【作者】 汪大宝；

【导师】 刘上乾；

【作者基本信息】 西安电子科技大学 ， 物理电子学， 2010， 博士

【摘要】 红外成像目标检测与跟踪系统是一种基于被动探测技术的光机电一体化系统,它具有隐蔽性好、抗干扰能力强等优点,被广泛地应用于红外告警和精确制导等武器装备系统中。在实际应用中,为了尽可能多地增加火控系统的预警时间、提高己方的安全系数,要求红外探测系统能够在尽可能远的距离捕捉到目标并获取目标的相关信息。然而,当目标距离较远时,目标在视场中是以小目标的形态出现的,并且信号微弱,以至被淹没在复杂的背景之中,导致目标检测跟踪非常困难。因此,如何在复杂背景条件下对红外弱小目标实施稳定的检测与跟踪成为当今一项高难的前沿研究技术。另一方面,随着目标机动性能的不断提高,在系统性能指标中,对弱小目标检测跟踪的实时实现也提出了更高的要求。可见,研究复杂背景条件下红外弱小目标检测跟踪算法及其实时实现技术,不仅具有重要的理论意义,而且具有重大的工程实用价值。作者在深入分析国内外红外弱小目标检测跟踪研究现状和研究进展的基础上,结合在研国防科研项目要求,提出了一套复杂背景条件下实时高效的红外弱小目标检测跟踪系统实现方案。由于复杂背景是影响弱小目标检测跟踪性能的重要因素,因此,若要稳定而可靠地对目标进行检测和跟踪,就必须首先对红外弱小目标图像进行有效的预处理。为此,本文以国防科研项目中可能遇到的几种复杂背景的图像预处理技术为切入点,同时对复杂背景下的弱小目标检测、机动目标跟踪以及系统的软硬件实现等关键技术进行了深入的研究,并提出了相应的技术方案,理论分析和实验结果均验证了其合理性和有效性。论文的研究工作及成果主要有:首先,针对不同类型的复杂背景,研究了几种相应的背景抑制算法。对于视场中存在灰度远大于目标的大面积云层背景的情况,提出了基于马尔可夫随机场模型的正则化背景抑制算法;为了有效抑制纹理结构复杂且强起伏的云层背景,提出了一种基于Facet小面图像模型双向扩散滤波的背景抑制算法;此外,为了提高系统的实时性能,设计了基于自适应差分量化理论的背景抑制方法,实现了在采集图像的同时,完成了背景抑制。实验结果表明,上述算法能够有效抑制多种情况的复杂背景,并且结构简单,实时性好,均可作为工程应用的备选方案。其次,经过背景抑制后,虽然图像的信噪比得到显著的提高,但仍有可能残留较强的背景,传统的目标检测算法难以快速而准确地从图像中检测出弱小目标。为此,本文设计了基于图像等高线图特征匹配的弱小目标检测算法。实验结果表明,该算法具有检测概率高、适用范围广、实时性好的优点。再次,由于实战中被探测的目标通常具有较高的机动性能,其运动模型具有很强的非线性,为了对该类目标实施快速而准确的跟踪,本文提出了一种多速率的交互式多模型卡尔曼粒子滤波快速跟踪算法。实验结果表明,该算法显著地改善了算法的运算量,并在一定程度上提高了跟踪精度,是一种具有工程实用价值的跟踪算法。最后,针对科研项目的具体性能指标要求,设计了一套基于FPGA与多DSP的高速并行图像处理硬件平台,实时地实现了上述红外弱小目标检测与跟踪算法。同时,为了便于扩展该硬件平台的功能,还为其设计了并行图像处理实时操作系统(Parallel Image Process Real Time Operation System, PIPRTOS),用于实现系统算法软件的分布式处理。原理样机的外场实验表明,该系统不仅可实时地运行上述红外弱小目标检测与跟踪算法,而且具有良好的稳定性和可扩展性。所研究的算法和硬件平台已组装成了“复杂背景下红外弱小目标检测与跟踪技术”原理样机,并在某机场进行了多次外场实验,其功能和性能均达到了设计要求,验证了我们所研究的软硬件方案的正确性和有效性。更多还原

【Abstract】 Infrared imaging targets detection and tracking system is an optical mechanical and electronic integration system based on passive detection technology, which has the advantage of self-hiding, and strong capacity of resisting disturbance. It’s widely applied in infrared warning and precision-guided weapons systems. In the actual application, in order to increase the early warning time of the fire control system (FCS) and enhance the safety factor of our own side, infrared detection system is required to capture the targets and obtain relevant information at far range from the detector as much as possible. However, the longer the distance of targets, the less the imaging area of targets and weaker the targets’ signal will be. Therefore, the targets may be submerged into background. So, the question that how weak small targets can be steadily detected and tracked under complex background has become the forefront and highly difficult techniques. On the other hand, as the mobility of targets grow, the further development of the real-time implementation of weak small targets detection and tracking is necessary. Therefore, researches on infrared weak small targets detection, tracking technology and its real-time implementation technology under complex backgrounds have great significance both theoretically and practically.Based on thorough study on the present state and the latest progress of weak small targets detection, combining with our research projects, an implementation scheme of weak small targets detection and tracking system under complex background is designed. The key technologies in target detection, maneuvering targets, implementation of hardware and software are studied deeply to the point of developing effective weak small targets detection and tracking project under complex background. As a result, a series of suitable technologies whose feasibility and effectiveness have been verified by practice have been put forward.The main contributions of this dissertation are as follows:Firstly, aiming at different types of complex background, several kinds of background suppression algorithms are proposed. For the case that the field of view contains highlighted and large area background, a novel markov random field (MRF) based adaptive regularizing filtering algorithm is proposed. To effectively suppress strong undulant background with complex texture, a new filtering method based on cubic fact image model and bidirectional diffusion is introduced. Besides, to improve the real-time capability of the system, a new background suppression method which is based on adaptive differencing quantization and can accomplish image acquisition and background suppression simultaneouly is proposed. Experiment results indicate that the image pre-processing algorithms can efficiently suppress the complex background with the advantage of its logical structure simple to be implemented in real-time system.Secondly, benefit from background suppression process, is that the signal-to-noise ratio (SNR) of the images is improved to some degree. However some residual background still exits, as a consequence, traditional targets detection algorithms have some difficulties in detecting accurately and fast. So, a fast detection algorithm for small targets based on iso-elevation contour map (IECM) feature matching is designed. Theoretical analysis and experiment results show that this method can provide good detection performance and promote the real time capacity.Thirdly, since the actual targets usually have high maneuverability, which have strong non-linear motion equation, a fast tracking method focusing on the purpose of tracking this type of targets fast and accurately, which is based on multi-rate interacting multiple model (IMM) algorithm, is proposed. Experiment results show that the algorithm significantly reduces the computational complexity and improves tracking accuracy to some extent, which has practical value in engineering.Lastly, aiming at the performance indicators of research project, a high-speed and parallel image signal processing system based on FPGA and multi DSP is presented, which can detect and track small targets in real-time. Meanwhile, to facilitate expanding the functions of system, parallel image process real time operation system (PIPRTOS) is designed for this hardware platform to accomplish the distributed processing of the software. Experiment results show that the system can not only detect and track small targets in real-time, but also have good stability and scalability.The proposed algorithms and hardware platform have been assembled into“infrared small targets detection and tracking technology”principle prototype, and several prototype field experiments has been done in certain airports. Experiment results show that the function and performance of the system have met the design requirement, which verifies the accuracy and effectiveness of proposed hardware and software programs.更多还原