【作者】 王朝辉；

【导师】 王岳环；

【作者基本信息】 华中科技大学 ， 模式识别与智能系统， 2016， 硕士

【摘要】 红外成像地面场景运动目标的跟踪在军事等领域应用很普及。由于地面场景很复杂、干扰具有很强的随机性、目标先验信息较为缺乏及目标运动很复杂,使得红外图像中地面背景下运动目标的跟踪成为国内外学者的研究热点。本文研究了地面场景下的运动目标的跟踪方法,主要工作如下:首先介绍了红外成像地面运动目标跟踪系统组成结构,分析了红外图像的特性,研究了几种抑制红外图像中噪声的方法。在缺少目标先验信息的情况下,提出了一种基于混合比和大津阈值分割的自适应模板选择方法。其次,在跟踪阶段,本文采用一种特征关联与模板匹配结合的运动目标跟踪方法。根据自适应模板选择阶段获取的目标信息是否与边界相关而采取相应的跟踪方法。当自适应模板选择阶段获取的目标信息与边界无关时,通过自适应加权系数的多特征融合方法对目标进行跟踪。当自适应模板选择阶段获取的目标信息与边界相关时,通过Kalman滤波与模板匹配相组合的方法对目标进行跟踪。另外,对跟踪波门内与目标相似的物体的跟踪方法进行了研究,本文在基于Kalman滤波与模板匹配的方法的基础上实现了对目标及与目标相似的物体的跟踪。然后,本文对跟踪阶段匹配模板的更新方法进行了研究。为了减少跟踪过程中匹配模板的漂移,使匹配模板更好地适应目标的变化,本文采用了一种自适应模板更新策略,保证了跟踪过程的准确稳定。最后,本文对红外成像下地面运动目标跟踪系统组成模块进行了介绍,同时简要介绍了PC端的系统调测软件,以及算法移植所做的部分修改与优化。更多还原

【Abstract】 The detection and tracking of infrared moving targets is widely used in many fields, such as military, surveillance system and so on. As the complexity of the ground background and the uncertainty of interference, it makes tracking of ground infrared target become the focus of scholars at home and abroad. The thesis studied the tracking of complex ground infrared target. The main achievement of this dissertation is as follows:First of all, we introduce the compositional structure of our infrared ground moving target tracking system,the characteristics of ground infrared image are deeply analyzed and studied, an algorithm of suppress noise method is researched and implemented. As the lack of target prior information, an adaptive template selecting method based on mixture ratio and Otsu threshold is proposed.Secondly, In the tracking stage,a moving target tracking method based on feature association and template matching is adopted. We select the corresponding tracking method according to the completeness of target information we get during the stage of adaptive template selection. When the target is isolated from the border, an tracking method based on adaptive weight multiple features fusion is presented. While the target is associated with the border, we improve the tracking method based on Kalman Filter and image matching. Additionally, we studied the tracking method for tracking target and similar objects, we use the tracking method based on Kalman Filter and image matching to track tracking target and similar objects.Thirdly, the updating method of template is studied..In order to reduce the drift of the matching template in the tracking process and make the matching template better adapt to the changes of the target, we use an adaptive template updating strategy to ensure the accuracy and stability.At last, we introduced the modules and state partition of Infrared ground moving target tracking system. Meanwhile, we will briefly introduce the system debugging software under the environment of PC and modification and optimization we make in the period of algorithm transplantation.更多还原