【作者】 梁敏；

【导师】 张科；

【作者基本信息】 西北工业大学 ， 导航、制导与控制， 2007， 硕士

【摘要】 随着红外技术的迅猛发展，红外弱小目标的检测与跟踪技术成为了红外成像系统中的一项核心技术。该技术是在红外图像处理的基础上，利用某种图像处理算法对杂乱背景和噪声环境下的目标进行自动检测与跟踪，算法性能的优劣对红外成像系统的作用距离大小和智能化程度至关重要。本文对红外制导的基本特征和弱小目标的检测与跟踪技术进行了系统研究，取得了一定的成果。其主要工作如下： 1．结合热辐射原理，在对红外序列图像建立数学模型的基础上，对红外序列图像和序列图像中弱小目标的特点进行了系统的研究和总结，并结合红外探测器本质，对红外序列图像中的噪声特点进行了研究和总结。 2．对红外图像预处理和背景抑制技术进行了研究，并针对传统红外图像分割算法精度低，及单纯基于边界信息或区域信息Level Set的图像分割算法实时性差等问题，提出了一种同时基于边界信息和区域信息的快速Level Set的红外图像分割算法。通过自适应系数T1和T2，较好的将边界信息和区域信息相结合，在保证分割精度的同时加快了图像分割速度。通过实验仿真并与其它红外图像分割结果相比较，证明了该算法的有效性和实时性。 3．对以往弱小目标检测技术进行了回顾和总结，并基于对红外图像中目标和背景特性的分析，提出了一种新的基于标记信息和梯度信息相融合的弱小目标检测方法。在红外图像空间，根据目标的特性，利用双重窗进行可能目标的搜索，并进行标记；同时并行的对原始图像进行梯度计算。对标记信息和梯度信息进行融合，进行目标的初次确定。再利用自适应门限对融合后的图像进行目标分割。实验结果证明了该方法能有效的检测出弱小目标。 4．对目标跟踪理论进行了系统的研究与分析，并针对跟踪过程中，由于目标机动的随机性，可能出现目标运动模型建立不准确的缺陷，提出了一种能更精确的反应目标机动信息的运动模型；并且针对弱小目标跟踪容易丢失的情况，提出了一种融合红外数据和毫米波数据的目标跟踪技术。将红外和毫米波两种传感器的工作频率设成互补状态，用红外传感器进行测角，毫米波传感器进行测距，并分别对观测空间进行独立赋值，然后对两传感器系统所获得的赋值数据利用D-S证据理论进行融合，最后进行了实验仿真。利用融合后更加精确的数据进行目标关联，形成航迹。该技术可以实现全天候工作。更多还原

【Abstract】 With the rapid development of IR imaging and its widest prospect in military application, the IR image processing technology has been the research hotspot in relative professional field. Infrared small dim moving target detection technology, one of the key techniques of infrared imaging system, adopts imaging processing algorithm whose performance is critical for action distance and intelligence of infrared imaging system, to detect target automatically which is under random background and heavy noise. In this paper, we studied on the IR guidance essential features and the limitation in the technology of detection and track for the IR small dim targets, and has got certain achievement. The main work as follows:1. Combine to the principle of heat radiation, we studied and summed on characteristic of the small dim targets in array IR image based on the mathematic model of the array IR image systematically. And union the IR detector nature, have been in progress to noise characteristic in IR array image studying and summing up.2. To infrared image, the pretreatment and background restrain a technology having been in progress studying. And aimed at poor precision of the traditional infrared image segmentation and the poor real time application of the Level Set method that is only based on boundary information or area information .A new Level Set method, based on boundary information and area information simultaneity ,used for infrared image segmentation was proposed . The boundary information and area information are combined excellently by using adaptive coefficient T1 and T2, and at the same time, the precision of image segmentation is guaranteed and the time-consuming is also reduced .The simulation results and comparison with other methods show that this method can segment the infrared image not only accurately but also practically.3. Have been in progress to target detecting in the past small and weak technology recollecting and summing up. And based on the analysis of the targets and background in IR image .An method of small dim targets detection based on fusion of the mark information and gradient information was proposed . In IR image space, according to targets’ characteristic, dual-window was used to search and mark possible targets .Calculate the gradient of originality IR image simultaneously.更多还原