基于QP_TR信任域方法的低信噪比序列图像目标跟踪*

运动平台上低信噪比序列图像中的目标跟踪面临着两大困难:平台运动导致图像存在全局平移,使得目标在相邻帧间脱离跟踪算法搜索窗;图像中的干扰使得跟踪窗口经常跳动而导致跟踪失败.鉴于QP_TR信任域算法的优良性能,针对上述两个问题提出了一种新的基于QP_TR信任域和Kalman滤波的跟踪算法.该算法利用QP_TR进行图像稳定和模板匹配,通过Kalman滤波器状态估计滤除干扰.与三步搜索方法相比,加大了搜索窗大小的同时减少了模板匹配的次数,提高了性能.在真实图像序列上进行的实验表明,该算法能有效地稳定运动图像,实现运动平台上低信噪比序列图像中目标的稳定跟踪.

Novel tracking algorithm based on QP_TR method for low S/N ratio image sequences

Two problems exist in a tracking system for low S/N ratio image sequences on moving platforms. The global shift introduced by ego motion often makes targets fall outside the search region. The noise in the images can distract the tracker, This paper proposed a new QPTR based tracking algorithm, which solved the above-mentioned problems by the combination of Kalman filtration. Image stabilization and template matching were done with QPTR, while the noise was eliminated by targot state estimation. Compared to the TSS, the new method had better performance in that it enlarges the search region and reduces the number of template matching operation at the same time, Experiments on real image sequences show that the proposed method can stabilize the sequences very well and track targets steadily despite of the ego motion and noise.