摘要
为了辅助舰载直升机安全着舰,一种基于计算机视觉的助降技术被提出。该技术包括:利用图像分割技术从飞行甲板的背景中识别助降陆标图案;采用Harris角点检测器,提取助降陆标角点的投影;根据透视投影几何原理,构造助降陆标角点和图像上相应投影点的函数关系;采用最小二乘法解算函数关系矩阵;对该关系矩阵进行分解,即可得到直升机相对飞行甲板的位姿参数。仿真实验表明:该方法能够实时给出舰载直升机的相对位置和姿态信息,而且在图像检测误差的干扰下,位姿估计仍然保持着较高的精度。
For assisting shipboard helicopter in landing safely, an assisted landing technology for shipboard helicopter was proposed based on computer vision. The contents of the technology include image segmentation for identifying the landmark from desk, extracting corner of landmark in image with Harris corner detecting implement, establishing the function between corner of landmark and its projection point in image based on perspective projection principle, computing functional relation matrix with least square technique, and decomposing relation matrix to get relative position and orientation between shipboard helicopter and ship. The experiments with simulated data show that the relative position and the attitude data for shipboard helicopter can be estimated by the approach in real-time, which remain high accuracy still under the interference of image testing error.
出处
《兵工学报》
EI
CAS
CSCD
北大核心
2007年第3期370-373,共4页
Acta Armamentarii
基金
国家自然科学基金资助项目(60473141)
关键词
信息处理技术
舰载直升机
图像处理
计算机视觉
位姿估计
information processing
shipboard helicopter
image process
computer vision
position and orientation estimation
