基于增量式子空间学习的红外目标跟踪研究

Study of Infrared Target Tracking Based on Incremental Subspace Learning

在对红外图像目标进行跟踪时,由于图像成像质量差致使目标的跟踪较为困难,采用一种基于增量式子空间学习的视觉跟踪方法,有效解决了红外图像中背景、目标运动方式复杂等问题;该方法基于粒子滤波,利用增量式主成分分析方法增量地学习一个有效的低维特征空间来适应目标外部结构的变化,鲁棒的跟踪红外图像中的目标;实验结果表明,该算法在具有复杂条件的红外图像中能够实现鲁棒的目标跟踪,跟踪成功率高达95%以上。

In the infrared image target tracking, due to poor quality of image formation the target tracking is very difficult, using a visual tracking methods based on the incremental learning space, effectively solved the infrared image problem in complex background or the complicated way of target motion. The method based on particle filter, using the Incremental Principal Component Analysis, incremental learning an effective low dimensional eigenspace representation to adapt to the changes in the external structure of target that expressed to follow up on the results, in order to robust track target in infrared image. The experimental results show that the algorithm in complicated conditions can be well implemented when tracking target of the infrared image.