红外序列图像帧间匹配算法研究

文章在现有单帧图像检测算法的基础上,提出了一种可同步跟踪运动目标的帧间匹配算法。算法采取对相邻两帧间像素点匹配,匹配过程与当前两帧外的其余帧无关,把匹配结果兴趣区域像素点对应的另一个全零矩阵位置值赋为1,顺利输入检测后的单帧图像,执行匹配操作,实时形成目标运动轨迹。该算法结构简单、存储量少、计算速度快,硬件平台要就低,可有效地完成对多个目标、交叉目标、迎头目标的检测与跟踪。根据仿真和实拍照片的试验,证实了算法理论上的优点,有较高的应用价值。

Research on an Algorithm of Matching between Infrared Image Sequences

Based on the research of many target detection algorithms used for single-frame image,this article presents an algorithm of matching between infrared image sequences.The algorithm merit lies in adopting to two adjacent frame images element match,the current match process has nothing to do with other frames;In the match result,we can obtain the interest region pixel point,then charge the corresponding pixel value of the pixel points in another entire zero matrix to 1;Inputs the pretreatment binary images gradually, performs matching operation and forms real time target trajectory synchronously. Especially, the match process does not change by the moving state or the number of targets, and the match process is no different between a single target and multi-targets, cross-targets, head-on targets. According to the simulation and the real images experiment, we have already confirmed the above-mentioned merits. The algorithm has the merits such as simple structure, less storage, high computing speed, so it has the high application value.