摘要
主要介绍了一种利用飞行器的可见光图像进行目标跟踪的方法;针对所研究图像的特点,对图像中含单多机情况选取不同的分割方式,运用形态学上的腐蚀、膨胀和标注等方法,确定出被跟踪飞行器的形心(即灰度质心),并用GM(1,1)灰色预测模型进行形心预测;然后根据给定的摄像机参数,得出飞行器实际形心的偏移量以及摄像机的偏转信息;仿真结果表明,该方法能够有效地识别图像中的目标,跟踪精度较高,具有良好的准确性和适应性。
An image sequence--based method for aircraft tracking is proposed. Different ways of segmentation are used for single and multi--air- crafts respectively after analyzing the studied images. Erosion,dilation in morphology and labeling for connecting region are used subtly to extract the eentriod of each aircrafts in process of multi-- aircrafts＇ tracking. The gray model GM （1,1） is used for predicting. The deviation of the aircrafts＇ actual center and the deflection of the video camera are acquired according to the given camera parameter. The results show that the algorithm is effective and adoptable.
出处
《计算机测量与控制》
CSCD
2008年第5期649-650,676,共3页
Computer Measurement &Control
关键词
飞行器图像
计算机视觉
目标跟踪
图像处理
形心
aircraft image
computer vision
object--tracking
image processing
centroid
