摘要
给出了一种以图像识别技术为基础的无人机自动着落导航的方法;采用DirectShow解码视频文件得到了连续的数字图像序列,同时设计了计算机同实际模拟图像信号的硬件接口,使系统很容易被扩展成实际的图像处理系统;为了提高计算机的使用效率和交互性,采用多线程技术设计了在Windows环境下的图像识别实验平台,并以某型无人机光测图像序列作为实验对象,实验结果表明本文设计的识别算法有较好的精度和较快的识别速度,为无人机自动着落导航的最终实现创造了较好的条件。
A method is presented used for unmanned aircraft＇s auto-landing navigation based on the technology of image recognition. By using DirectShow, the video file is deoded to get the continuous image sequence. Furthermore, a hardware interface that can transform the real analog image signal to digital signal is added to the system. Therefore, the system can be converted to a image processing system. In order to improve the computer＇s efficiency and alternation, a multi-threaded program technology is adopted in Windows environment to design the image recognition simulation platform, and a experiment based on an unmanned aircraft＇s optical image sequence is designed as well. The result of the experiment showed that the recognition algorithm has a high performance, this provided a good condition for the craft to land automatically.
出处
《计算机测量与控制》
CSCD
2006年第8期1093-1095,共3页
Computer Measurement &Control
关键词
图像识别
视觉导航
无人机
多线程
image recognition
visual navigation
unmanned aircraft
multi- threaded
