摘要
采用基于视觉信息的姿态估计方法对微小型无人直升机位姿估计进行了研究.在分析视觉信息与直升机运动关系的基础上,可以看出,稳定解算航拍图像序列之间的单应矩阵是整个方法的关键.而解算单应矩阵需得到序列图像之间的特征匹配点.为了获得稳定的、抗噪性强的同平面特征匹配点,方法采用了基于尺度不变特征变换(SIFT)算法和基于随机抽样一致性(RANSAC)算法的匹配策略.在一套真实的微型无人直升机系统上的实验证明,通过该方法得到的位姿信息可以达到无人直升机自主飞行所需的精度要求.
The vision-based pose and position estimation approach was proposed for mini unmanned helicopter (MUH). The relationship between vision information and MUH movements was analyzed. The ho- mography between the sequential images captured from MUH was the key of the approach. The homography was solved according to the series of matching feature points between the sequential images. The scale invariant feature transform (SIFT) features and a matching strategy based on random sample consensus (RANSAC) algorithm were employed to obtain the stable and anti-noise points in the same plane. The experimental results of real flights verified that the approach can meet the MUH flight precision needs.
出处
《浙江大学学报(工学版)》
EI
CAS
CSCD
北大核心
2009年第1期18-22,共5页
Journal of Zhejiang University:Engineering Science
基金
国家"863"高技术研究发展计划资助项目(2006AA10Z204)
关键词
微型无人直升机
位姿估计
尺度不变特征变换
mini unmanned helicopter (MUH)
pose and position estimation scale invariant feature transform (SIFT)