
授予单位	中国科学院沈阳自动化研究所
导师	尹健 ; 史泽林 ; 徐保树
关键词	红外弱小目标辐射累积目标检测空间反演运动模型红外噪声
其他题名	Research on Detection Method for Dim Moving Targets in Infrared Video
学位专业	模式识别与智能系统
中文摘要	本文以提高目标探测距离的实际需求为牵引，开展红外弱小目标检测方法研究。旨在从光电成像系统噪声干扰问题出发，解决弱小目标检测面临的难题。分析光电系统中的噪声特性，找出制约弱小目标检测性能的主要因素；分析远距离目标的运动模型和成像几何特征，找出目标在图像空间中的主要特征，进而提出用能量累积与空间反演方法解决远距离目标检测问题的思路。第一章介绍研究背景与意义，归纳总结了国内外红外弱小目标检测方法的研究现状。第二章介绍了与本文有关的红外辐射和红外成像系统知识，分析了红外成像系统中噪声的统计特性以及光学系统对目标成像的影响，进一步分析了信噪比与探测距离以及信噪比与检测概率之间的关系等基础理论问题。第三章提出了一种基于辐射累积与空间反演的弱小目标检测方法。分析目标在原始图像空间中的运动模型，建立目标的位置空间和运动空间，以运动空间中的不同运动矢量控制原图像序列能量累积，建立新的图像空间。基于Neyman-Pearson定理，采用恒虚警率判决法在新图像空间中检测准目标点。定义运动空间的体密度函数度量，给出由位置空间和运动空间局部极值判定目标点的目标确认方法。进一步由运动空间到原图像空间的反演，得到目标在原图像空间中的真实位置向量。针对匀速直线（Constant Velocity，CV）运动模型和匀加速直线（Constant Acceleration，CA）运动模型分别给出了对应的检测方法。第四章为提高检测方法的实施效率，设计并实现了一套基于FPGA的红外视频弱小目标检测方法，提高了检测的实时性。本文提出的检测方法不仅对CV运动模型和CA运动模型的目标有良好的检测效果，而且可扩展到更高阶的目标运动模型，方法具有普适性。
英文摘要	From the actual demand in improving detection range of dim target detection, this paper carries out the study of IR dim target detection method. The aim is to solve the noise jamming in electro-optical imaging system which severely restricts the dim target detection. By analyzing the noise characteristics in electro-optical imaging system, the dominant factor which restricts the dim target detection is found. By analyzing the kinematic model and geometric characteristics of dim targets in IR images, dim targets’ main characteristic in the image space is found. Further, a method, which is based on the radiation accumulation and space inversion, is proposed to detect extremely remote IR target. Chapter I introduces the research background and significance, and summarizes the research status in the field of IR dim target detection at home and aboard. Chapter II introduces the related and fundamental researches in the field of IR radiation and IR imaging system. The statistical characteristic of noise in IR imaging system is analyzed and the effect of optical system on remote target imaging is also analyzed in this chapter. Furthermore, the relationship between the signal to noise ratio (SNR) and the detection range and the relationship between SNR and detection ratio are analyzed. A radiation accumulation and space inversion based dim target detection method is proposed in chapter III. First, after analyzing the kinematic model of dim target in original image space, a target location space and a target motion space are established. Radiation accumulation operation, controlled by vectors from the motion space, is applied to the original image space. Then, a new image space is established. Second, based on Neyman-Pearson (NP) theorem, constant false-alarm ratio judging is implemented in the new image space to obtain quasitargets. Third, Volume density function is defined in the motion space. Target confirming method is given by determining local extremum of location space and motion space. Moreover, the inversion from the motion space to the original image space will confirm the actual location vectors of target in original image space. Finally, aiming at constant velocity (CV) kinematic model and constant acceleration (CA) kinematic model, detection methods are given respectively. In order to improve the implementing efficiency of the detection method, a FPGA based detection method for dim moving targets in IR Video is designed and achieved in chapter IV. The proposed method improves the real-time performance of the detection. The proposed method in this paper has an excellent detection results on targets with CV kinematic model and CA kinematic model. By extending the kinematic model of dim targets, the method can also deal with the detection of moving dim targets in high dimensional motion space. The method is universal for the detection of dim moving targets.
语种	中文
产权排序	1
页码	119页
内容类型	学位论文
源URL	[http://ir.sia.cn/handle/173321/19622]
专题	沈阳自动化研究所_光电信息技术研究室
推荐引用方式 GB/T 7714	马天磊. 红外视频图像运动弱目标检测方法研究[D]. 中国科学院沈阳自动化研究所. 2016.