基于立体视觉和光流的无人机避障算法研究
发布时间：2018-09-17 09:07
【摘要】：无人机在飞行过程中必须能够检测障碍物并且准确躲避。为使无人机在飞行过程中更加安全,提出了一种双目立体视觉和光流相结合的避障方法。双目立体视觉通过边缘索引算法来获取可靠的视差值,并根据视差线汇聚角度得到空间深度信息,进而辨别物体的远近;基于SIFT的光流法能得到障碍物相对于摄像头的每一个时刻的运动速度。为了更快地得到更加准确的位置信息,将立体视觉和光流结合在一起。实验结果表明,该方法能有效提高避障的效率和精度。
[Abstract]:UAVs must be able to detect obstacles and avoid them accurately during flight. In order to make UAV more safe in flight, a new obstacle avoidance method combining binocular stereo vision and optical flow is proposed. Binocular stereo vision obtains reliable visual difference by edge indexing algorithm, and obtains spatial depth information according to the angle of parallax convergence, and then distinguishes the distance and proximity of objects. The optical flow method based on SIFT can get the moving speed of obstacles relative to the camera at every moment. In order to obtain more accurate position information more quickly, stereo vision and optical flow are combined. Experimental results show that this method can effectively improve the efficiency and accuracy of obstacle avoidance.
【作者单位】： 南京航空航天大学自动化学院;
【基金】：国家自然科学基金(61673209,61533008) 航空科学基金(2016ZA52009) 中央高校基本科研业务费专项资金(NJ20160026) 大学生创新训练计划项目基金
【分类号】：TP391.41;V279
，
本文编号：2245386
[Abstract]:UAVs must be able to detect obstacles and avoid them accurately during flight. In order to make UAV more safe in flight, a new obstacle avoidance method combining binocular stereo vision and optical flow is proposed. Binocular stereo vision obtains reliable visual difference by edge indexing algorithm, and obtains spatial depth information according to the angle of parallax convergence, and then distinguishes the distance and proximity of objects. The optical flow method based on SIFT can get the moving speed of obstacles relative to the camera at every moment. In order to obtain more accurate position information more quickly, stereo vision and optical flow are combined. Experimental results show that this method can effectively improve the efficiency and accuracy of obstacle avoidance.
【作者单位】： 南京航空航天大学自动化学院;
【基金】：国家自然科学基金(61673209,61533008) 航空科学基金(2016ZA52009) 中央高校基本科研业务费专项资金(NJ20160026) 大学生创新训练计划项目基金
【分类号】：TP391.41;V279
，
本文编号：2245386
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2245386.html
