摘要
基于仿生双目理论提出一种仿双眼异向运动的变视轴夹角双目视觉系统,通过对实时采集的目标图像的处理,研究其双目摄像机夹角的在线自标定,实现获取无人旋翼机的实时高度信息。实验结果表明,该方法获得的光轴夹角信息相对误差小于5%,能准确控制双目摄像机的转动,避免由于双目视觉盲区所造成的着陆误差,系统所标定出的旋翼机高度信息也为其安全着陆提供了有力保障。
Based on the binocular stereo vision theory,a binocular vision system is presented,whose axis angle can be changed real time.Through dealing with real-time acquisition of the target image,the axis angle of stereo vision sensor can be calibrated,and the real-time height of unmanned helicopter can be achieved.Experimental results show that the calibration error of the axis angle is less than 5%.It can control stereo vision sensor rotate accurately,avoid the landing error caused by the blind spot of the binocular vision,and provide the security for the unmanned helicopter autonomous landing.
出处
《计算机工程》
CAS
CSCD
北大核心
2010年第19期193-194,197,共3页
Computer Engineering
基金
国家"863"计划基金资助项目(2007AA04Z225)
国家自然科学基金资助项目(60605028)
关键词
无人旋翼机
仿生视觉
自主着陆
摄像机标定
unmanned helicopter
binocular vision
autonomous landing
camera calibration