一种稳健的空中目标跟踪方法

Robust Object Tracking for Realistic Air Target

提出了一种稳健的空中目标跟踪方法。该方法基于MAD构造了一种新的梯度特征相似度量算法,在梯度特征空间对目标进行匹配定位。为适应跟踪过程中目标的大小变化,利用自适应调整模板尺寸的方法在跟踪过程中调整目标模板大小,增强了对具有强机动特点的空中目标跟踪的稳定性。仿真结果表明,跟踪算法能够适应飞机在短期跟踪过程中由机动动作产生的快速形变,以及由形变带来的目标自身灰度上的剧烈变化和在长期跟踪过程中的大小变化,实现了对空中目标的稳定跟踪。

A robust method for tracking realistic air target with rapid changes in image sequences was proposed. A novel similarity measure of gradient features and adaptive window method was used to track air target with changes in viewpoint,pose,illumination and scale. By giving each pixel different power in target template,the similarity measure of gradient features was proposed based on MAD. And special attention was paid to adjust the size of target template while tracking. The basic idea of adaptive window method is to maintain the occupancy rate of the target gradient feature within a specified range. Experimental results using battleplane sequences confirm that the proposed algorithm is robust to the rapid changes in short term and the size change in long term of air target.