弹射试验中的目标识别与跟踪

Target Recognition and Tracking in Test of Ejection

在弹射座椅测试中,为了更好地对图像数据进行分析处理,需要对火箭橇和飞行员进行全程识别跟踪。首先采用帧间差分法以及形态学滤波得到火箭橇的质心坐标,以质心坐标确定初始搜索窗口后利用Camshift算法对火箭橇进行跟踪。采用指定区域比较颜色直方图的方法来检测弹出飞行员,并且完成对其跟踪。利用搜索框的大小来判断目标分离,当飞行员与座椅分离后,采用Opencv中块识别函数对飞行员及座椅进行识别,识别后再利用Camshift算法完成后续跟踪。实验结果表明,整个识别跟踪过程均达到良好效果。

In the test of ejection seat,to further analyze the sequence, several algorithms are implemented to track the rocket sled and the pilot in this paper.Inter-frame difference algorithm and morphology filters are used to acquire the center of rocket sled.With the establishment of initial search window,Camshift algorithm can well track the rocket sled.A rectangular window is set to check the color histogram of the area which is used to capture the ejected pilot.