摘要
折反射全景视觉导航是自主移动机器人研究领域一项新兴技术,但由该视觉传感器生成的原始图像是严重畸变的,不适合机器人的目标定位和跟踪等任务。基于此,提出一种全景图像中心定位方法,将双线性内插值法应用于图像的采样,实现双曲面折反射全景图像的柱面还原解算。实验结果显示,还原后的图像人机交互性能有一定的提高。
Navigation with catadioptric omni-directional vision is a new technology in the domain of auto-mobile robot, but the image getting from catadioptric sensor is distorted and not suitable for target locating or tracking, so cylindrical calculation should be employed on the original omni-directional image. In this paper, one kind of method to find the center coordinates of the omni-directional image is presented, which is the base for cylindrical calculation, and quadratic interpolation method is used in the process of cylindrical calculation. According to the experiments, an improved cylindrical calculated image is more suitable for human-computer interaction.
出处
《计算机工程》
CAS
CSCD
北大核心
2010年第12期204-206,共3页
Computer Engineering
基金
江苏省"六大人才高峰"基金资助项目(07-E-024)
关键词
折反射
全景图像
中心定位
柱面解算
catadioptric
omni-directional image
center location
cylindrical calculation
